Using Unmanned Aerial Vehicle Remote Sensing and a Monitoring Information System to Enhance the Management of Unauthorized Structures

applied sciences

Article Using Unmanned Aerial Vehicle Remote Sensing and a Monitoring Information System to Enhance the Management of Unauthorized Structures

Yuanrong He 1,2, Weiwei Ma 1,*, Zelong Ma 1, Wenjie Fu 1, Chihcheng Chen 3, Cheng-Fu Yang 4,* and Zhen Liu 3

1 Big Data Institute of Digital Natural Disaster Monitoring in Fujian, Xiamen University of Technology, Xiamen 361024, China; [email protected] (Y.H.); [email protected] (Z.M.); [email protected] (W.F.) 2 Key Laboratory of Ecological Environment and Information Atlas Fujian Provincial University, Putian University, Putian 351100, China 3 School of Information Engineering, Jimei University, Xiamen 361021, China; [email protected] (C.C.); [email protected] (Z.L.) 4 Department of Chemical and Materials Engineering, National University of Kaohsiung, Kaohsiung 081168, Taiwan * Correspondence: [email protected] (W.M.); [email protected] (C.-F.Y.)

Received: 12 October 2019; Accepted: 13 November 2019; Published: 18 November 2019

Featured Application: Unauthorized construction is common in residential areas and can result in various types of harm, such as pieces of buildings falling off in bad weather and causing damage. These incidents may lead to injury or death, as well as structural damage, increasing financial loss to governments and negatively affecting the local population’s livelihoods. In this study, we investigated a method that uses an unmanned aerial vehicle with remote sensing as well as a monitoring information system to enhance the detection and prevention of unauthorized construction.

Abstract: In this research, we investigated using unmanned aerial vehicle (UAV) photographic technology to prevent the further expansion of unauthorized construction and thereby reduce postdisaster losses. First, UAV dynamic aerial photography was used to obtain dynamic digital surface model (DSM) data and elevation changes of 2–8 m as the initial sieve target. Then, two periods of dynamic orthophoto images were superimposed for human–computer interaction interpretation, so we could quickly distinguish buildings undergoing expansion, new construction, or demolition. At the same time, mobile geographic information system (GIS) software was used to survey the ﬁeld, and the information gathered was developed to support unauthorized construction detection. Finally, aerial images, interpretation results, and ground survey information were integrated and released on WebGIS to build a regulatory platform that can achieve accurate management and eﬀectively prevent violations.

Keywords: unmanned aerial vehicle (UAV); dynamic digital surface model (DSM); geographic information system (GIS)

1. Introduction Over the past 40 years since China’s reform and opening up, the pace of urban construction has dramatically increased [1]. Between 2000 and 2005 alone, the total area of China’s provincial capitals increased by 90.15%. Between 2000 and 2010, China’s urban area grew by 101.04% [2,3].

Appl. Sci. 2019, 9, 4954; doi:10.3390/app9224954 www.mdpi.com/journal/applsci Appl. Sci. 2019, 9, 4954 2 of 18

At the same time, the volume of unauthorized construction has also increased in direct proportion, which has seriously affected urban management and increased the workloads of urban supervision departments. Those engaging in unauthorized construction have not obtained the required permission for land use, structural planning, and construction. Since the promulgation of the urban planning law, unauthorized construction is regarded as illegal building [4]. Unauthorized structures such as corrugated iron houses in coastal cities are disastrous when a typhoon strikes, as the high winds cause the iron sheeting to fly through the air, destroying vehicles, shattering glass, and leading to other adverse consequences, including injuries and deaths. Quickly and effectively identifying unauthorized construction has become a key problem for urban planning departments. To identify illegal changes to buildings in urban areas, high-resolution aerial images can be useful for clearly showing the basic external characteristics of structures. In the past, many methods had been investigated to detect changes in buildings. Matikainen and Hyyppa presented a method to automatically detect alterations to buildings using airborne laser scanner and digital aerial image data [5]. In a different approach, Huertas extracted data on buildings in old images and put the information into a database created using Site Model; new images were then compared with the models in the database to identify changes [6]. Agrostis applied an algorithm to a geographic information system (GIS) database to detect changes in buildings. The important core of the algorithm was measurement of the minimum square distance between original building boundaries and a vector polygon from the GIS in new images [7]. Watanabe used aerial images from different shooting angles and at different times. By eliminating the influence of parallax and building shadow caused by different shooting angles, the corresponding roof of a building in the two images was scanned to detect variation [8]. The buildings’ change detection can also be used in different areas. For example, Vu used the specificity of LIDAR data to detect building changes in earthquake-prone areas [9]. Recently, Marin proposed a new method for employing multitemporal very high resolution (VHR) synthetic aperture radar (SAR) images to detect changes, using the regular back-scattering characteristics of buildings to detect new or completely demolished buildings [10]. Bouziani used existing geographic databases and prior knowledge to automatically detect building changes in urban environments from high-resolution images, thereby identifying illegal construction and monitoring urban growth [11]. Tian proposed a solution to monitor changes in buildings based on height variations in stereo imagery and digital surface models (DSMs). The change was detected via a stereo-matching methodology that used the joints of height changes and Kullback-Leibler divergence similarity measurements between original and later images [12]. Guo used a multilevel random forest (RF) classifier to classify VHR images and compared the changes in two datasets to detect building alterations, using GIS data [13]. Although the above-mentioned studies have achieved good results in the identification of unauthorized construction, they cannot precisely prevent unauthorized construction. Thus, it is necessary to find a method of effectively detecting and managing unauthorized construction. In this study, we took the streets in the Siming District of Xiamen City as a research target, and we used multiperiod data and images from an unmanned aerial vehicle (UAV) with remote sensing to generate a database. We then developed an initial screening tool that gathered DSM data from UAV remote sensing to automatically recognize changes in building elevation [14]. This was used to compare two orthogonal images in an interactive way to extract boundaries that indicated new construction, or the extension or demolition of buildings. When the tilt photographs of oblique buildings and 720◦ panoramic aerial photography were used simultaneously to manually verify the automatic extraction of results, we built a database of illegal buildings. In this study, the constructed supervisory systems included 3D visualization of important and key zones, the buildings’ monitoring application software (APP) of named “dynamic inspect”, and a WebGIS supervisory platform. We were thereby able to create a completely new supervision model that uses “seeing from the sky”, “managing on the ground”, and “searching on the Internet” to detect illegal building [15]. Appl. Sci. 2019, 9, 4954 3 of 18

2. Identification and Supervision of Unauthorized Construction- Ground object recognition and change detection based on remote sensing images constitute a complex process, especially for small targets such as buildings. Low-resolution remote sensing images are insufficient for extracting the features of ground objects and hence less useful for detecting changes in buildings to identify unauthorized construction. The high-resolution data sources used for change detection have some limitations; for example, it is difficult to obtain images from the QuickBird satellite, and high-resolution aerial images take large amounts of memory to store. At present, although some cities have carried out building change detection and research to identify illegal buildings, and have tried to establish a business application system for finding illegal buildings, there still is no complete theoretical and technological system for identifying illegal buildings, because these applications have not been widely promoted [7,16]. The basic idea of identifying illegal urban construction activity in this manner is to take the distribution of urban buildings as the monitoring object, then use UAV remote sensing technology to quickly obtain images and spatial distribution information for buildings through image processing, pattern recognition, and other methods. The results are then overlaid with planning map data that only experts can interpret, allowing managers to monitor current building information and identify unauthorized construction. This method can effectively deter unauthorized construction activities and aid in the supervision of overall urban planning. It also can provide high-quality, intuitive spatial information support for dealing with urban disasters and emergencies, and promote the sustainable development of urban planning and construction in China. In this study, using UAV remote sensing for the dynamic monitoring of construction and the collection of monitoring information involved three main modules. The first was the data processing and illegal building identification module. The aim of this module was to complete data acquisition and illegal building identification. High-resolution UAV remote sensing images were captured from the research objects. At the same time, automatic preliminary screening of changes in building height was carried out based on Model Builder in ArcGIS, and the changes were confirmed by manual visual interpretation. Then, unauthorized construction was verified using a 3D model and 720◦ panorama imaging to obtain accurate data on illegal buildings. In the second module, we handled the data based on the WebGIS network information management system. This provided a dynamic supervision platform for a series of functions: target image identification, field verification, investigation and penal process monitoring, and historical tracing. We employed the architecture of WebGIS in the OpenGeo Suite and related technologies to analyze the design and integration process of PostGIS based on an object-oriented spatial database. By building a complete prototype system, it is possible to use WebGIS technology for digital information management related to unauthorized urban construction. In the third module, we used a mobile grid management terminal system for data communication. This mobile intelligent terminal provided information support for daily operations so data could then be collected and analyzed in a timely manner for presentation to grid managers. The built detection and management platform is shown in Figure1. Appl. Sci. 2019, 9, 4954 4 of 18 Appl. Sci. 2019, 9, x FOR PEER REVIEW 4 of 17

