ISBN: 978-1-4799-5302-8

2014 6th International Conference on Information Technology and Electrical Engineering

2014 6th International Conference on Information Technology and Electrical Engineering

Department of Electrical Engineering and Information Technology Faculty of Engineering, Universitas Gadjah Mada Jalan Grafika 2 Yogyakarta 55281, ICITEE 2014 Partners and Sponsors

Organized by

Co-organized by

Technically Co-Sponsored by

Supported by Welcome Message from the General Chair

On behalf of the organizing committee, it is our pleasure to welcome you to Yogyakarta, Indonesia, for our annual conference. This is the 6th conference that is held by the Department of Electrical Engineering and Information Technology, Faculty of Engineering, Universitas Gadjah Mada. This year, the conference is differently called as Joint conference 2014 as there will be 4 parallel conferences, including: 1. ICITEE (International Conference of Information Technology and Electrical Engineering) 2014, 2. CITEE (Conference of Information Technology and Electrical Engineering) 2014, 3. RC-CIE (Regional Conference on Computer and Information Engineering) 2014, and 4. CCIO (Conference on Chief Information Officer) 2014.

The joint conference’s theme is “Leveraging Research and Technology through University-Industry-Government Collaboration” emphasizes on the enhancement of research in a wide spectrum, including information technology, communication and electrical engineering, as well as e-services, e-government and information system. The conference is expected to provide excellent opportunity to meet experts, exchange information, and strengthen the collaboration among researchers, engineers, and scholars from academia, government, and industry.

In addition, the conference committee has invited five renowned keynote speakers, Prof. Marco Aiello from University of Groningen (RuG), Netherland, Prof. Einoshin Suzuki from Kyushu University, Prof. Yoshio Yamamoto from Tokai University, Prof. Jun Miura from Toyohashi University of Technology, and Prof. Kazuhiko Hamamoto from Tokai University, Japan. The conference committee also invited Tony Seno Hartono from National Technology Officer of Microsoft Indonesia and Dr. Ing. Hutomo Suryo Wasisto (Associate Team Leader in MEMS/NEMS and Sensor Group) Technische Universität Braunschweig, Germany as Invited speaker to present their current research activities.

This conference is technically co-sponsored by IEEE Indonesia Section. Furthermore, it is supported by JICA, AUN/SEED-Net, Ministry of Communication and Information Technology of The Republic of Indonesia, and King Mongkut’s Institute of Technology Ladkrabang, Thailand.

As a General Chair, I would like to take this opportunity to express my deep appreciation to the organizing committee members for their hard work and contribution throughout this conference. I would also like to thank authors, reviewers, all speakers, and session chairs for their support to Joint Conference 2014.

In addition to the outstanding scientific program, we hope that you will find time to explore Yogyakarta and the surrounding areas. Yogyakarta is city with numerous cultural heritages, natural beauty, and the taste of traditional Javanese cuisines, coupled with the friendliness of its people.

Lastly, I would like to welcome you to Joint Conference 2014 and wish you all an enjoyable stay in Yogyakarta.

Sincerely,

Hanung Adi Nugroho, Ph.D. General Chair of Joint Conference 2014

-i- Welcome Message from the TPC Chair

On behalf of the technical program committee (TPC), we warmly welcome you to the 6th International Conference on Information Technology and Electrical Engineering (ICITEE 2014) in the cultural city of Yogyakarta, Indonesia. The committee has organized exciting technical programs for ICITEE 2014 with conference theme of “Leveraging Research and Technology through University-Industry Collaboration.” As an annual International conference, ICITEE provides excellent platform to share innovative idea and experiences, exchange information, and explore collaboration among researchers, engineers, practitioners and scholars the field of information technology, communications, and electrical engineering.

All 163 submitted papers from 18 countries throughout the world went through a rigorous review process and each paper was evaluated by at least three independent reviewers in accordance with standard blind review process. Based on the results of the rigorous review process, 78 papers have been selected, which constitute the acceptance rate of 47.9%. These papers have been grouped into 5, ranging from the fields of information technology, communications, power systems, electronics, and control systems. Besides those regular sessions, ICITEE 2014 also features world-class keynote/plenary speeches and distinghuish-invited speakers that reflect the current research and development trends in the aforementioned fields.

We are deeply indebted to all of our TPC members as well as our reviewers, who volunteered a considerable amount of their time and expertise to ensure a fair, rigorous, and timely review process. Many thanks should be given to our keynote and invited speakers who will share their experience in this conference. Last but not least, our sincere gratitude should be given to all authors for submitting their work to ICITEE 2014, which has allowed us to assemble a high quality technical program.

Welcome to Yogyakarta and hope you will enjoy a wonderful experience in this traditional city of Indonesia.

