LT. CDR. MELVIN]. UMBACH* US J~SSA CjJasl & Geodetic Survey Rockville, J[d. 20852 for Metric Photogrammetry

The very slight degradation in resolution of bar targets is more than offset by the higher resolution of natural and cultural features.

(.flbslrael on nexi page)

rNTRODUCTION planimetric detail. The superior interpret­ ability of natural and cultural planimetric COAST AND GEODETIC SURVEY has THE detail when color aerial photography was routinely used color aerial photography used considerably uced the amoun t of for analog aerotriangulation, analytic aero­ time that was formerly spent on this phase. triangulation, and map compilation during Between 1958 and 1961, th~. last fi\"e years. Color film is presently from relati\'ely 10\\' altitudes (1,500 to 2,200 utIlIzed for approximately 60 percent of its meters) was used as a su pplemen t to higher­ aerial photography. This percentage will in­ altitude panchromatic photography-the crease as laboratory facilities for processing high-altitude photography being used for and printing color materials are expanded. aerotriangulation and general map compila­ and infrared photography will tion, and low-altitude color being used to probably continue to be used for the delinea­ photoidentify and locate important chart tion of the low-water line in the Coast Survey features such as landmarks, hydrographic coastal mapping program. Panchromatic survey signal sites, aids to navigation, sub­ emulsions on glass plates will be used for the merged hazards to navigation, and the deline­ ma!ority of geodetic point and boundary lo­ ation of shoal areas. catIOn work, wherein specific Each successive year brought new progress targets are to be precisely located by photo­ and a more promising future for the metric g.rammetric methods. All other aerotriangula­ application of color aerial photography. By tlOn and map compilation requirements will 1961, color aerial films with higher resolving be fulfilled through the use of color photog­ raphy. When the Coast and Geodetic Survey first used color aerial photography nine years ago, the available films were extremely slow and had such low resolution characteristics that very low-altitude photography was required in order to record ground images adequately. Even wi th this tem porary handicap, the addi­ tional dimension of color proved to be an in­ valuable asset which became mandatory for the Coastal Mapping Program. Field edit or field completion of photogrammetrically com­ piled manuscripts was an extremely costly phase of the mapping program when panchro­ matic photography was used because it was necessary for a field man to walk along most of the shoreline to clarify and interpret the * Presented at the Annual Convention of the American Society of Photogrammetry, Washing­ ton, D. c., March 1967 under the title "Color and Metric Photogrammetry." LT. CDR. MELVIN J. UMBACH 265 266

power a nd greater speed becamc a\'ailable. of a total color system which could completely These films proved to be adequate for pri­ rcplace thc panchromatic systcm for aero­ mary mapping photography and control triangulation and stereoscopic map compila­ densification by aerotriangulation in coastal tion. areas at flight altitudes up to 4,500 meters, In J963 color photography was successfully Between 1961 and 1963, color emulsions introduced to the Bureau's Airport Survey were available only on acetate bases. whe"eas Program, and used for aerotriangulation panchromatic films were available on the where a Right height of approximately 5,500 more stable p)lyester and polycarbonate meters is required to obtain adequate ground bases. In addition, a color emulsion on glass­ coverage with a minimum amount of ground plate disapositives was not available, making control survey work, it necessary to do aerotriangulation and Precision analytic photogrammetry has stereoscopic compilation with black-and­ been developed to the point that, aside from white disapositives made from the original the day-to-day instability of the comparators color films. Despite these temporary draw­ used to measure coordi nates (at most, 2 backs, the metric quality of color film was microns), all errors in aerotriangulation are

ABSTRACT: The Coast and Geodetic Survey has been using color photography for aerotriangulation and map compilation during the last five years. Extensive tests have continually been performed during tltis period to evaluate the metric fidelity of several brands of color film. These tests included: gn:d f:xposure stndies for the determination of differential film distortion properties and diapositive plate emulsion creep; strip aerotriangulation error propagation studies using true film distortion values applied to fictitious photographic coordinates; air­ borne image resolution tests; and analytic aerotriangulation accuracy tests using recent color and panchromatic aerial photography taken over the JlcClure, Ohio, camera calibration test site, The results of these tests indicate no significant #fference in the metric stability between tlte color and the panchromatic film systems.