FigureFigure 1. 1.The The builtbuilt detectiondetection andand managementmanagement platform. DSM: DSM: digital digital surface surface model; model; DOM: DOM: digital digital orthophoto map orthophoto map

3. Aerial Aerial Images Images Capture Capture and Data Processing 3.1. Preimage Data Acquisition 3.1. Preimage Data Acquisition The ﬁrst aerial images obtained from the research target were captured from 19 to 22 February The first aerial images obtained from the research target were captured from 19 to 22 February 2017. The actual working time was 3 days, and the image spatial resolution was better than 10 cm. 2017. The actual working time was 3 days, and the image spatial resolution was better than 10 cm. The acquired image data were processed into an orthophoto map by software 32 Gbit capacity. As The acquired image data were processed into an orthophoto map by software 32 Gbit capacity. As ArcGIS software was used for image processing, we were able to add place names and administrative ArcGIS software was used for image processing, we were able to add place names and administrative boundaries to the images, enabling us to make initial grid blocks according to streets and the jurisdictions boundaries to the images, enabling us to make initial grid blocks according to streets and the of community organizations. After that, we divided the image data into 81 areas to make secondary jurisdictions of community organizations. After that, we divided the image data into 81 areas to make grid blocks, then printed out 82 maps, one of which was of the total area. The map scale for each block secondary grid blocks, then printed out 82 maps, one of which was of the total area. The map scale was about 1:1000. These results could be used by the grid managers to “see from the sky”. for each block was about 1:1000. These results could be used by the grid managers to “see from the sky”.3.2. Image Data Network Sharing

3.2. ImageThe volume Data Network of data Sharing had the potential to create bottlenecks that would slow down performance in the computer and at the interface with the GIS software, so we needed to publish on the web. Managers usedThe a web volume browser of data or mobilehad the phone potential to view to create aerial bottlenecks photos of that big-data would results. slow down The web performance platform inmainly the computer enabled basic and dataat the input, interface grid datawith management, the GIS software, grid member so we needed management, to publish daily on operational the web. Managerssupervision used of grid a web personnel, browser and or statistical mobile phone analysis to ofview events aerial in thephotos grid andof bi ofg-data management results. functionsThe web platformin the background. mainly enabled Based basic on high-resolution data input, grid aerial data images management, of the GIS grid visual member integrated management, development daily operationalenvironment, supervision we were able of togrid eff ectivelypersonnel, improve and statistical the ability analysis to manage of integratedevents in the grid grid monitoring and of managementdata and provide functions an eff ectivein the onlinebackgr technologyound. Based for on regulating high-resolution unauthorized aerial images construction. of the GIS visual integrated development environment, we were able to effectively improve the ability to manage integrated4. Identification grid ofmonitoring Unauthorized dataConstruction and provide an effective online technology for regulating unauthorized construction. The second set of aerial images obtained from the research target was captured on 3 March 2017. Using aerial imagery postprocessing software, we generated a digital orthophoto map (DOM), DSM, 4. Identification of Unauthorized Construction 3D images, and other data. With this approach, as long as the DSM of the building is obtained and changesThe insecond the DSM set of are aerial detected images by obtained comparison from of th imagese research from target the two was dates, captured preliminary on 3 March filtering 2017. Usingcan be aerial done imagery by computer. postprocessing On the basis software, of preliminary we generated computer a digital selection, orthophoto supplemented map (DOM), by manual DSM, 3D images, and other data. With this approach, as long as the DSM of the building is obtained and changes in the DSM are detected by comparison of images from the two dates, preliminary filtering Appl. Sci. 2019, 9, x FOR PEER REVIEW 5 of 17

Appl. Sci. 2019, 9, 4954 5 of 18 can be done by computer. On the basis of preliminary computer selection, supplemented by manual visual inspection, a 3D image and a 720° panoramic view can provide artificial verification, and then visualthe unexpected inspection, stronghold a 3D image can and be aobtained 720◦ panoramic and the corresponding view can provide thematic artificial maps verification, can be produced. and then the unexpected stronghold can be obtained and the corresponding thematic maps can be produced. 4.1. Preliminary Automatic Screening of Unauthorized Construction 4.1. Preliminary Automatic Screening of Unauthorized Construction In this paper, the program was coded in several steps for preliminary identification of illegal buildings,In this and paper, these the main program steps was are coded shown in severalin Figure steps 2. First, for preliminary the projection identification transformation of illegal of buildings,preprocessed and DSM these model main data steps was are required shown in in order Figure to2 make. First, DSM the data projection with different transformation phases have of preprocessedaccurate superposition. DSM model Second, data was different required DSM in data order with to make different DSM phases data with were di calculatedfferent phases by using have accurateArcGIS built-in superposition. grid calculator. Second, Third, different if the DSM difference data with value diff waserent smaller phases than were the calculated set threshold by using one, ArcGISit was judged built-in as grid a non-illegal calculator. building, Third, if theotherwise difference it was value an was illegal smaller building. than theAnd set fourth, threshold the one,result it wasdata judgedof the asbuilding a non-illegal area was building, extracted otherwise by superimposing it was an illegal the building. masking And data fourth, with the the result data dataof the of thebuilding building area. area was extracted by superimposing the masking data with the data of the building area.

Figure 2. Automatic extraction flow ﬂow chart of building elevation change detection.

The term “building changes” refers to newnew construction,construction, capping, or demolition of buildingsbuildings within aa certaincertain period period of of time time in ain specific a specific area. area. These These changes changes are mainly are mainly reflected reflected in building in building height, asheight, the results as the inresults Figure in3 Figurea,b show. 3a and By using3b show. the DSMBy using model the ofDSM a long model time of series a long in time the observationseries in the areaobservation and combining area and this combining with a spatial this analysis with a algorithm,spatial analysis we extracted algorithm, a change we distributionextracted adiagram change ofdistribution regional building diagram height, of regional as shown building in Figure height,4a. as Further shown analysis in Figure of this4a. Further diagram analysis enabled of us this to setdiagram a reasonable enabled number us to set and a reasonable interval of number height and difference interval classification of height difference values. Finally,classification we obtained values. aFinally, reclassification we obtained diagram a reclassification of changes indiagram buildings’ of changes heights. in Inbuildings’ combination heights. with In actual combination observation, with asactual the buildingsobservation, were as constructedthe buildings at were different construc timeted diff erencesat different and time their differences heights were and less their than heights 1.5 m, theywere couldless than be recognized 1.5 m, they as could non-illegal be recognized areas, so suchas no datan-illegal were areas, filtered so out.such The data final were result filtered is shown out. inThe Figure final 4resultb by selectingis shown diin ffFigureerent symbol4b by selecting styles for different the appropriate symbol styles layers. for Thethe appropriate results in Figure layers.4 showThe results that we in canFigure accurately 4 show and that intuitively we can accura presenttely dynamic and intuitively changes inpresent buildings’ dynamic heights changes within in a projectbuildings’ area heights for a specific within timea project period. area By for rapidly a specific recognizing time period. and By filtering rapidly changesrecognizing in buildings and filtering in a monitoredchanges in region,buildings including in a monitored new buildings, region, illegal including construction, new buildings, and demolition, illegal weconstruction, can effectively and reducedemolition, the e ffweort can required effectively for early reduce artificial the effort visual requ discriminationired for early of artificial objects that visual can discrimination provide cluesfor of furtherobjects that artificial can provide visual discrimination. clues for further artificial visual discrimination.