With best regards,

TPC Chair

-ii- Committees

Advisory Board Committee Adha Imam Cahyadi (Universitas Gadjah Mada, Indonesia) Adhi Susanto (Universitas Gadjah Mada, Indonesia) Adhi Susanto (Universitas Gadjah Mada, Indonesia) Dadang Gunawan (Universitas Indonesia, Indonesia) Alagan Anpalagan (Ryerson University, Canada) Yanuarsyah Haroen (Institut Teknologi Bandung, Amirthalingam Ramanan (University of Jaffna, Sri Indonesia) Lanka) Kuncoro Wastuwibowo (IEEE Indonesia Section) Andy Warner (Google) T. Haryono (Universitas Gadjah Mada, Indonesia) Anto Satriyo Nugroho (BPPT, Indonesia) Chanboon Sathitwiriyawong (KMITL, Thailand) Anton Satria Prabuwono (Universiti Kebangsaan Hidekazu Murata (Kyoto University, Japan) Malaysia, Malaysia) Ruttikorn Varakulsiripunth (Thai-Nichi Institute of Ardyono Priyadi (Institute of Technology Sepuluh Technology, Thailand) Nopember, Indonesia) Lukito Edi Nugroho (Universitas Gadjah Mada, Armein Z. R. Langi (Bandung Institute of Technology, Indonesia) Indonesia) Son Kuswadi (PENS, Indonesia) Awinash Anand (Kyushu University, Japan) Azwirman Gusrialdi (University of Central Florida, USA) Boonprasert Suravkratanasakul (KMITL, Thailand) Organizing Committee Chalermsub Sangkavichitr (KMUTT, Thailand) Sarjiya (Universitas Gadjah Mada, Indonesia) Chanboon Sathitwiriyawong (KMITL, Thailand) Eka Firmansyah (Universitas Gadjah Mada, Chotipat Pornavalai (KMITL, Thailand) Indonesia) Cuk Supriyadi Ali Nandar (BPPT, Indonesia) Hanung Adi Nugroho (Universitas Gadjah Mada, Dhany Arifianto (Institute of Technology Sepuluh Indonesia) Nopember, Indonesia) I Wayan Mustika (Universitas Gadjah Mada, Dhomas Hatta Fudholi (La Trobe University, Australia) Indonesia) Eiji Okamoto (Nagoya Institute of Technology, Japan) Adha Imam Cahyadi (Universitas Gadjah Mada, Ekachai Leelarasmee (Chulalongkorn University, Indonesia) Thailand) Sigit Basuki Wibowo (Universitas Gadjah Mada, Esa Prakasa (LIPI, Indonesia) Indonesia) F Danang Wijaya (Universitas Gadjah Mada, Kuntpong Woraratpanya (KMITL, Thailand) Indonesia) Prapto Nugroho (Universitas Gadjah Mada, Fahkriy Hario P (Universitas Brawijaya, Indonesia) Indonesia) Fikri Waskito (Nanyang Technological University, Teguh Bharata Adji (Universitas Gadjah Mada, Singapore) Indonesia) Gamantyo Hendrantoro (Institute of Technology Sumet Prabhavat (KMITL, Thailand) Sepuluh Nopember, Indonesia) Natapon Pantuwong (KMITL, Thailand) Gunawan Wibisono (Indonesia University, Noor Akhmad Setiawan (Universitas Gadjah Mada, Indonesia) Indonesia) Gusti Agung Ayu Putri (, Indriana Hidayah (Universitas Gadjah Mada, Indonesia) Indonesia) Harris Simaremare (Universite de Haute Alsace, Kitsuchart Pasupa (KMITL, Thailand) France) Avrin Nur Widiastuti (Universitas Gadjah Mada, Haiguang Wang (Institute for Infocomm Research, Indonesia) Singapore) Teerapong Leelanupab (KMITL, Thailand) Haruichi Kanaya (Kyushu University, Japan) Iswandi (Universitas Gadjah Mada, Indonesia) Heroe Wijanto (Institut Teknologi Telkom, Indonesia) Budi Setiyanto (Universitas Gadjah Mada, Indonesia) Hutomo Suryo Wasisto (Technische Universität Bimo Sunarfri Hantono (Universitas Gadjah Mada, Braunschweig, Germany) Indonesia) I Ketut Gede Dharma Putra (Udayana University, Yusuf Susilo Wijoyo (Universitas Gadjah Mada, Indonesia) Indonesia) Agus Bejo (Universitas Gadjah Mada, Indonesia) I Made Yulistya Negara (Institute of Technology Husni Rois Ali (Universitas Gadjah Mada, Indonesia) Sepuluh Nopember, Indonesia) Azkario Rizky Pratama (Universitas Gadjah Mada, I Nyoman Satya Kumara (Udayana University, Indonesia) Indonesia) Lilik Suyanti (Universitas Gadjah Mada, Indonesia) I Putu Agung Bayupati (Udayana University, Nawang Siwi (Universitas Gadjah Mada, Indonesia) Indonesia) Ida Ayu Dwi Giriantari (Udayana University, Indonesia) Igi Ardiyanto (Toyohashi University of Technology, Technical Program Committee Japan) Issarachai Ngamroo (KMITL, Thailand) Addy Wahyudie (United Arab Emirates University, Ivanna Timotius (Satya Wacana Christian University, UEA) Indonesia)

-iii- Iwan Setiawan (Satya Wacana Christian University, Surapan Airphaiboon (KMITL, Thailand) Indonesia) Surin Kittitornkun (KMITL, Thailand) Jaziar Radianti (University of Agder, Norway) Suvapadee Aramvith (Chulalongkorn University) Joko Siswantoro (Universitas Surabaya, Indonesia) Teerapong Leelanupab (KMITL, Thailand) Kang-Hyun Jo (Ulsan University, Korea) Thanisa Numnonda (KMITL, Thailand) Kazunori Hayashi (Kyoto University, Japan) Thippaya Chintakovid (KMUNB, Thailand) Kazuto Yano (ATR, Japan) Udayanto Dwi Atmojo (The University of Auckland, Khoirul Anwar (Japan Advanced Institute of Science New Zealand) and Technology, Japan) Uke Kurniawan Usman (ITTelkom, Indonesia) Kitsuchart Pasupa (KMITL, Thailand) Umar Khayam (Bandung Institute of Technology, Koji Yamamoto (Kyoto University, Japan) Indonesia, Indonesia) Kuntpong Woraratpanya (KMITL, Thailand) Virach Sornlertlamvanich (NECTEC, Thailand) Lesnanto Multa Putranto (Hokkaido University, Wanchalem Pora (Chulalongkorn University) Japan) Wawta Techataweewan (Srinakharinwirot University, Maleerat Sodanil (KMUNB, Thailand) Thailand) Mamiko Inamori (Tokai University, Japan) Wayan Gede Ariastina (Udayana University, Indonesia) Manop Phankokkruad (KMITL, Thailand) Wimol San-Um (Thai-Nichi Institute of Technology, Marco Aiello (University of Groningen, Netherland) Thailand) Maulahikmah Galinium (Swiss German University, Widyawan (Universitas Gadjah Mada, Indonesia) Indonesia) Yasushi Kato (Tsuruoka National College and Mauridhi Hery Purnomo (Institute of Technology Technology, Japan) Sepuluh Nopember, Indonesia) Yi Ren (National Chiao Tung University, Taiwan) Mustansar Ghazanfar (University of Southampton, Yoshikazu Washizawa (The University of UK) Electro-Communications, Japan) Natapon Pantuwong (KMITL, Thailand) Yoshimitsu Kuroki (Kurume National College of Nawat Kamnoonwatana (Mahidol University, Technology, Japan) Thailand) Nidapan Sureerattanan (Thai-Nichi Institute of Technology, Thailand) Nihan Tran (University of Melbourne, Australia) Nol Premasathian (KMITL, Thailand) Noopadol Maneerat (KMITL, Thailand) Nopporn Chotikakamthorn (KMITL, Thailand) Olarn Wongwirat (KMITL, Thailand) Pattaracha Lalitrojwong (KMITL, Thailand) Pichai Aree (Thamassat University, Thailand) Poramote Wardkein (KMITL, Thailand) Pornvalai Chotipat (KMITL, Thailand) Ramesh Pokharel (Kyushu University, Japan) Rohana Sapawi (University Malaysia Sarawak, Malaysia) Rony Seto (Institute of Technology Sepuluh Nopember, Indonesia) Rukmi Sari Hartati (Udayana University, Indonesia) Ruttikorn Varakulsiripunth (Thai-Nichi Institute of Technology, Thailand) Sakchai Thipchaksurat (KMITL, Thailand) Saiyan Saiyod (Khon Kaen University, Thailand) Selo Sulistyo (Universitas Gadjah Mada, Indonesia) Singha Chaveesuk (KMITL, Thailand) Sisdarmanto Adinandra (Universitas Islam Indonesia, Indonesia) Somjet Suppharangsan (Burapha University, Thailand) Sompong Valuvanathorn (Ubon Ratchathani University, Thailand) Sooksan Panichpapiboon (KMITL, Thailand) Soradech Krootjohn (KMUNB, Thailand) Sorawat Chivapreecha (KMITL, Thailand) Sumet Prabhavat (KMITL, Thailand) Sunisa Rimcharoen (Bhurapa University, Thailand) Sunu Wibirama (Tokai University, Japan) Supakit Nootyaskool (KMITL, Thailand) Supawan Amanab (KMITL, Thailand) Supot Nitsuwat (KMUNTB, Thailand) Surachai Chaitusaney (Chulalongkorn University, Thailand)