firmly established to be entirely adequate for attributable to unknown or uncontrollable constructing maps meeting national accuracy errors in the cent,'al perspective of the aerial standards when aerotriangulation was per­ photograph diapositi\'e. During the last fl\'e formed wi th first-order stereoplotters and years conside"able effort has becn expended comparators at photographic scales as small on the isolation, analysis, and compensation as 1/3 of the scale of the published map. of systematic errors in the aerial photo­ The two missing links in the complete in­ graph. tegrated color mapping system were pro­ DISTOIlTlO~ vided in 1962 and 1963. First, the direct FlU' STUDIES viewing stereoplotter for map compilation Film distortion, universally rccognized as was introduced to the photogrammetric one of the prime sources of photogrammetric community; then glass disapositi\'es with error, was gi\'en much attention throughout color emulsions that permitted the preserva­ our program to impro\'e the metric fidelity of tion of a stable reproduction of the original the central perspective.' The results of this aerial exposure were developed and marketed, majo,' effort led to the unmistakable conclu­ Shortly thereafter, color film became avail­ sion that the color films marketed by the two able on the same polyester and polycarbonate leading manufacturers are metrically equal base material that is presently used with and indistinguishable from the relatively panchromatic aerial film, During this period stable Plus-X Aerographic panchromatic the resolving power of aerial color films was film on a 0.004-inch polyester base. This con­ also improved to the point of nearly equaling clusion holds for the case where contact dia­ that ofLthe standard panchromatic aerial positives are used in com'entional optical­ films for all combinations of natural mechanical stereoplotters, as well as the including black-and-white resolution targets, analytic aerotriangulation case where mathe­ These developmen ts ful filled the req uiremcnts matical models for the compensation of film COLOR FOR :\1ETIUC I'IfOTOGR.\MMI·:TRY 267