Appl. Sci. 2019, 9, 4954 6 of 18 Appl. Sci. 2019, 9, x FOR PEER REVIEW 6 of 17

Appl. Sci. 2019, 9, x FOR PEER REVIEW 6 of 17

(a) (b)

(a) Figure 3. Identification of (a) unauthorized construction and (b) undiminished(b) unauthorized structures. Figure 3. Identiﬁcation of (a) unauthorized construction and (b) undiminished unauthorized structures.

Figure 3. Identification of (a) unauthorized construction and (b) undiminished unauthorized structures.

(a) (b)

Figure 4. Identiﬁcation of building height: (a) change distribution map and (b) building height Figure 4. Identification of building height: (a) change distribution map and (b) building height variation reclassiﬁcation. variation reclassification.(a) (b) First, we pretreated the multiphase DSM model data for the same research area, including data First, we pretreated the multiphase DSM model data for the same research area, including data exploration, projection transformation, and so forth, to ensure all the pretreated time-phase data was exploration, projection transformation, and so forth, to ensure all the pretreated time-phase data was founded on the same accurate coordinate system for subsequent spatial analysis. Next, we used a grid foundedFigure on 4. the Identification same accurate of building coordinate height: system (a) change for subsequent distribution spatial map and analysis. (b) building Next, heightwe used a computing tool to subtract the pre- and postprocessed phase data [17]. The formula is as follows: grid variationcomputing reclassification. tool to subtract the pre- and postprocessed phase data [17]. The formula is as follows:

First, we pretreatedResult the= current-image multiphase DSM ProjectRaste model dataearly-image for the same ProjectRaste research area, including data (1) Result = current-image ProjectRaste −– early-image ProjectRaste (1) exploration, projection transformation, and so forth, to ensure all the pretreated time-phase data was where current-image ProjectRaste is the post-temporal DSM model (representing the current state), founded on the same accurate coordinate system for subsequent spatial analysis. Next, we used a andwhere early-image current-image ProjectRaste ProjectRaste is the is antecedent the post-tem phaseporal DSM DSM model model (representing (representing the the past current state). state), The grid computing tool to subtract the pre- and postprocessed phase data [17]. The formula is as follows: diandfference early-image between ProjectRaste the early-and is the current-phase antecedent ph DSMase DSM model model can be(representing displayed for the the past same state). region. The difference between the early-and current-phase DSM model can be displayed for the same region. A A positive result indicatesResult = ancurrent-image increase in height,ProjectRaste which – early-image may be due ProjectRaste to new construction or illegal (1) buildings.positive result A negative indicates result an indicatesincrease in a lowerheight, elevation, which may which be maydue beto thenew result construction of demolition. or illegal We buildings. A negative result indicates a lower elevation, which may be the result of demolition. We wherecan thereby current-image obtain a building ProjectRaste height is distributionthe post-tem map.poral DSM model (representing the current state), can thereby obtain a building height distribution map. and early-imageThrough the ProjectRaste above process, is the we achievedantecedent automatic phase DSM extraction model of (representing all elevation changesthe past instate). a certain The Through the above process, we achieved automatic extraction of all elevation changes in a differenceregion during between a certain the early-and time period. current-phase As the main DS objectiveM model of can this be project displayed was tofor examine the same built region. areas, A certain region during a certain time period. As the main objective of this project was to examine built potentialpositive result disturbance indicates factors an increase such as in woodland, height, which shrubs, may and be mountainous due to new areasconstruction were eliminated, or illegal areas, potential disturbance factors such as woodland, shrubs, and mountainous areas were buildings.and we applied A negative masking result tools indicates to the builta lower area. elevation, A difference which in may phase be heightthe result less of than demolition.1.5 m wasWe eliminated, and we applied masking tools to the built area. A difference in phase height less± than ±1.5 canconsidered thereby aobtain nonbuilding a building change height interval distribution and was map. filtered out. The test data were then reclassified m was considered a nonbuilding change interval and was filtered out. The test data were then at intervalsThrough of the 1 and above 1.5 m.process, A distribution we achieved map ofautomatic building extraction height change of all was elevation thus obtained changes and in a reclassified at intervals of 1 and 1.5 m. A distribution map of building height change was thus certaindata processing region during model a certain was created time period. to generate As the the main appropriate objective data of this processing project was tools. to examine The tools built can obtained and a data processing model was created to generate the appropriate data processing tools. areas,be shared potential for further disturbance development factors and such improvement. as woodland, It also shrubs, has a goodand mountainous interface to facilitate areas were bulk The tools can be shared for further development and improvement. It also has a good interface to eliminated,data processing. and we applied masking tools to the built area. A difference in phase height less than ±1.5 facilitate bulk data processing. m wasAfter considered computer-automated a nonbuilding processing, change interval new DSM and data was were filtered obtained: out. The elevation test data diff wereerences then in After computer-automated processing, new DSM data were obtained: elevation differences in reclassifiedtwo-stage images. at intervals For convenient of 1 and application,1.5 m. A distribution the data’s attributesmap of building can be classified height change using a systemwas thus in two-stage images. For convenient application, the data’s attributes can be classified using a system whichobtained a thresholdand a data value processing is assigned model a was color created for the to range generate area. the With appropriate that, managers data processing can see a tools. lot of in which a threshold value is assigned a color for the range area. With that, managers can see a lot of extracted patches on the preprocessing images. Visual comparison of base maps allows managers to see elevation changes in the houses of a given area. The ArcGIS recognition function is then used to check the buildings’ elevation and guide the follow-up visual interpretation work. A thematic map of preliminary screening results can also be made, as Figure 5 shows. Appl. Sci. 2019, 9, 4954 7 of 18 extracted patches on the preprocessing images. Visual comparison of base maps allows managers to see elevation changes in the houses of a given area. The ArcGIS recognition function is then used to check the buildings’ elevation and guide the follow-up visual interpretation work. A thematic map of Appl. Sci. 2019, 9, x FOR PEER REVIEW 7 of 17 preliminaryAppl. Sci. 2019, 9 screening, x FOR PEER results REVIEW can also be made, as Figure5 shows. 7 of 17

(a) (b) (a) (b)

Figure 5. ((a) Patches elevation query and ( b) patches elevation drawing. Figure 5. (a) Patches elevation query and (b) patches elevation drawing.

4.2. Artificial Artificial Visual Interpretations 4.2. Artificial Visual Interpretations Next, wewe establishedestablished the the database database and and the the new new corresponding corresponding elements, elements, including including the illegalthe illegal and Next, we established the database and the new corresponding elements, including the illegal demolitionand demolition buildings. buildings. Computer Computer processing processing enables enables images images to be extracted to be extracted when obvious when additionobvious and demolition buildings. Computer processing enables images to be extracted when obvious oraddition removal or ofremoval buildings of buildings has occurred, has occurred, but the variations but the variations of floors inof buildings floors in buildings (the blocks (the shown blocks in addition or removal of buildings has occurred, but the variations of floors in buildings (the blocks Figureshown6 in) areFigure not 6) easily are not visible easily in visible the image. in the The image. elevation The elevation of the DSM of the chart DSM can chart be checkedcan be checked using shown in Figure 6) are not easily visible in the image. The elevation of the DSM chart can be checked ArcGISusing ArcGIS recognition recognition to confirm to confirm architectural architectura changesl changes have occurred. have occurred. Figure6 showsFigure some 6 shows results some of using ArcGIS recognition to confirm architectural changes have occurred. Figure 6 shows some artificialresults of visual artificial interpretation. visual interpretation. results of artificial visual interpretation.