-iv-

Technical Sessions

 Session 1. Software Engineering, Services, and Information Technology

TS 1 – 01 A Hybrid Technique for Enhancement of Periductal Fibrosis Ultrasound Images for Cholangiocacinoma Surveillance ...... 2 Pichet Wayalun (Khon Kaen University, Thailand); Saiyan Saiyod (Khon Kaen University, Thailand) TS 1 – 02 A Real Time Mission-Critical Business Intelligence for Development of Mixture Composition on Aromatherapy Product Based on Customer Personality Type ...... 7 Taufik Djatna (Bogor Agricultral UnIversity, Indonesia); Ida Bagus Dharma Yoga Santosa (Bogor Agriculture University, Indonesia) TS 1 – 03 An Infrastructure-less Occupant Context-Recognition in Energy Efficient Building ...... 13 Azkario Rizky (Universitas Gadjah Mada, Indonesia); Widy Widyawan (, Indonesia); Guntur Putra (Universitas Gadjah Mada, Indonesia) TS 1 – 04 An Integrated Model of Customer Repurchase Intention in B2C E-commerce ...... 19 Saowakhon Homsud (King Mongkut’s Institute of Technology Ladkrabang, Thailand); Singha Chaveesuk (King Mongkut’s Institute of Technology Ladkrabang, Thailand) TS 1 – 05 An Intuitive User Interface for Motion Retrieval on a Mobile Multi-touch Device ...... 25 Natta Tammachat (King Mongkut's Institute of Technology Ladkrabang, Thailand); Natapon Pantuwong (King Mongkut's Institute of Technology Ladkrabang, Thailand) TS 1 – 06 Automated Document Classification for News Article in Bahasa Indonesia Based on Term Frequency INVERSE Document Frequency (TF-IDF) Approach ...... 29 Ari Aulia Hakim (Swiss German University, Indonesia); Alva Erwin (Swiss German University, Indonesia); Kho Eng (Swiss German University, Indonesia); Maulahikmah Galinium (Swiss German University, Indonesia); Wahyu Muliady (Akon Teknologi, Indonesia) TS 1 – 07 Automatic Leaf Color Level Determination for Need Based Fertilizer Using Fuzzy Logic on Mobile Application ...... 33 Kestrilia Prilianti (Universitas Ma Chung, Indonesia) TS 1 – 08 Automatic Multi-Document Summarization for Indonesian Documents Using Hybrid Abstractive-Extractive Summarization Technique ...... 39 Glorian Yapinus (Swiss German University, Indonesia); Alva Erwin (Swiss German University, Indonesia); Maulahikmah Galinium (Swiss German University, Indonesia); Wahyu Muliady (Akon Teknologi, Indonesia) TS 1 – 09 Autonomous Monitoring Framework with Fallen Person Pose Estimation and Vital Sign Detection ...... 44 Igi Ardiyanto (Toyohashi University of Technology, Japan); Jun Miura (Toyohashi University of Technology, Japan) TS 1 – 10 Benchmarking of Feature Selection Techniques for Coronary Artery Disease Diagnosis ...... 50 Noor Akhmad Setiawan (Universitas Gadjah Mada, Indonesia); Dwi Wahyu Prabowo (Universitas Gadjah Mada, Indonesia); Hanung Adi Nugroho (Universitas Gadjah Mada, Indonesia) TS 1 – 11 Boosting Performance of Face Detection by Using an Efficient Skin Segmentation Algorithm ...... 55 Mohammad Reza Mahmoodi (Isfahan University of Technology, Iran); Sayed Masoud Sayedi (Isfahan University of Technoly, Iran) TS 1 – 12 C2C E-Commerce Trust Level Measurement and Analysis ...... 61 Sayid Ali Hadi (Swiss German University, Indonesia); James Purnama (, Indonesia); Moh. A. Soetomo (Swiss German University, Indonesia); Maulahikmah Galinium (Swiss German University, Indonesia) TS 1 – 13 Calories Analysis of Food Intake Using Image Recognition ...... 67 Natta Tammachat (King Mongkut's Institute of Technology Ladkrabang, Thailand); Natapon Pantuwong (King Mongkut's Institute of Technology Ladkrabang, Thailand) TS 1 – 14 Contrast Measurement for No-Reference Retinal Image Quality Assessment ...... 71 Hanung Adi Nugroho (Universitas Gadjah Mada, Indonesia); Titin Yulianti (Universitas Gadjah Mada, Indonesia); Noor Akhmad Setiawan (Universitas Gadjah Mada, Indonesia); Dhimas Arief D (Universitas Gadjah Mada, Indonesia) TS 1 – 15 Digital Image Hashing Using Local Histogram of Oriented Gradients ...... 75