graph, require several explanations and 11------220 mm------_·I qualification. + + The first three values for reseau spacings shown on the abscissa can be correlated wi th the number of fiducial images used for film distortion compensation. TnfLIlite reseau spacing, or zero reseau marks, simulates the condition of uncompensated film, i.e., film used with analog plotting equipment where the compensation for differential distortion +ft is impossible. The four-corner fiducial mark +f+ case is identical to a 22-cm. reseau spacing. and the 11 cen ti meter reseau very nearly approxi mates the eigh t fid ucial mark case. A true ll-cm. reseau would require a centrally located mark. A very important difference exists between optically imaged fid ucial marks and reseau grid images within the photograph area. + Each fid ucial mark is optically projected through a short focal-length lens system by FIG. 1. Calibration grid for r.lm an individual source thereby producing distortion studies. a high-contrast, high-resolution image on a uniform background. On the other hand, distortion are defined, depending upon the grid spacings less than ll-cm. require the use camera used, through the displacemen t of of a camera reseau in contact with the emul­ four or eight fiducial marks exposed on the sion or projected either by the camera lens or photographic format. through the film base. Cn any case, the multi­ Exhaustive film distortion studies were tone terrain on an exposed photograph re­ conducted by means of a glass plate which places the high-con trast, high-resolution had throughout the photographic format image on a uniform background that is pro­ numerous grid intersections whose coordi­ duced by internally projected fiducial marks nates were precisely calibrated to a standard and the high-quality grid images used for error of approximately 1 micron2 (see Figure this laboratory test. 1). This glass grid was contact printed The shaded portion at the left side of the throughou t \'arious sam pie panchroma tic chart indicates the standard-error zone for and color film rolls, through the use of a the measurement of these high-quality pro­ vacuum frame. Six exposures of the grid from jected fid ucial marks. The shaded portion to each sample roll were then transferred to the right of the 6-cm. grid spacing is, in OUI' glass diapositi\'es following the normal film opinion, a conservative estimate of the stan­ negati\'e aging period of six weeks and sub­ dard-error zone for aerial camera reseau sequently measured with a precisely cali­ image measurement. This limitation is brated Mann comparator. pointed out here to indicate that one ap­ The grid distortion measurement data were parently cannot take advantage of the small initially used to compare the ability of color and panchromatic aerial films to preserve ;;;- 10 -r------, their metric characteristics through the pro­ z ~ - PANCHROMATIC SAMPLE cessing phases. Secondly, these data pro\'ided u i"l'C"~- the means to test the capabilities of color and i ~:-:-..... •--COlORSAMPlE... COLOR SAMPLE panchromatic films to respond to prO\'en g 5 :-"'S;;, .• techniques of film distorted compensation. w ~ • o .....~ Figure 2A demonstrates the gross ftl m dis­ ~ o tortion encountered along \\·ith effects and Z ""•.,..,.,.:-:.':'::-:.:: .'7'. ::-••7: ;'; effectiveness of various fiducial mark systems '" and reseau spacings on the ability to recover 22 em 11 em 6 Uti .. em 2,m I,m a grid coordinate at any point within the en­ 101 14 1 lSI FIDUCIAL MARKS tire photographic format for each film sam­ FIG. 2A. Reseau spacing versus standard ple. The results of this test, as shown on the error of all poi nts. 268 PHOTOGRAlIlMETRTC ENGINEER.J NG gain in accuracy which could theoretically be 3S-r------, -- PANCHROMATIC SAMPLE obtained by using reseau spacings of less -- COLOR SAMPLE than 6 m...... COLOR SAMPLE From the metric standpoint, the pass-point errors are of prime concern because they are the source of error propagation throughout an aerotriangulation bridge performed in either the analog or analytic mode. Figure 2B demonstrates the standard error of recovery of a point in the most probable pass point .... areas for each film sample after various types of distortion compensation have been ap­ a'" plied. '" '"w The data from the grid studies were fur­ o ...... ther utilized to construct three fictitious '"« .... , o I , strips of six aerial photographs with 60 per­ Z , ~ , . cent endlap for the purpose of observing V1 , error propagation caused solely by true LINEAR NON.lINEAR differential film distortion and random com­ parator measuring error. As the grids were contact printed in the laboratory under con­ AFFINE 8 AFFINE trolled conditions, any addi tional propaga­ FIDUCIAL FIDUCIAL tion of error throughout the simulated strip caused by atmospheric refraction, charac­ FIG. 3. Standard error of position determined by aerolriangulatioll applying the film-distortion teristic camera lens errors, or other unknown compensation technique. external sources of systematic errors, is ab­ sent. Rou tine Coast and Geodetic Survey analyt­ ic aerotriangulation techniques were ap­ relatively free, unrestrained manner, whereby plied to the simulated strip coordinates for the photogrammetric bundles are not dis­ the photogrammetric solution of ground co­ torted or deformed to fit other control points ordinates. Figure 3 shows a typical horizontal spaced throughout the strip. The standard and vertical error propagation in the com­ errors of the plate coordinates resulting from puted ground coordinates at unit scale for the analytic relative orientation phase of the each of the three types of film tested. For aerotriangulation process were well under 1 this particular error-propagation plot, the micron. Therefore, image displacemen t caused image coordinates were compensated for film by film distortion and measuring error was, distortion through an eight-fiducial-mark ad­ in fact, transformed into the orientation justment. Horizontal control points were parameter errors that caused the obsel'ved selected at the midpoints of the ends of the error propagation. strip and vertical control points at each cor­ No significant difference existed in the error ner. The selection of this particular array of propagation properties of the panchl'Omatic control permits the propagation of error in a sample and the two colol' samples used in this test. Figure 4 shows the resultant three-di­ -.;; '0 mensional standard errors of position for the z " panchromatic and two color stri ps after ~ - PANCHROMATIC SAMPLE u -- COLOR SAMPLE various types of film distortion compensation i· ...• COLOR SAMPLE and techniques for adjustment to ground § control. '"w , The results demonstrate: (1) the improve· ~ ment in accuracy if eight fiducial marks are « 0 z used to compensate differential film distor­ ~ tion; (2) the merits of a non-linear adjust­ V1 0 22 em 11 em ment of model coordinates to ground control; 101 1"1 (81 FIDUCIAL ..... ARKS and (3) the lack of significant difference in FIG. 2B. Reseau spacing versus standard photogrammetric accuracy between pan­ error of pass points. chromatic and color photography. COLOR FOR METRIC PHOTOGRAMMETRY 269