Figure 6. Example of artificial artiﬁcial visual interpretation results. Figure 6. Example of artificial visual interpretation results.

4.3. Two Kinds of 3D Verification 4.3. Two Kinds of 3D Verification

Three-Dimensional Image Verification Three-Dimensional Image Verification Oblique photographs of the aerial images were obtained using postprocessing software, Oblique photographs of the aerial images were obtained using postprocessing software, enabling us to create 3D images. With a 3D image model, we could take measurements of buildings, enabling us to create 3D images. With a 3D image model, we could take measurements of buildings, such as height, area, and so on, as Figure 7 shows. The oblique photographic data were then output such as height, area, and so on, as Figure 7 shows. The oblique photographic data were then output into open scene graph binary (OSGB, one of the formats for oblique photographic data) format for into open scene graph binary (OSGB, one of the formats for oblique photographic data) format for 3D data; this is an open-source format that can support the secondary application of a variety of 3D 3D data; this is an open-source format that can support the secondary application of a variety of 3D software. Managers could then release the exported OSGB format data to the online map browser software. Managers could then release the exported OSGB format data to the online map browser Appl. Sci. 2019, 9, 4954 8 of 18

4.3. Two Kinds of 3D Veriﬁcation Appl. Sci. 2019, 9, x FOR PEER REVIEW 8 of 17 4.3.1. Three-Dimensional Image Veriﬁcation namedOblique “Local photographs space”. Publ ofishing the aerial this images information were obtained online usingwould postprocessing enable inhabitants software, to enablingidentify usunauthorized to create 3D construction. images. With This a 3D method image could model, deter we illegal could takeactivity measurements and at the same of buildings, time facilitate such the as height,handling area, of lawful and so on,judgments, as Figure which7 shows. could The be oblique used to photographic educate people data about were thenlegal output procedures into open and sceneconsequences graph binary related (OSGB, to illegal one buildings. of the formats for oblique photographic data) format for 3D data; this is an open-source format that can support the secondary application of a variety of 3D software. Managers720° Panoramic could Verification then release the exported OSGB format data to the online map browser named “LocalA space”.720° panorama Publishing is thisachieved information by combining online would horizontal enable 360° inhabitants and vertical to identify 360° unauthorizedimages. The construction.shooting and production This method speeds could are deter very illegal fast. activityImages andare then at the collected same time as data facilitate from the a website, handling and of lawfula web browser judgments, can whichclearly couldand intuitively be used to display educate the people results about from legalthese procedures images. In general, and consequences the results relatedare viewable to illegal quickly buildings. and easily on a mobile phone or a PC [18,19].

Figure 7. Building measurement based on 3D model.

4.3.2.4.4. Verification 720◦ Panoramic of Precision Veriﬁcation Extraction

AIn 720this◦ panoramapaper, the is automatic achieved byextraction combining program horizontal of “Building 360◦ and vertical Elevation 360 ◦Change”images. Thewas shooting used to andextract production the suspected speeds illegal are verybuilding fast. spots Images on arethe thenmap collectedwith height as changes data from larger a website, than 1.5m. and aBased web browseron the extraction can clearly data, and we intuitively could use display 3D images the resultsand panoramic from these images images. to verify In general, the suspected the results illegal are viewableconstruction quickly spots, and and easily then on we a could mobile obtain phone the or i allegal PC [18 construction,19]. spots for assignment. This is a human–computer interaction verification process. For that, we took Siming District of Xiamen City 4.4. Verification of Precision Extraction as the experimental object to verify the accuracy. According to the field verification, 27 illegal buildingsIn this could paper, be found the automatic in this block. extraction A total program of 26 suspected of “Building illegal Elevationbuildings spots Change” on the was map used were to extractextracted the in suspected this experiment, illegal building 21 of which spots were on the actual map illegal with heightbuildings, changes 5 of largerwhich thanwere 1.5m. misjudged, Based onand the 6 of extraction which were data, lost. we The could accuracy use 3D of images human–computer and panoramic interaction images toextraction verify the of suspectedillegal buildings illegal constructionwas 78%, which spots, can and meet then the wemanagers’ could obtain requirements. the illegal construction spots for assignment. This is a human–computer interaction verification process. For that, we took Siming District of Xiamen City as the5. WebGIS-Based experimental object Violation to verify Monitoring the accuracy. Network According Information to the field Management verification, System 27 illegal buildings couldThe be foundmain method in this block. for managing A total of illegal 26 suspected building illegal is to buildings use a regulatory spots on platform the map wereto supervise extracted a inseries this of experiment, functions, 21such of whichas target were image actual recognition, illegal buildings, field verification, 5 of which investigation were misjudged, and andprocess 6 of whichmonitoring, were lost.and Thehistorical accuracy tracking of human–computer of illegal buildings. interaction Therefore, extraction the adopted of illegal digital buildings information was 78%, whichmanagement can meet system the managers’ needs accurate requirements. analyses of positioning and spatial information about illegal entities, which are the main functions of the GIS. Compared with a traditional GIS software platform, the system platform built with WebGIS can access data from any platform and any place. The user- friendly interface can be customized with simple, clear, and intuitive images. It can effectively balance a high volume of spatial data for processing calculation and maximize the utilization of computer Appl. Sci. 2019, 9, 4954 9 of 18

5. WebGIS-Based Violation Monitoring Network Information Management System The main method for managing illegal building is to use a regulatory platform to supervise a series of functions, such as target image recognition, field verification, investigation and process monitoring, and historical tracking of illegal buildings. Therefore, the adopted digital information management system needs accurate analyses of positioning and spatial information about illegal entities, which are the main functions of the GIS. Compared with a traditional GIS software platform, the system platform builtAppl. Sci. with 2019 WebGIS, 9, x FOR can PEER access REVIEW data from any platform and any place. The user-friendly interface can9 of be17 customized with simple, clear, and intuitive images. It can effectively balance a high volume of spatial dataresources. for processing It can also calculation use a spatial and database maximize for the data utilization management of computer and effectively resources. improve It can alsoefficiency use a spatialand accuracy. database for data management and effectively improve efficiency and accuracy.

5.1.5.1. System Functions and Design AccordingAccording toto thethe actualactual demandsdemands ofof illegalillegal constructionconstruction management,management, thethe systemsystem isis divideddivided intointo thethe followingfollowing fourfour functionalfunctional modules,modules, asas shownshown inin FigureFigure8 8..