-v-

Iwan Setyawan (Satya Wacana Christian University, Indonesia); Ivanna Timotius (Satya Wacana Christian University, Indonesia) TS 1 – 16 Emoticon-based Steganography for Securing Sensitive Data ...... 79 Tohari Ahmad (Institut Teknologi Sepuluh Nopember (ITS), Indonesia); Gregory Sukanto (Institut Teknologi Sepuluh Nopember (ITS), Indonesia); Hudan Studiawan (Institut Teknologi Sepuluh Nopember, Indonesia); Waskitho Wibisono (Institut Teknologi Sepuluh Nopember, Indonesia); Royyana Ijtihadie (Institut Teknologi Sepuluh Nopember (ITS), Indonesia) TS 1 – 17 Evaluation of Edge Orientation Histograms in Smile Detection ...... 85 Ivanna Timotius (Satya Wacana Christian University,Indonesia); Iwan Setyawan (Satya Wacana Christian University, Indonesia) TS 1 – 18 ICUMSA Identification of Granulated Sugar Using Discrete Wavelet Transform and Colour Moments ...... 90 Alfiah Rizky Diana Putri (Universitas Gadjah Mada, Indonesia); Adhi Susanto (Universitas Gadjah Mada, Indonesia); Litasari Litasari (Universitas Gadjah Mada, Indonesia) TS 1 – 19 Identification of Malignant Masses on Digital Mammogram Images ...... 96 Hanung Adi Nugroho (Universitas Gadjah Mada, Indonesia); Faisal N (Gadjah Mada University, Indonesia); Indah Soesanti (Universitas Gadjah Mada, Indonesia); Lina Choridah (Universitas Gadjah Mada, Indonesia) TS 1 – 20 Measuring Domain Decomposition Effect in Estuary Model Parallelization Using High Performance Computer...... 102 Santosa Sandy Putra (UNESCO IHE - Institute for Water Education, The Netherlands) TS 1 – 21 Mobile Tourism Services Model: A Contextual Tourism Experience Using Mobile Services ...... 108 Ridi Ferdiana (Universitas Gadjah Mada, Indonesia); Bimo Sunarfri Hantono (Universitas Gadjah Mada, Indonesia) TS 1 – 22 Real Time Key Element Extraction for Design of In-Flight Meal Services Based on Passenger's Personality Traits ...... 114 Taufik Djatna (Bogor Agricultral UnIversity, Indonesia); Hety Handayani Hidayat (IPB, Indonesia) TS 1 – 23 Real Time Static Hand Gesture Recognition System Prototype for Indonesian Sign Language ...... 120 Rudy Hartanto (Universitas Gadjah Mada, Indonesia); Adhi Susanto (Universitas Gadjah Mada, Indonesia); Paulus Santosa (Gadjah Mada University, Indonesia) TS 1 – 24 Release of Masking and FAME Performance Evaluation to Improve Speech Intelligibility on Cochlear Implant ...... 126 Sena Sukmananda Suprapto (Institut Teknologi Sepuluh Nopember, Indonesia); Dhany Arifianto (Institut Teknologi Sepuluh Nopember, Indonesia); Sekartedjo Sekartedjo (Institut Teknologi Sepuluh Nopember, Indonesia) TS 1 – 25 Statistical Analysis of Popular Open Source Software Projects and Their Communities ...... 132 Andi Wahju Rahardjo Emanuel (Universitas Kristen Maranatha, Indonesia) TS 1 – 26 Text-Background Decomposition for Thai Text Localization and Recognition in Natural Scenes ...... 138 Ungsumalee Suttapakti (King Mongkut's Institute of Technology Ladkrabang, Thailand); Kuntpong Woraratpanya (King Mongkut's Institute of Technology Ladkrabang, Thailand); Kitsuchart Pasupa (King Mongkut's Institute of Technology Ladkrabang, Thailand); Pimlak Boonchukusol (King Mongkut's Institute of Technology Ladkrabang, Thailand); Taravichet Titijaroonroj (King Mongkut's Institute of Technology Ladkrabang, Thailand); Rattaphon Hokking (King Mongkut's Institute of Technology Ladkrabang, Thailand); Yoshimitsu Kuroki (Kurume National College of Technology, Japan); Yasushi Kato (Tsuruoka National College of Technology, Japan) TS 1 – 27 The Study of Utilization of SIP in Mobile Monitoring Abnormal Events Wireless Sensor Network ...... 144 Andreo Yudertha (Gadjah Mada University, Indonesia); Widy Widyawan (Gadjah Mada University, Indonesia); Sujoko Sumaryono (Gadjah Mada University, Indonesia) TS 1 - 28 TIS Dishub DIY: An Implementation of Traveler Information System in Special Region of Yogyakarta ...... 150 Daniel Febrian Sengkey (Gadjah Mada University, Indonesia); Sayuri Egaravanda (Universitas Gadjah Mada, Indonesia); Lukito Nugroho (Universitas Gadjah Mada, Indonesia) TS 1 – 29 Website Quality Assessment for Portal Hospital Indonesia Using Gap Analysis ...... 156 Muhammad Adipridhana (Swiss German University, Indonesia); Maulahikmah Galinium (Swiss German University, Indonesia); Heru Ipung (Swiss German University, Indonesia)

 Session 2. Wireless Communications, Networking and Vehicular Technology

TS 2 – 01 3D Artificial Material Characterization Using Rectangular Waveguide ...... 164 Danang Wibowo (ITB, Indonesia); Achmad Munir (Institut Teknologi Bandung, Indonesia) TS 2 – 02 Design on FPGA of the IEEE 802.11p Standard Baseband OFDM Section Model ...... 168