,,~ 1.-;---":'':'-',:, . _ r:'--'-'-'-'-l I: '. I: , '. I: ': I, I'

VERTICAL -- PANCHROMATIC SAMPLE ...... COLOR SAMPLE ____ COLOR SAMPLE FIG. 4. Error propagation caused by film distortion.

STABILITY OF COLOR GLASS DIAPOSITIVES photography. Modern wide-angle camera lenses are now designed for the use of the en­ The diapositive printing process was also tire , a quality which is as considered as a possible source of loss of necessary for high-q uali ty panchromatic metric fidelity in the color system. Due to photography as it is for color. Three well­ the time required to conduct physically a set known aerial cameras with modern color cor­ of coordinate measuremen ts of a plate or rected lenses ha\'e been tested wi th color stereoscopic model, the work. is normally film to determinc the inAuence of residual p.erformed on stable glass-plate reproduc­ tions of the original exposures. It is, there­ chromatic aberration on the metric fidelity of aerial photography taken with panchro­ fore, important that diapositive plates be matic and color cmulsions. Each camera ex­ kept relatively free from image shift through­ hibited a slight lack of spectral balance for out the exposure, chemical processing, drying, white objects imaged beyond 40 degrees from and measurement phases of their use. the optical axis. Howe\'er, no geometrical The color glass diapositives now marketed were tested and found to be metrically equal displacement could bc detected between im­ ages of red, , bluc, gray, or white tar­ to the standard microAat glass diaposi ti ves used in the panchromatic system. The pro­ gets. I t is in teresti ng to note that the order in cedures and techniques used for thesc tests which the three sensitized color layers arc were nearly identical to those described in arranged on the film base ( on top, grecn the previous section for the evaluation of in the middle, and red on the bottom) tends gross film distortion. The standard crror of a to compensate, ra/lter than contribute to, image coordinate after transfer to a color glass diapositive was found to be Icss than 2 d1:splacement caused by chromatic aberration. If, in the futurc, a more thorough labora­ microns. This standard error level is within one's capability to recover a point on tory test indicates that significant image displacements occur due to chromatic lens panchromatic diapositives if comparator aberration, the condition must be common pointin~, calibration error, and emulsion creep for both color and panchromatic emulsions. IS conSIdered. I t is not surprising that this Ob\'iously, the color of the object must be high metric quality was achieved a the color known in order to com pensate properly for emulsion is coated on a film base which is this type of systematic distortion. subsequen tly laminated or fastened to the These tests, as well as others designed to glass by an adhesive during the plate manu­ determi ne and establish the optimu m size, facturing process. shape, and reAecti ve characteristics of targets for premarking ground points, verified the AERIAL CA~IERAS A 'D COLOR PHOTOGRAPHY importance of using small, symmetrical, Aerial camera lens distortions can be as­ medium-contrast targets to a\'oid metric sumed to be negligible when comparing the error caused by image spread which is also a metric fideli ty of panchromatic and color function of photographic exposure. This pre- 270 PHOTOGRAMMETl{IC ENGINEI~RING caution applies as well to panchromatic glass plate expo ures. Each flight line con­ photography. sists of a 1; 35,OOO-scale, sixty-percen t endlap The metric q uali ty of aerial photography stereotriplet, the center photograph of which is influenced to some degree by the resolving covers a densely controlled four-mile square power, or ability, of the photographic emul­ area containing 85 geodetic control points. sion to record fine detail. I t has been our Pinpoint exposures with both film and glass experience that the resolution of aerial plates provided a nearly iden tical coverage of color film, if used in wide-angle photogram­ conjugate photographs in each strip. metric