Figure 8. Functional modules of designed system. Figure 8. Functional modules of designed system. (1) Comprehensive demonstration module (1) Comprehensive demonstration module BasedBased onon remoteremote sensing data of UAV, everyevery illegalillegal constructionconstruction projectproject cancan bebe processedprocessed intointo thethe productsproducts ofof orthophotoorthophoto images,images, digitaldigital surfacesurface models,models, 3D3D models,models, andand panoramicpanoramic images.images. These productsproducts cancan displaydisplay thethe inspectioninspection areaarea inin aa moremore comprehensivecomprehensive displaydisplay modemode andand showshow thethe developmentdevelopment status of illegalillegal buildingsbuildings in a moremore vividvivid andand ﬂexibleflexible way.way. It is oneone wayway ofof mixingmixing managementmanagement andand supervisionsupervision forfor thethe phenomenonphenomenon ofof illegalillegal buildings.buildings. (2) Image database and management module (2) Image database and management module The main task of this module is to manage and schedule the preprocessed UAV remote sensing imagesThe and main it can task provide of this module basic services is to manage for the and entire schedule management the preprocessed system. ItUAV mainly remote includes sensing the imagesestablishment and it canof the provide image basic database, services the forimport the entire of multiformat management image system. data, Itthe mainly browsing includes and thedisplaying establishment of image of the data, image and database, the location the import and scope of multiformat of image imagedata, data,and other the browsing information and displayingmanagement. of image data, and the location and scope of image data, and other information management. (3) Vector database and management module (3) Vector database and management module This module is mainly used for the management and scheduling of the relevant vector data generatingThis module from the is mainlyprocessing used of fororiginal the management data and vector and data scheduling of buildings’ of the planning relevant provided vector data by generatingrelevant departments. from the processing By superimposing of original the data vector and data vector on datathe images of buildings’ and comparing planning them provided with the by relevantplanning departments. approval maps, By superimposingthey can accura thetely vector identify data the on illegal the images buildings. and comparing them with the planning(4) GIS approval analysis maps, and monitoring they can accurately module identifyof illegal the buildings illegal buildings. By using the accurate positioning and analysis capability of spatial information of the GIS (4)system, GIS analysis the information and monitoring such as module the location, of illegal area, buildings and center point of the illegal buildings can be calculated.By using Then, the accuratewe can exclude positioning and screen and analysis out the capability initially extracted of spatial informationillegal buildings’ of the information, GIS system, theand information we can determine such as the location,geographical area, locations and center of point the illegal of the illegalbuildings, buildings the types can beof calculated.the illegal Then,buildings, we can and exclude other information and screen outto prepare the initially for the extracted next step illegal of monitoring. buildings’ information,The monitoring and work we can of determineillegal buildings the geographical is mainly divided locations into of six the steps: illegal segmentation buildings, thegrid, types polygon of the establishment, illegal buildings, polygon and otherconfirmation, information task to distribution, prepare for thefield next verification, step of monitoring. and archive The confir monitoringmation. work The ofmonitoring illegal buildings area is isgridded mainly and divided divided, into sixand steps: the management segmentation polygon grid, polygon of illegal establishment, buildings polygonis created conﬁrmation, (a polygon represents one household of illegal building). After confirming that there is no error in the polygon information, the task is dispatched to the mobile station's grid members, and the grid members can verify the illegal construction in the field, confirm the relevant information, and then return the data through the mobile terminal to archive.

5.2. System Construction This section uses the architecture of WebGIS present in OpenGeo Suite and related technologies to analyze the design and integration process of PostGIS based on an object-oriented spatial database. Appl. Sci. 2019, 9, 4954 10 of 18 task distribution, field verification, and archive confirmation. The monitoring area is gridded and divided, and the management polygon of illegal buildings is created (a polygon represents one household of illegal building). After confirming that there is no error in the polygon information, the task is dispatched to the mobile station’s grid members, and the grid members can verify the illegal construction in the field, confirm the relevant information, and then return the data through the mobile terminal to archive.

5.2. System Construction

Appl. Sci.This 2019 section, 9, x FOR uses PEER the REVIEW architecture of WebGIS present in OpenGeo Suite and related technologies10 of to17 analyze the design and integration process of PostGIS based on an object-oriented spatial database. In otherIn other words, words, we elaborate we elaborate on how on to how comprehensively to comprehensively use a WebGIS-based use a WebGIS-based prototype forprototype analyzing for a digitalanalyzing information a digital information management management system to detect system unauthorized to detect unauthorized construction construction [20,21]. [20,21].

5.2.1.5.2.1. SystemSystem ArchitectureArchitecture andand DevelopmentDevelopment TechnologyTechnology AtAt present,present, thethe WebGISWebGIS systemsystem mainlymainly usesuses thethe browserbrowser/server/server (B(B/S)/S) mode.mode. The BB/S/S modemode simpliﬁessimplifies clientclient software,software, developsdevelops thethe system’ssystem’s functions,functions, andand maintainsmaintains andand updatesupdates recognizingrecognizing methodsmethods onon thethe application server.server. The database serverserver is composed of a three-layer architecture, andand thethe constructedconstructed data data is placedis placed in the in server’sthe server's database database to form to a clientform layera client and layer the middle and the application middle layer.application Compared layer. to Compared the B/S mode to the and B/S the mode traditional and the standalone traditional client standalone mode, it isclient easier mode, to manage it is easier and maintain,to manage makes and maintain, the leastdemands makes the on least the client, demands and ison convenient the client, to and use is and convenient popularize. to WebGISuse and implementationpopularize. WebGIS in this implementation study was based in this on thestudy development was based on of the open-sourcedevelopment software of the open-source OpenGeo Suite,software which OpenGeo provides Suite, a comprehensive which provides web-based a comprehensive mapping web-based and data-sharing mapping solution. and data-sharing Software developmentsolution. Software was donedevelopment mainly withwas done GeoServer, mainly PostGIS, with GeoServer, OpenLayers, PostGIS, and OpenLayers, other components. and other In addition,components. OpenGeo In addition, Suite provided OpenGeo other Suite components, provided suchother as components, GeoExplore andsuch GeoWebCache. as GeoExplore When and OpenGeoGeoWebCache. Suite was When being OpenGeo used, the Suite “Dashboard” was being wasused the, the platform “Dashboard” for logging was the in andplatform editing. forFigure logging9 showsin and theediting. software’s Figure architecture. 9 shows the software’s architecture.

Figure 9.9. Hierarchical relationship of the OpenGeoOpenGeo SuiteSuite seriesseries software.software.

5.2.2.5.2.2. SpatialSpatial DatabaseDatabase DesignDesign andand IntegrationIntegration InIn thethe processprocess ofof designingdesigning aa spatial database, twotwo aspects require attention:attention: the building-related geographicgeographic informationinformation data, data, such such as as the the UAV UAV aerial ae imagerial image data, data, and theand preliminary the preliminary screening screening vector datavector and data related and basicrelated data basic such data as point such ofas interest point of (POI). interest At the (POI). same At time, the thesame landscape time, the architecture landscape dataarchitecture needs to data use theneeds management to use the ofmanagement spatial database of spatial for conducting database for the conducting analyses of the detailed analyses needs. of Relevantdetailed needs. departments Relevant will departme constructnts a standardwill construct mode a that standard suits their mode processing that suits system their processing to achieve thesystem dynamic to achieve monitoring the dynamic of unauthorized monitoring construction. of unauthorized To achieve construction. these goals, To these achieve steps these should goals, be followedthese steps in should designing be followed the database: in designing the database: (1)(1) ConceptualConceptual modelmodel ofof spatialspatial data:data: SpatialSpatial datadata classificationclassification includesincludes vectorsvectors andand grids.grids. AA three-levelthree-level approachapproach toto thethe datadata modelmodel isis required,required, namely,namely, conceptualconceptual model,model, logicallogical model,model, andand physical model. Traditional relational databases cannot be used directly to manage spatial data; instead, data models and methods must be extended. The conceptual model is the first level of abstraction from the real world to the machine world. (2) Logic model of spatial data: A logical model can be understood by mapping the real world to a computer or database system. In such a model, the database areas are: emerged hierarchical model, mesh model, relational model, and object-relational model. This article mainly relates to the object-relational model, which results from combining relational database technology and object- oriented programming. An object-relational database generally has the following functions: ① extend data types, ② support complex objects, ③ support the concept of inheritance, and ④ provide common rules and systems. PostGIS is based on predefined data types and extended geometry types. Predefined data types are stored and managed using existing data types in relational Appl. Sci. 2019, 9, 4954 11 of 18 physical model. Traditional relational databases cannot be used directly to manage spatial data; instead, data models and methods must be extended. The conceptual model is the first level of abstraction from the real world to the machine world. (2) Logic model of spatial data: A logical model can be understood by mapping the real world to a computer or database system. In such a model, the database areas are: emerged hierarchical model, mesh model, relational model, and object-relational model. This article mainly relates to the object-relational model, which results from combining relational database technology and object-oriented programming. An object-relational database generally has the following functions: O1 extend data types, O2 support complex objects, O3 support the concept of inheritance, and O4 provide common rules and systems. PostGIS is based on predefined data types and extended geometry types. Predefined data types are stored and managed using existing data types in relational databases, whereas extended data types are stored and managed using extended geometry data types.