-vi-

Budi Setiyanto (Universitas Gadjah Mada, Indonesia); Rani Aji (Universitas Gadjah Mada, Indonesia); Afatika Adianti (Universitas Gadjah Mada, Indonesia); Addin Suwastono (Universitas Gadjah Mada, Indonesia) TS 2 – 03 Development of Embedded Gateway for Wireless Sensor Network and Internet Protocol Interoperability ...... 174 Sigit Basuki Wibowo (Gadjah Mada University, Ireland); Guntur Putra (Universitas Gadjah Mada, Indonesia); Bimo Sunarfri Hantono (Universitas Gadjah Mada, Indonesia) TS 2 – 04 Dynamic CFO Reduction in Various Mobilities Based on Extended Kalman Filter for Broadband Wireless Access Technology ...... 178 Asri Diliyanzah (, Indonesia); Rina Pudjiastuti (Telkom University, Indonesia); Budi Syihabuddin (Telkom University, Indonesia) TS 2 – 05 Experimental Study on Improved Parametric Stereo for Bit Rate Scalable Audio Coding...... 184 Ikhwana Elfitri (, Indonesia); Rahmadi Kurnia (Andalas University, Indonesia); Defry Harneldi (Andalas University, Indonesia) TS 2 – 06 FDTD Method for Scattering Parameters Extraction of Rectangular Waveguide Loaded with Anisotropic Dielectric Material ...... 189 Achmad Munir (Institut Teknologi Bandung, Indonesia); Maulana Randa (Badan Penelitian Dan Pengembangan Kementerian Pertahanan RI, Indonesia) TS 2 – 07 FSS-based Planar Bandpass Filter Using Strip Slotted-Lines ...... 194 Eric Simbolon (Bandung Institute of Technology, Indonesia); Achmad Munir (Institut Teknologi Bandung, Indonesia) TS 2 – 08 High Gain RF Amplifier for Very Low Frequency Receiver Application ...... 199 Rahmat Putera (Institut Teknologi Bandung, Indonesia); Achmad Munir (Institut Teknologi Bandung, Indonesia) TS 2 – 09 Investigation on Objective Performance of Closed-loop Spatial Audio Coding ...... 203 Ikhwana Elfitri (Andalas University, Indonesia); Rahmadi Kurnia (Andalas University, Indonesia); Fitrilina Fitrilina (Andalas University, Indonesia) TS 2 – 10 Performance of Anti-Jamming Techniques with Bit Interleaving in OFDM-Based Tactical Communications .... 209 Pradini Puspitaningayu (Universitas Negeri Surabaya, Indonesia); Gamantyo Hendrantoro (Sepuluh Nopember Institut of Technology, Indonesia) TS 2 – 11 Performance of Repeat-Accumulate Codes (RAC) for Decode-and-Forward Wireless Relay Channel ...... 214 Daryus Chandra (Universitas Gadjah Mada, Indonesia); Adhi Susanto (Universitas Gadjah Mada, Indonesia); Sri Suning Kusumawardani (Universitas Gadjah Mada, Indonesia) TS 2 – 12 Reorganizing Fingerprint Information Using Intersection Technique for RFID-based Indoor Localization System ...... 220 I Wayan Mustika (Universitas Gadjah Mada, Indonesia); Sisongkham Phimmasean (NUOL, Laos) TS 2 – 13 RSSI Based Analysis of Bluetooth Implementation for Intra-Car Sensor Monitoring ...... 225 Eka Firmansyah (UGM, Indonesia); Lafiona Grezelda (Gadjah Mada University, Indonesia); Iswandi Iswandi (Gadjah Mada University, Indonesia)

 Session 3. Power Systems

TS 3 – 01 A Probabilistic Approach to Analyze and Model the Simultaneity of Power Produced by Wind Turbines in a Wind Farm ...... 232 Kaveh Malekian (Chemnitz University of Technology, Germany); Anne Göhlich (Chemnitz University of Technology, Germany); Liana Pop (Chemnitz University of Technology, Germany); Wolfgang Schufft (University of Technology Chemnitz, Germany) TS 3 – 02 An Improved Maximum Efficiency Control for Dual-Motor Drive Systems ...... 239 Luiz Rizki Ramelan (Universitas Gadjah Mada, Indonesia); Eka Firmansyah (UGM, Indonesia); Tian-Hua Liu (National Taiwan University of Science and Technology, Taiwan); Shao-Kai Tseng (National Taiwan University of Science and Technology, Taiwan); Jing-Wei Hsu (National Taiwan University of Science and Technology, Taiwan) TS 3 – 03 CCT Computation Method Based on Critical Trajectory Using Simultaneous Equations for Transient Stability Analysis ...... 245 Ardyono Priyadi (ITS, Indonesia); Ony Qudsi (Politeknik Elektronika Negeri Surabaya, Indonesia); Mauridhi Purnomo (Institut of Technology Sepuluh Nopember, Indonesia) TS 3 – 04 Comparison of Economic Models for Two Differently Configured Uninterrupted Power Supply Systems From User Electricity Bill Perspective ...... 251 Awais Yousaf (The University of Lahore, Pakistan); Onaiza Yousaf (The University of Lahore, Pakistan); Durdana Yousaf (Lahore Electric Supply Company, Pakistan)