cameras, is nearer to that of panchro­ Standard processing techniques were used matic films than the film manufacturers have to develop the panchromatic and color film indicated in their literature. This may, of strips. After curing the film negatives for six course, be attributable to characteristics of weeks, each exposure was transferred to a the particular wide-angle camera-film com­ glass diapositive plate. Diapositives were also binations used by the Coast and Geodetic printed from the glass-plate exposures be­ Survey flight missions. Recently, variable­ cause the target images on the negatives were altitude color and panchromatic aer'ial photo­ too dense for coordinate measurement of the graphs were taken of the color and bar re­ highest preci ion. solution targets at \iVright-Patterson Air Image coordinate measurements were car­ Force Base. These photographs demon­ ried out in the stereo mode on a Zeiss PSK strated a maximum resolving power of 48 stereocomparator. Previous calibration of lines per millimeter for color film and 54 this instrument showed little or no X- Y-scale line per millimeter for panchromatic film. differential or non-orthogonality of the After examination of many color aerial axes. The measured coordinates were then photographs of various types of terrain, it is processed through the Coast and Geodetic our opinion that the very slight degradation Survey Coordinate Refinement Computer in the resolution of black-and-white repeti­ Program3 which compensates for systematic tive bar targets is more than offset by the errors caused by atmospheric refraction, lens higher resolution of a large percentage of distortion, and film distortion for the film natural colors which are more frequent in samples. both man-made and natural cui tural fea­ Photogrammetric ground coordinates for tures. each targeted station were computed from the refined plate coordinates of each stereo­ AIRBORNE METRIC TESTS triplet by the Coast and Geodetic Survey OF COLOR PHOTOGRAPHY Block Adjustment Computer Program.4 A '1'\\'0 stri ps of near-vertical color aerial minimum of two horizontal control points photography were recen tly taken over the and three vertical control points were held McClure, Ohio, camera test site for a prac­ in the adjustment thereby permitting a free, tical study of the metric capabili ties of the unrestrained photogrammetric solution. To total color system under operational condi­ eliminate the propagational effects of local tions. Three strips of panchromatic film and errors in the iden tification of any of the con­ three strips of panchromatic glass-plate trol stations used in the unrestrained solu­ photography were also taken over the same tion, a least squares, three-dimensional linear flight line for use as a basis of comparison. adj ustmen t based on all known con trol The Ohio camera test site, originally de­ poin ts was made. This fi nal adj ustment per­ signed and established as the calibration formed on the photogrammetric block ad­ area for the obsolete nine-lens aerial camera, justment of the ground coordinates for each includes 107 targeted triangulation stations strip of each sample tended to normalize the within a five-mile square. Circular white con­ results for a more comprehensive comparison crete pads, six and eight feet in diameter, and interpretation. mark the control stations which are of first­ Table 1 shows the standard errors of the and second-order accuracy. Second-order level plate residuals after block adjustment, and lines provide precise elevations for each sta­ the standard errors of the photogram metri­ tion and marker. cally determined ground coordinates resulting The photography was taken with a Wild from each panchromatic film, color film, and RC-8 aerial camera equipped with a Uni­ glass plate stereotriplet. versal Aviogon lens and eight fiducial marks. The standard errors of plate coordinate A special glass plate camera back was used residuals cannot be considered as absolute in lieu of the camera film magazine for the indicators of aerotriangulation precision. CULUR FUR J\lETRIC Pl-]UTUGR.\J\lJ\lETRY 271