– Enable PostGIS (includes raster) CREATE EXTENSION postures; – Enable Topology CREATE EXTENSION postgis_topology; – Fuzzy matching needed for Tiger CREATE EXTENSION fuzzystrmatch; – Enable US Tiger Geocoder CREATE EXTENSION postgis_tiger_geocoder;

(2) Building the map server: GeoServer structures mainly use the default Jetty application server of OpenGeo. Because GeoServer is built on Java 2 Platform Enterprise Edition (J2EE), users need to install and conﬁgure the Java environment before installing OpenGeo. They also need to start Geoserver-related services after installation, then open OpenGeo’s Dashboard console through a browser. At the same time the needed vector data are imported and connected with GeoServer and PostGIS database, the user selects Import Data-> postGIS and enters the correct parameters, including the database name, port number, user name, password, and so on. After the data import is completed, the user can employ GeoExplorer to modify the layer style, including color, symbol, and transparency. Multiple layers can further build the layer group to improve computational eﬃciency, as shown in Figure 10. For importing image data, GeoServer provides an extended model that can directly import the data into the remote sensing images and automatically generate a mosaic. So, to acquire UAV aerial data, the user can go directly along the path (Import Data-> Mosaic) to select the appropriate storage path for completing the data import. After the import is completed, the user can employ the GeoWebCache tool to set the appropriate parameters, including hierarchy levels and image formats. After an image pyramid is built, the user can preview it, as shown in Figure 11. Appl. Sci. 2019, 9, x FOR PEER REVIEW 11 of 17

databases, whereas extended data types are stored and managed using extended geometry data types.

5.2.3. System Development and Implementation (1) Building the server environment: Establishing a server environment mainly involves the installation of OpenGeo Suite, a PostGIS database, and QGIS. All the software needs to be installed in the same directory. After the database is constructed, users need to load the functions of the extension module provided by PostGIS, which are mainly a series of processing functions based on the OpenGIS standard. After connection to the database, the server environment is realized by opening the “console” box in the PgAdmin menu and then executing the following SQL statement: --Enable PostGIS (includes raster) CREATE EXTENSION postures; --Enable Topology CREATE EXTENSION postgis_topology; --Fuzzy matching needed for Tiger CREATE EXTENSION fuzzystrmatch; --Enable US Tiger Geocoder CREATE EXTENSION postgis_tiger_geocoder; (2) Building the map server: GeoServer structures mainly use the default Jetty application server of OpenGeo. Because GeoServer is built on Java 2 Platform Enterprise Edition (J2EE), users need to install and configure the Java environment before installing OpenGeo. They also need to start Geoserver-related services after installation, then open OpenGeo's Dashboard console through a browser. At the same time the needed vector data are imported and connected with GeoServer and PostGIS database, the user selects Import Data-> postGIS and enters the correct parameters, including the database name, port number, user name, password, and so on. After the data import is completed, the user can employ GeoExplorer to modify the layer style, including color, symbol, and

Appl.transparency. Sci. 2019, 9, 4954 Multiple layers can further build the layer group to improve computational efficiency,12 of 18 as shown in Figure 10.

(a) (b)

Figure 10. a b Figure 10.Layer Layer modiﬁcation modification of ( of) the(a) the vector vector layer layer group group and and ( ) the(b) the label label layer. layer. Appl. Sci. 2019, 9, x FOR PEER REVIEW 12 of 17 For importing image data, GeoServer provides an extended model that can directly import the data into the remote sensing images and automatically generate a mosaic. So, to acquire UAV aerial data, the user can go directly along the path (Import Data-> Mosaic) to select the appropriate storage path for completing the data import. After the import is completed, the user can employ the GeoWebCache tool to set the appropriate parameters, including hierarchy levels and image formats. After an image pyramid is built, the user can preview it, as shown in Figure 11.

Figure 11. Image release.

(3) Front-endFront-end access page development: In In this this study, we used the OpenLayers 3 and LeafLetLeafLet frameworks toto develop develop the the front front page. page. When When the remotethe remote sensing sensing data for data the for demonstration the demonstration of research of targetresearch cases target is introduced, cases is introduced, the page resultsthe page show results that show the core that part the ofcore the part Map of display the Map is thedisplay definition is the ofdefinition the Map of element. the Map Then, element. the remote Then, sensingthe remote data sensing can define data the can current define map the neededcurrent tomap load needed the layer. to Inload this the method, layer. In requisite this method, points ofrequisite interest, points roads, of and interest, other layersroads, canand be other added layers to order. can be The added display to oforder. different The display layers can of bediffer controlledent layers through can be the controlled layer’s panel, through and the the layer’s final e ffpanel,ect is shownand the in final Figure effect 12. is shown in Figure 12.

Figure 12. Vector layer release.

6. Grid Management Mobile System Construction

6.1. Function Design of Grid Management on Mobile Terminal System As grid management has become routine, large-scale, and unevenly distributed, it is necessary to introduce mobile terminals as information support tools for daily operations, to prepare grid managers for timely on-site data collection and analysis. The system framework of the mobile law enforcement platform for illegal constructions mainly consists of three parts, namely, mobile terminal, PC terminal, and server. The functions of the mobile terminal include main functions such as home page, grid task, illegal construction record, location navigation, and opinion feedback. The PC terminal accesses the monitoring system through the browser, mainly for data management and display and task distribution and management of the mobile terminal, so that law enforcement Appl. Sci. 2019, 9, x FOR PEER REVIEW 12 of 17

Figure 11. Image release.

(3) Front-end access page development: In this study, we used the OpenLayers 3 and LeafLet frameworks to develop the front page. When the remote sensing data for the demonstration of research target cases is introduced, the page results show that the core part of the Map display is the definition of the Map element. Then, the remote sensing data can define the current map needed to load the layer. In this method, requisite points of interest, roads, and other layers can be added to Appl.order. Sci. The2019 display, 9, 4954 of different layers can be controlled through the layer’s panel, and the final13 effect of 18 is shown in Figure 12.

Figure 12. Vector layer release.

6. Grid Management Mobile System Construction

6.1. Function Design of Grid Management on Mobile Terminal System 6.1. Function Design of Grid Management on Mobile Terminal System As grid management has become routine, large-scale, and unevenly distributed, it is necessary to As grid management has become routine, large-scale, and unevenly distributed, it is necessary introduce mobile terminals as information support tools for daily operations, to prepare grid managers to introduce mobile terminals as information support tools for daily operations, to prepare grid for timely on-site data collection and analysis. The system framework of the mobile law enforcement managers for timely on-site data collection and analysis. The system framework of the mobile law platform for illegal constructions mainly consists of three parts, namely, mobile terminal, PC terminal, enforcement platform for illegal constructions mainly consists of three parts, namely, mobile and server. The functions of the mobile terminal include main functions such as home page, grid task, terminal, PC terminal, and server. The functions of the mobile terminal include main functions such illegal construction record, location navigation, and opinion feedback. The PC terminal accesses the as home page, grid task, illegal construction record, location navigation, and opinion feedback. The monitoring system through the browser, mainly for data management and display and task distribution PCAppl. terminal Sci. 2019, 9accesses, x FOR PEER the REVIEW monitoring system through the browser, mainly for data management13 ofand 17 and management of the mobile terminal, so that law enforcement personnel can carry out their work display and task distribution and management of the mobile terminal, so that law enforcement morepersonnel easily. can The carry server out mainlytheir work serves more mobile easily. terminals, The server such mainly as uploading serves mobile evidence terminals, to the such server, as obtaininguploading personal evidence tasks, to the historical server, obtaining remote sensingpersonal images tasks, historical of illegal constructionremote sensing areas, images and of so illegal forth. Theconstruction physical structureareas, and of so the forth. system The isphysical shown instructure Figure 13of .the system is shown in Figure 13.

Figure 13. The physical structure of the system.