-vii-

TS 3 – 05 Development of a Power Flow Software for Distribution System Analysis Based on Rectangular Voltage Using Python Software Package ...... 255 Lukmanul Hakim (Universitas Lampung, Indonesia); Muhamad Wahidi (Universitas Lampung, Indonesia); Trisno Handoko (Universitas Lampung, Indonesia); Herri Gusmedi (Universitas Lampung, Indonesia); Noer Soedjarwanto (Universitas Lampung, Indonesia); Federico Milano (University College Dublin, Ireland) TS 3 – 06 Efficiency Improvement of a Solar Power Plant Using Combine Cycle: An Experimental Study on a Miniaturized Solar Power Station ...... 260 Bishwajit Banik Pathik (American International University-Bangladesh, Bangladesh); Nipu Datta (American International University-Bangladesh, Bangladesh); Muhammad Najebul Ahmed (American International University-Bangladesh, Bangladesh); Roksana Liya (American International University-Bangladesh, Bangladesh); Nazia Zaman (American International University-Bangladesh, Bangladesh) TS 3 – 07 Flower Pollination Algorithm for Optimal Control in Multi-Machine System with GUPFC ...... 265 Mohammad Musofa Mulya, Pambudy (Gadjah Mada University, Indonesia) TS 3 – 08 Frequency Dependent Model of Underground Cables for Harmonic Calculations in Frequency Domain ...... 271 Kaveh Malekian (Chemnitz University of Technology, Germany); Uwe Schmidt (Dresden University of Technology, Germany); Abdullah Hoshmeh (Chemnitz University of Technology, Germany); Ali Shirvani (TU Chemnitz, Germany) TS 3 – 09 Fuzzy Logic Principles for Wind Speed Estimation in Wind Energy Conversion Systems ...... 278 Agus Naba (, Indonesia) TS 3 – 10 Investigation and Modeling of Transient Voltage Stability Problems in Wind Farms with DFIG and Crowbar System ...... 282 Kaveh Malekian (Chemnitz University of Technology, Germany); Uwe Schmidt (Dresden University of Technology, Germany); Ali Shirvani (TU Chemnitz, Germany); Wolfgang Schufft (University of Technology Chemnitz, Germany) TS 3 – 11 Magnetic Flux Distribution Due to the Effect of Stator-Rotor Configuration in the Axial Machine ...... 290 Danang Wijaya (UGM, Indonesia); Nobal Rahadyan (Universitas Gadjah Mada, Indonesia); Husni Ali (UGM, Indonesia) TS 3 – 12 Maximum Power Point Tracking Algorithm for Photovoltaic System Under Partial Shaded Condition by Means Updating β Firefly Technique ...... 296 Yanuar Safarudin (Institut Teknologi Sepuluh Nopember, Indonesia); Ardyono Priyadi (ITS, Indonesia); Mauridhi Purnomo (Institut of Technology Sepuluh Nopember, Indonesia); Margo Pujiantara (ITS, Indonesia) TS 3 – 13 Multi-Resolution Complex Image Method of Horizontal Multilayer Earth ...... 301 Qi Yang (Wuhan University, P.R. China) TS 3 – 14 On the Potential and Progress of Renewable Electricity Generation in ...... 307 Satya Kumara (Udayana University, Bali, Indonesia); Wayan G. Ariastina (Udayana University, Indonesia); I Sukerayasa (Udayana University, Indonesia); Ida Giriantari (Udayana University, Bali, Indonesia) TS 3 – 15 Optimal Configuration of PV-Wind turbine-Grid-Battery in Low Potency Energy Resources ...... 313 D Fittrin (Universitas Gadjah Mada, Indonesia); D Wijaya (Universitas Gadjah Mada, Indonesia); Sasongko Pramono Hadi (Gadjah Mada University, Indonesia) TS 3 – 16 Optimal Solution of Reliability Constrained Unit Commitment Using Hybrid Genetic Algorithm-Priority List Method ...... 319 Sarjiya Sarjiya (Gadjah Mada University, Indonesia); Arief Budi Mulyawan (Gadjah Mada University, Indonesia); Andi Sudiarso (Gadjah Mada University, Indonesia) TS 3 – 17 Partial Discharge Analysis Using PCA and ANN for the Estimation of Size and Position of Metallic Particle Adhering to Spacer in Gas-Insulated System ...... 325 Firmansyah Nur Budiman (Universitas Gadjah Mada, Indonesia); Yasin Khan (King Saud University, Saudi Arabia) TS 3 – 18 Quantum Neural Network for State of Charge Estimation ...... 331 Kevin Gausultan Hadith Mangunkusumo (Universitas Gadjah Mada, Indonesia); Danang Wijaya (UGM, Indonesia); Yung-Ruei Chang (Institute of Nuclear Energy Research, Atomic Energy Council, Taiwan); Yih-Der Lee (Institute of Nuclear Energy Research, Taiwan); Kuo Lung Lian (National Taiwan University of Science and Technology, Taiwan) TS 3 – 19 Reducing Induction Motor Starting Current Using Magnetic Energy Recovery Switch (MERS) ...... 336 Danang Wijaya (UGM, Indonesia); Sholihatta Aziz (UGM, Indonesia); Hartanto Prabowo (UGM, Indonesia) TS 3 – 20 The Dynamic Performance of Grid-Connected Fixed-Speed Wind Turbine Generator ...... 342 Husni Rois Ali (UGM, Indonesia) TS 3 – 21 TVAC PSO for Modal Optimal Control POD and PSS Coordination in UPFC ...... 347 Rian Fatah Mochamad (UGM, Indonesia); Sasongko Pramono Hadi (Gadjah Mada University, Indonesia); Mokhammad Setyonegoro (UGM, Indonesia)

-viii-

 Session 4. Electronics, Circuits, and Systems

TS 4 – 01 A Face Detector Based on Color and Texture ...... 354 Mohammad Reza Mahmoodi (Isfahan University of Technology, Iran); Sayed Masoud Sayedi (Isfahan University of Technoly, Iran) TS 4 – 02 Analysis of Single Excitation Signal for High Speed ECVT Data Acquisition System ...... 360 Arbai Yusuf (CTECH Labs Edwar Technology Co., Indonesia); Imamul Muttakin (CTECH Labs Edwar Technology Co., Indonesia); Wahyu Widada (CTECH Labs Edwar Technology Co., Indonesia); Warsito P. Taruno (CTECH Labs Edwar Technology Co., Indonesia) TS 4 – 03 Pulley's Clamping Force and Axial Position Measurements for Electro-mechanical Continuously Variable Transmission in Automotive Applications ...... 366 Bambang Supriyo (Universiti Teknologi Malaysia, Malaysia); Kamarul Tawi (Universiti Teknologi Malaysia, Malaysia); Mohd Che Kob (Universiti Teknologi Malaysia, Malaysia); Izhari Mazali (Universiti Teknologi Malaysia, Malaysia); Mohd Che Kob (Universiti Teknologi Malaysia, Malaysia) TS 4 – 04 Reconfigurable Hardware Implementation of Gigabit UDP/IP Stack Based on Spartan-6 FPGA ...... 370 Mohammad Reza Mahmoodi (Isfahan University of Technology, Iran); Sayed Masoud Sayedi (Isfahan University of Technoly, Iran); Batul Mahmoodi (Telecommunication Company of Isfahan, Iran) TS 4 – 05 The Performance of Three-Phase Four-Wire Grid-Connected Inverter with Enhanced Power Quality ...... 376 Susatyo Handoko (Universitas Gadjah Mada, Indonesia); Sasongko Pramono Hadi (Gadjah Mada University, Indonesia); Suharyanto Suharyanto (Gadjah Mada University, Indonesia); Eka Firmansyah (UGM, Indonesia) TS 4 – 06 Underwater Sound Propagation Characteristics At Mini Underwater Test Tank with Varied Salinity and Temperature ...... 381 Niken Yuwono (Institut Teknologi Sepuluh Nopember, Indonesia); Dhany Arifianto (Institut Teknologi Sepuluh Nopember, Indonesia); Endang Widjiati (Institut Teknologi Sepuluh Nopember, Indonesia); Wirawan Wirawan (Institut Teknologi Sepuluh Nopember, Indonesia)