TABLE 1. AEROTR1ANGULATION BLOCK ADJUSTMENT STANDARD ERRORS

Sa1ll pie Strip Plate Ground-Horizontal Ground- VertiCIl I

1l1icrons meters meters Color film-strip #1 1.3 0.223 0.236 2 1.6 0.291 0.298

1l1ICrons meters meters Panchromatic film-strip #1 1.7 0.256 0.225 2 1.5 0.233 0.230 3 1.9 0.229 0.457

InicrOll:-; Illeters meters Glass plate-strip #t .2 0.226 0.166 2 .1 0.214 0.186 3 .1 0.215 U.20U

The point residual itself reflects the failure plate multiplied by the photographic scale of the point to satisfy the collinearity condi­ Although these values do not rigorously de­ tion. Because the analytic solution is rela­ fine ground position precision, they serve as tively elastic, some of the errors on the plate good approxi mations as the maxi mum differ­ will be transformed into orientation param­ ence in terrain elevation throughout the en­ eter adjustments which naturally tend to tire test area is less than ten feet. reduce the magnitude of the standard errors. To what degree or magnitude the actual stan­ CONCLUSIONS dard errors of plate residuals have been re­ duced in these block adjustments is unknown. The superiority of color aerial photography In addition to a block adjustment of each for the accurate interpretation of natural sample strip, a least-squares resection was and cultural planimetric detail cannot be performed on the cen ter photograph of each challenged. stereotriplet for a further comparison of the The results of the film-distortion studies, metric quality between panchromatic and the color-diapositive stability tests, and the color aerial film. statistics based on the aerotriangulation of Table 2 shows the standard errors of pia te color aerial photography over the Ohio test residuals after resection of the \'arious center area, have provided concrete evidence that plates. The standard errors of posi tion on the color aerial film is metrically equivalent to ground are simply the standard errors on the panchromatic aerial film.

TABLE 2. AEIWTRIANGULATlON RESECTION STAI\"DAIlD ERRORS

Center Exposure Plate Ground-H or izontal

microns meters Color film strip #1 7.3 0.256 2 8.1 0.28+

microns meters Panchromatic film strip #1 8.+ 0.29+ 2 8.5 0.298 3 9.2 0.322

ll11crOilS 1l1eler~ Glass-plate strip #1 6.6 0.231 2 6.4 0.22+ 3 6.5 0.228 272 PHOTOGRAMMETRIC ENGINEERING

REFEIlENCES 3. Keller, M., and Tewinkel, H. c., "Aerotrian­ gulation: rmage Coordinate Refinement," Coast 1. Lampton, B. F., "Film Distortion Compensa­ and Geodetic Survey Technical Bulletin No. 25, tion," PhotogrmlL'llletric Engineering, Vol. XXXI, March,1965. :\'0.5, September, 1965. -t. Keller, M., and Tewinkel, G. c., "Aerotriangu­ 2. Lampton, B. F., and Umbach, M. J., "Film lation: Block Adjustment in the Coast and Distortion Compensation Errectiveness," Photo­ Geodetic Survey," Coast and Geodetic Survey grammetric Engineering, Vol. XXXII, No.6, Technical Memorandwl1 No.5, December 1967. i\'ovember, 1966.

RAVEN presents a new STRONGER POINTER expressly for PHOTOGRAMMETRISTS It makes stronger pencil points terminate precisely on the center axis of the lead holder without fuss or bother. The points are actually stronger (about 20%) because fine striations spiral, rather than go round the point. It is unexcelled for drafting pencils and also makes excellent chisel points. See it in use at the March ASP-ACSM convention with a WilD 88 stereo­ plotter, or write for descriptive literature and prices. RAVEN MFG. & SALES CO. 5390 W. 1st Ave. Denver, Colo. 80226