The mainmain functions functions of theof mobilethe mobile terminal terminal module module include homeinclude page, home grid page, task, illegalgrid constructiontask, illegal record,construction location record, navigation, location and navigation, opinion feedback.and opinion feedback. (1) Home page (1) Home page The home page is a map that mainly shows the current location of the grid members and the surroundingThe home illegal page buildings. is a map that mainly shows the current location of the grid members and the surrounding(2) Grid illegaltask buildings. The main function of the grid task module is to perform task notification. Through the network information management system of violation monitoring based on WebGIS, each grid member can receive the currently assigned monitoring tasks in the grid task module. As long as the grid member selects a polygon, he can use the positioning module, navigating himself to the task location. (3) Illegal construction record The illegal construction record module is investigated for on-site forensics. The module displays information such as the districts, geographic coordinates, areas, and types of the illegal buildings. At the same time, it will retrieve the historical images and current images of the illegal buildings in the server. The grid members can compare the images’ information with the actual situation to determine whether it is illegal. Also, the grid members can use the module for photo samplings, data modifications, and other works. (4) Location navigation This module displays all illegal buildings’ positions within the grid members’ management area. By selecting a location, the navigation function can be provided, so that the grid members can find the illegal buildings more quickly and improve work efficiency. At the same time, the positioning navigation module will transmit the current positions of the grid members to the PC terminal in real time, and the management personnel can manage the current work progress of the grid members. (5) Opinion feedback This module is for users who have experience of the mobile law enforcement platform to put forward suggestions for improvement to the developers.

6.2. Function Realization of Mobile Terminal In this paper, the HTML5 + PhoneGap framework was adopted to implement the development of an APP on the user’s side, and at the same time, back-end logic computing service was provided through J2EE software, which not only improves the stability but also takes into account the user’s habits and updates the service accordingly.

6.2.1. Illegal Building Information Verification Preparation Before government staff enter community buildings to collect information, they need a general understanding of the architectural regulation area and familiarity with the terrain and the path of the Appl. Sci. 2019, 9, 4954 14 of 18

(2) Grid task The main function of the grid task module is to perform task notiﬁcation. Through the network information management system of violation monitoring based on WebGIS, each grid member can receive the currently assigned monitoring tasks in the grid task module. As long as the grid member selects a polygon, he can use the positioning module, navigating himself to the task location.

(3) Illegal construction record The illegal construction record module is investigated for on-site forensics. The module displays information such as the districts, geographic coordinates, areas, and types of the illegal buildings. At the same time, it will retrieve the historical images and current images of the illegal buildings in the server. The grid members can compare the images’ information with the actual situation to determine whether it is illegal. Also, the grid members can use the module for photo samplings, data modiﬁcations, and other works.

(4) Location navigation This module displays all illegal buildings’ positions within the grid members’ management area. By selecting a location, the navigation function can be provided, so that the grid members can ﬁnd the illegal buildings more quickly and improve work eﬃciency. At the same time, the positioning navigation module will transmit the current positions of the grid members to the PC terminal in real time, and the management personnel can manage the current work progress of the grid members.

(5) Opinion feedback This module is for users who have experience of the mobile law enforcement platform to put forward suggestions for improvement to the developers.

6.2.1. Illegal Building Information Verification Preparation Before government staff enter community buildings to collect information, they need a general understanding of the architectural regulation area and familiarity with the terrain and the path of the query location. After confirming the location, they go there carrying the APP-related equipment (for example, cell phones) for grids’ search. Using APP-based positioning to navigate, they can reach the specified location. When entering the work area, they then call out the correct projective images of the marked buildings and inspect the spots marked as having changed.

6.2.2. Mobile APP Site Information Collection As the information gathered by the Grid Inspector APP includes the precise location of a household (including the specific house number) and the architectural changes observed (including photos and other evidence), sensitive personal information related to the vital interests of many people is collected. The system therefore requires that grid members use their own accounts’ passwords to log in and start building information collection maps. After each grid’s operator logs in, they can view their inspection records to check their own inspection places and quickly find the nondetection places. This means that grid operators can have an accurate grasp of the historical records for their regions of responsibility, giving them a clear understanding of the changes in their regulatory regions, and then can effectively monitor to detect Appl. Sci. 2019, 9, 4954 15 of 18 unauthorized and illegal construction. These results prove that the mobile APP can help grid operators monitor newly built or illegal buildings and prevent dangerous demolitions, so operators can smoothly apply regulations and enforce laws. Grid operators’ daily tasks will be to reach the grid site and use the APP to select grids, times, and statuses, then enter selective auxiliary information. Via photo records, they can also add corresponding notes. Because the system provides geographical locations and positional navigation, grid operators can arrive at the correct positions at the right time to collect information. Using the “inquiry on the ground” approach can greatly increase the operators’ efficiency and the reliability and timeliness of data management. In addition, the system keeps track of grid staff’s daily work, so histories and inspection results can be tracked through the web-based system and improve daily grid management.

6.2.3. Building Monitoring APP and Web-Side Data Synchronization Association For information dissemination, monitoring and management terminals should provide mobile data interfaces so that grid operators can synchronize handling the processing steps and the data management system. This system can provide decision makers with real-time information from the web-based platform, helping to achieve the regular output of work reports. After grid members have collected site information and associated data, the managers can synchronize to discern relevant information and data, including the names and times of events, speciﬁc house numbers, photos, and events that have occurred on the site. Management staﬀ can use the associated building information to determine whether action should be taken.

6.2.4. Illegal Building Query Management Systems

In this study, the management system summarized the DOM, DSM, and 720◦ panoramas, the measurement model using tilt photography, law enforcement records, and photo data to develop an oﬃce’s network platform. This could then be used to check, in real time, the law enforcement records for a particular address. These results show that the investigated system can achieve the goals of “seeing from the sky”, “managing on the ground”, and “searching on the Internet”, essential tools for stopping and preventing illegal construction.

7. Discussion In this paper, the DSM data of multitemporal images were used to identify the elevation changes of illegal buildings with remote sensing data of UAV in different periods. With the excellent spatial information analysis ability and data management ability of WebGIS, this research constructed the network information management system for monitoring illegal buildings. By utilizing the timeliness and convenience of mobile terminal, the system was constructed and the grid management operation level of the system was improved. Table1 compares the proposed method with other studies; when remote sensing data of UAV is used, the images with a data accuracy of 0.06 m/pix can be obtained, which is much better than that of satellite remote sensing images of 2.5 m/pix. The improvements of images’ resolution are beneficial to the extraction information of illegal buildings and higher resolutions can lead to having higher extraction accuracy. In terms of the extraction methods of illegal buildings, this research started from the height determinations of the buildings, screened out the changes of the buildings, and then extracted the illegal buildings by human–computer interaction to ensure the accuracy of the illegal building extraction, but the level of automation was low. Through the acquisition and processing of dynamic aerial images, relative information releasing by network, grid management, and collection of relevant building information by mobile APP, the effective supervision of urban illegal buildings can be reached. A three-dimensional model monitoring system based on sky, ground, and Internet can provide strong technical supports for urban constructions. Appl. Sci. 2019, 9, 4954 16 of 18

Table 1. Comparison table of illegal buildings’ identiﬁcation and monitoring research. UAV: unmanned aerial vehicle

Project In This Paper Others [22,23] Basic data UAV remote sensing data (0.06 m/pix) Satellite remote sensing data (2.5 m/pix) Convenience for remote sensing Easy Difficult data Post comparison of multitemporal image classification, change detection Multitemporal elevation difference combining strategies at the pixel and the Illegal building extraction method calculation, man–machine interaction feature levels, building variation pixels were detected by k-means clustering technique Accuracy of illegal buildings’ 78% 75% [23] extraction Dimension of illegal building Three-dimensional Two-dimensional supervision

The identification and monitoring of illegal buildings is a complex task involving multiple disciplines and the resolution of various key issues, such as the corrections of images’ geometry and radiation, feature extraction and description, knowledge expression and integration analysis of multisource data, and knowledge. Along with the support of basic theories such as computer technology, remote sensing technology, image processing, and pattern recognition, the geographic information system for the unified management of geographic information and other attributes’ information has also developed rapidly. Based on these technologies, carrying on spatial information mining and knowledge discovery are the problems that need to be solved at present, and they are some of the hotspots and difficulties of GIS research. How to achieve high-level integration of existing technologies and methods to achieve more efficient work is also an important issue for further research.