 Session 5. Control Systems

TS 5 – 01 A Neural Network Structure with Parameter Expansion for Adaptive Modeling of Dynamic Systems ...... 388 Erwin Sitompul (, Indonesia) TS 5 – 02 A New Approach in Self-Generation of Fuzzy Logic Controller by Means of Genetic Algorithm ...... 394 Erwin Sitompul (President University, Indonesia); Iksan Bukhori (President University, Indonesia) TS 5 – 03 Double Target Potential Field ...... 400 Ferry Manalu (Universitas Katolik Indonesia Atma Jaya, Indonesia) TS 5 – 04 Robust Residual Generation for Sensor Fault Isolation in Systems with Structured Uncertainty ...... 405 Samiadji Herdjunanto (Gadjah Mada University, Indonesia); Adhi Susanto (Universitas Gadjah Mada, Indonesia); Oyas Wahyunggoro (UGM, Indonesia) TS 5 – 05 Design of Decoupled Repetitive Control for MIMO Systems ...... 411 Edi Kurniawan (Indonesian Institute of Sciences, Indonesia); Riyo Wardoyo (Indonesian Institute of Sciences, Indonesia); Oka Mahendra (Indonesian Institute of Sciences, Indonesia)

-ix-

ICITEE 2014 Yogyakarta, 7-8 October 2014 ISBN: 978-1-4799-5302-8 Real Time Static Hand Gesture Recognition System Prototype for Indonesian Sign Language

Rudy Hartanto, Adhi Susanto, and Paulus Insap Santosa Department of Electrical Engineering and Information Technology, Gadjah Mada University (UGM) Jalan Grafika No. 2, Yogyakarta, 55281 Indonesia Email: [email protected], [email protected], [email protected]

Abstract— Sign language uses gestures instead of speech represented in the form of images, while the dynamic hand sound to communicate. However, it is rare that the normal people gesture defined as the sign of moving hand gesture represented try to learn the sign language for interacting with deaf people. by the image sequence or video. Therefore, the need for a translation from sign language to written or oral language becomes important. Many research works related to sign languages have been In this paper, we propose a prototype system that can done such as the American Sign Language (ASL) [1] , the recognize the hand gesture sign language in real time. We use British Sign Language [2], the Japanese Sign Language [3], HSV (Hue Saturation Value) color space combined with skin and so on. However, very few works has been done in detection to remove the complex background and create Indonesian Sign Language recognition until date. In this paper segmented images. Then a contour detection is applied to localize we proposed a prototype system that can recognize the and save hand area. Further, we use SURF algorithm to detect Indonesian alphabets in real-time. A single camera is used to and extract key point features and recognize each hand gesture capture hand gesture pose, which then will be recognized by sign alphabet by comparing with these user image database. the system and translate to an Indonesian alphabet. Based on the experiments, the system is capable to recognize hand gesture sign and translate to Alphabets, with recognize rate Rios-Soria et.al [4] had doing a research to recognise hands 63 % in average. gestures to interact with system. They proposed hands gesture detection algorithm, such as filters, border detection, convex Keywords—sign language; real time; HSV color space; hull detection and using standard webcam to capture hand complex background; contour detection; key point feature command image. They can use their hands to interact with this such as pick them up, transform their shape, or move them. The I. INTRODUCTION system is capabel to recognise six different hand gestured Sign language is one of the main communication media for captured using a webcam in real time. the deaf people worldwide. Sign language users around the Singha [5] had proposed a hand gesture recognition system world are quite a lot. Each country even each region have their based on Karhunen-Loeve transform. Their system consist of own sign language. However, it is rare that the normal people five steps: skin filtering, palm cropping, edge detection, feature try to learn sign language for interacting with the deaf people. extraction, and classification. The system were tested using 10 This becomes a cause of isolation of the deaf people. Hence, different hand gestures and capable to recognize up to 96%. the need for a translation from sign language into written or Safaya [6] using DVS (dynamic vision sensor) camera to oral language becomes important. capture image of hand gesture an then recognized. DVS is Sign language uses gestures instead of speech sound to different from conventional camera, its only respond to pixels communicate. Gestures are forms of body language or non- with temporal luminance difference. That is can reduce the verbal communication that commonly use combination of computational cost of comparing consecutive frames to track shape/patterns, orientation and movement of the hands, face moving object. Chaudhary [7] uses Neural Network algorithm and expression patterns of the lips. Sign language recognition to recognize static sign gesture. The static gesture image will is considered the most structured of all categories of gesture. be convert into Lab color first, and then segmented using Sign language is an area of research involving the fields of threshold tehcnique. After cropping the hand area and pattern recognition, computer vision, and natural language converted to binary image for feature extraction, finally its will processing. The implementation of sign language using hand be used to train a feed-forward back propagation network. gestures still facing many problems, due to the complexity of For sign language recognition many research had been gesture recognition, and naturally, there is a high variation in done, such as real-time static hand gesture recognition for ASL the performance of user gestures, such as their physical conducted by [8]. In this system, the hand will be capture using features and environmental conditions. Further hand gesture webcam at fixed position. The system consist of four stages can be classified into two categories, namely static and that are pre processing, region extraction, feature extraction, dynamic hand gestures. Static hand gesture is defined as the and feature matching. [9] had introduced an efficient and fast configuration of the sign of hand gestures and poses that are algorithm for identification of the number of fingers opened in