8. Conclusions In this study, we successfully used UAV remote sensing for the dynamic monitoring of construction and the collection of monitoring information. This involved three main modules: processing data and identifying illegal buildings; handling the data based on the WebGIS network information management system; and using a mobile grid management terminal system for data communication. The main method for managing illegal buildings was to use a regulatory platform to supervise a series of functions: target image recognition, field verification, investigation and process monitoring, and historical tracking of illegal buildings. When aerial imagery postprocessing software was used, we were able to generate a digital orthophoto map, a digital surface model (DSM), 3D images, and other data. On the basis of preliminary computer selection, supplemented by manual visual inspection, a 3D image and a 720◦ panoramic view provided artificial verification, then the unexpected stronghold was obtained and the corresponding thematic maps were produced. After computer-automated processing, new DSM data were obtained from elevation differences in two-stage images, and illegal buildings could be identified. The investigated algorithm achieved the goals of “seeing from the sky”, “managing on the ground”, and “searching on the Internet”, essential tools for stopping and preventing illegal construction.

Author Contributions: Investigation, Y.H., W.M., Z.M., and C.C.; Methodology, Y.H., Z.L., C.-F.Y., and W.F.; Formal analysis, Y.H., Z.L., W.F., and C.C.; Writing—original draft preparation, Y.H., W.M., C.-F.Y., and Z.M.; Writing—review and editing, C.-F.Y., Y.H., W.M., and C.C. Funding: This study was supported by Longyan City Science and Technology Plan Foreign Cooperation Project of China (Grant No.2018LYF7006) and Open Fund Programs of Key Laboratory of Ecological Environment and Information Atlas (Putian University) Fujian Provincial University of China (Grant No.ST17002). This work was also supported by projects under the Nos. MOST 108-2221-E- 390-005 and MOST 108-2622-E-390-002-CC3. Conﬂicts of Interest: The authors declare no conﬂict of interest. Appl. Sci. 2019, 9, 4954 17 of 18

References

1. Yu, L.; Xie, B.; Chan, E. How does the Built Environment Influence Public Transit Choice in Urban Villages in China? Sustainability 2019, 11, 148. [CrossRef] 2. Deng, Y.; Qi, W.; Fu, B.; Wang, K. Geographical transformations of urban sprawl: Exploring the spatial heterogeneity across cities in China 1992–2015. Cities 2019, 102415. Available online: https://www. sciencedirect.com/science/article/abs/pii/S0264275119300307 (accessed on 18 September 2019). [CrossRef] 3. Liu, Y.; Fan, P.; Yue, W.; Song, Y. Impacts of land finance on urban sprawl in China: The case of Chongqing. Land Use Policy 2018, 72, 420–432. [CrossRef] 4. Akrofi, M.M.; Avogo, F.A.; Wedam, E.A. Sustainable urban development and the challenge of unauthorized buildings: Assessing demographic drivers in peri-urban areas of Ghana. Neth. J. Hous. Environ. Res. 2019, 34, 571–598. [CrossRef] 5. Matikainen, L.; Hyyppa, J.; Ahokas, E.; Markelin, L.; Kaartinen, H. Automatic Detection of Buildings and Changes in Buildings for Updating of Maps. Remote. Sens. 2010, 2, 1217–1248. [CrossRef] 6. Huertas, A.; Nevatia, R. Detecting changes in aerial views of man-made structures. Image Vis. Comput. 2000, 18, 583–596. [CrossRef] 7. Agouris, P.; Mountrakis, G.; Stefanidis, A. Automated spatiotemporal change detection in digital aerial imagery. In Proceedings of the Society of Photo-Optical Instrumentation Engineers (SPIE), Bellingham, WA, USA, 27 July 2000; pp. 2–12. 8. Watanabe, S.; Miyajima, K. Detecting building changes using epipolar constraint from aerial images taken at different positions. In Proceedings of the 2001 International Conference on Image Processing (ICIP), Thessaloniki, Greece, 7–10 October 2001; pp. 201–204. 9. Vu, T.; Matsuoka, M.; Yamazaki, F. LIDAR-based change detection of buildings in dense urban areas. In Proceedings of the 2005 IEEE International Geoscience and Remote Sensing Symposium, Seoul, Korea, 29–29 July 2005; pp. 3413–3416. 10. Marin, C.; Bovolo, F.; Bruzzone, L. Building change detection in multitemporal very high resolution SAR images. IEEE Trans. Geosci. Remote Sens. 2015, 53, 2664–2682. [CrossRef] 11. Bouziani, M.; Goïta, K.; He, D.-C. Automatic change detection of buildings in urban environment from very high spatial resolution images using existing geodatabase and prior knowledge. ISPRS J. Photogramm. Remote. Sens. 2010, 65, 143–153. [CrossRef] 12. Tian, J.; Cui, S.; Reinartz, P. Building change detection based on satellite stereo imagery and digital surface models. IEEE Trans. Geosci. Remote Sens. 2013, 52, 406–417. [CrossRef] 13. Guo, Z.; Du, S. Mining parameter information for building extraction and change detection with very high-resolution imagery and GIS data. GISci. Remote Sens. 2017, 54, 38–63. [CrossRef] 14. Zarco-Tejada, P.J.; Diaz-Varela, R.; Angileri, V.; Loudjani, P. Tree height quantification using very high resolution imagery acquired from an unmanned aerial vehicle (UAV) and automatic 3D photo-reconstruction methods. Eur. J. Agron. 2014, 55, 89–99. [CrossRef] 15. Tian, J.; Reinartz, P.; D’Angelo, P.; Ehlers, M. Region-based automatic building and forest change detection on Cartosat-1 stereo imagery. ISPRS J. Photogramm. Remote. Sens. 2013, 79, 226–239. [CrossRef] 16. Hussain, M.; Chen, D.; Cheng, A.; Wei, H.; Stanley, D. Change detection from remotely sensed images: From pixel-based to object-based approaches. ISPRS J. Photogramm. Remote. Sens. 2013, 80, 91–106. [CrossRef] 17. Chaabouni-Chouayakh, H.; Reinartz, P. Towards Automatic 3D Change Detection inside Urban Areas by Combining Height and Shape Information. Photogramm. Fernerkund. Geoinf. 2011, 2011, 205–217. [CrossRef] 18. Xiao, X.; Javidi, B.; Martinez-Corral, M.; Stern, A. Advances in three-dimensional integral imaging: Sensing, display, and applications. Appl. Opt. 2013, 52, 546–560. [CrossRef][PubMed] 19. Ai, M.; Hu, Q.; Li, J.; Wang, M.; Yuan, H.; Wang, S. A Robust Photogrammetric Processing Method of Low-Altitude UAV Images. Remote. Sens. 2015, 7, 2302–2333. [CrossRef] 20. Aye, Z.C.; Jaboyedoff, M.; Derron, M.H.; Van Westen, C.J.; Hussin, H.Y.; Ciurean, R.L.; Frigerio, S.; Pasuto, A. An interactive web-GIS tool for risk analysis: A case study in the Fella River basin, Italy. Nat. Hazards Earth Syst. Sci. 2016, 16, 85–101. [CrossRef] 21. Aye, Z.C.; Jaboyedoff, M.; Derron, M.-H.; Van Westen, C.J. Prototype of a Web-based Participative Decision Support Platform in Natural Hazards and Risk Management. ISPRS Int. J. Geo-Inf. 2015, 4, 1201–1224. [CrossRef] Appl. Sci. 2019, 9, 4954 18 of 18

22. Li, D.; Wang, M.; Hu, F. Utilizing Chinese high-resolution satellite images for inspection of unauthorized constructions in Beijing. Chin. Sci. Bull. 2009, 54, 2524–2534. [CrossRef] 23. Moghadam, N.K.; Delavar, M.R.; Hanachee, P. Automatic urban illegal building detection using multi-temporal satellite images and geospatial information systems. ISPRS-Int. Arch. Photogramm. Remote. Sens. Spat. Inf. Sci. 2015, 40, 387–393. [CrossRef]

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).