-120- ISBN: 978-1-4799-5302-8 Yogyakarta, 7-8 October 2014 ICITEE 2014 a gesture representing an alphabet of the Binary Sign white pixel represents the object (hand gesture) and black pixel Language. They build an intelligent system using image represents the background. processing, machine learning and artificial intelligence C. Contour Detection concepts to take visual inputs of sign language’s hand gestures and generate easily recognizable form of outputs. Contour is a sequence of pixels with same value obtained from the differences between pixels with its neighbors in II. THE PROPOSED METHOD binary image resulted from skin detection between white A. Research Objectives pixels (the object that has the colors resemble hands) and black pixels (objects other than hands or background). This The objective of this research is making a system prototype method however, has the possibility of a high error when an that can recognize Indonesia Sign Language alphabets in real time. We propose a new approach and algorithms for hand object in an image has a color that is almost the same. These gesture recognition that consist of two-step. The first step is yield more than one contours, which is hand contours and part hand gesture pose recognition process and the next step is of the background, which has same color as user’s hand. identification of transition from one pose to another pose. Fig. Practically, the biggest contour is the user’s hand. Therefore, 1 shows the proposed system, starting with the identification of need to be sure that the algorithm will compare the size of all hand gesture image captured by ordinary webcam and contours and choose only one contour in the frame, which is separates it from their background. The segmentation and the biggest contour (i.e. The user's hand). background elimination is accomplished using HSV color Fig. 2 shows the flow chart to detect the biggest contour space combine with skin color detection algorithm, while the (that is hand area) and make boundary box around the contour recognition process is accomplished using SURF (Speeded Up as a region of interest (ROI). The ROI rectangle size will Robust Features) algorithm. The features of hand gesture depend on the size of hand image (the biggest contour image will then compare with feature of hand gesture image detected). This ROI will then be save to the buffer and used as from database. an input to the next stage.

Start Skin Detection & Image Background Acquisition Removal Make Contour on Binary Image from Segmentation Process

Contour Feature Detection & Extraction Finding ROI No The biggest contour area? Yes No, Next Image Match Alphabet ? Yes

The biggest Area = Hand Contour Area Feature Hand Gesture Extraction Image Database

Find the center of Contour & Draw Rectangle around the contour Fig. 1. The Proposed System. B. Skin Detection and Background Removal Display Hand Contour, Center The hand gesture pose will be captured using ordinary Point and The Boundary Box webcam. We use skin color detection algorithm to detect, segmented hand area, and then eliminate the image background as well. To improve the reliability from the change of Save Image Inside The Boundary environment illumination, we convert RGB color space to HSV Box to The Buffer color space to isolate brightness (value or luminance) component V, with color components (hue or tint). Fig. 2. Contour detection and find ROI. The skin color detection can be obtained by determining the range of H, and S values, according to the value of the user's D. Feature Extraction skin color to be used as a threshold to determine whether the pixel value is white (hand image) or black (background). Eq. Feature extraction process is done by using SURF (1) shows the algorithm of skin color detection and threshold algorithm. The SURF detector is based on the determinants of process to convert to binary-segmented image. the Hessian matrix. Suppose that a continuous function of two variables such that the value of the function at (x, y) is given (1) by f (x, y). Then the Hessian matrix H is the matrix partial derivative of a function f [10]. Where H1 and H2 as well as S1 and S2 are the range of skin color value of user hand. The output is a binary image where

-121- ICITEE 2014 Yogyakarta, 7-8 October 2014 ISBN: 978-1-4799-5302-8

2 det(Happrox) = DxxDyy – (0,9 Dxy) (5) (2) To be able to detect the point of interest using the determinant of Hessian then it is need to introduce the notion The determinant of a matrix in (2) is known as the of a scale-space first, which is a continuous function that can discriminant, which can be obtained by: be used to find the extrema across all possible scales. In (3) computer vision, it is typically implemented as the image pyramid where the input image is convolves iteratively with The discriminant value is used for classifying the minima and Gaussian kernel and repeatedly sub-sampled to reduce the maxima value of the function by second order derivative test. size. Due to the processing time of the kernels used in SURF Since the determinant is the product of the eigenvalue of the does not change in size, the scale of space can be created by Hessian, then we can use the sign of the result to classify the applying kernel of increasing size to the original image. This points. The eigenvalue will have different sign and the point is allows multiple layers of the scale space pyramid to be not a local extremum, if the determinant is negative; and both processed simultaneously and eliminates the need to sub- eigenvalues are either positive or negative if the determinant is sample the image so it will increase the performance. Fig. 5 positive and this point is classified as an extremum. illustrates the difference between the traditional scale-space To apply the theory to work in images, then we first need structures with SURF. to replace the value of the function f (x, y) by the pixel image intensity I (x, y). Next, calculate the second-order partial derivatives of the image that can be obtained by performing a convolution with an appropriate kernel. We choose the second-order scale normalized Gaussian filter because it allows the analysis over scale and space. Since the Gaussian is an isotropic function, convolution with kernel allows unchanging against rotation. Further the Hessian matrix H can be calculated, as a function for both space x = (x, y) and scale ơ Fig. 4 Filter Pyramid. The traditional approach to constructing a scale- . space (left). The SURF approach (right) leaves the original image unchanged and varies only the filter size. [10] (4) The scale-space is divided into a number of octaves, where an octave refers to a series of response map that cover doubling of scale. The lowest level of scale-space is obtained from the Where Lxx (x, y) refers to the convolution of the second order output of 9x9 filters as shown on Fig. 4. These filters Gaussian derivative with the image at point x = (x, y) correspond to a real value Gaussian with ơ = 1.2. The next layer is obtained by up scaling the filters while maintaining the and similarly for Lyy and Lxy. These derivatives are known as ratio of filter layout remains the same. With increasing in the the Laplacian of the Gaussian. size of the filter, so is the value of the associated Gaussian Fig.3 shows that the weight applied to each of the filters scale. Since the layout ratio is, remain constant, so this scale is made simple. For the Dxy filter, the black areas are can be calculated: weighting with 1, while the white areas with value of -1 and the rest were not weighted at all. The Dxx and Dyy filters are weighted similarly but with white area have a value of-1 and the black area with a value of 2. This simple weighting allows rapid calculation of the area but in using these weights is necessary to include the difference in response between the original and approximation kernel. E. Interest Points Descriptor SURF descriptor describes how the intensity of pixels is distributed within a scale dependent neighborhood of each point of interest detected by Fast-Hessian. This approach use Haar wavelets filter to increase robustness and decrease

Fig.3 Laplacian of Gaussian Approximation. [11] computing time. [10] Extraction of the descriptor can be divided into two On left to right of Fig. 3 is discretized and cropped different tasks. First, each interest point is assigned a Gaussian second order partial derivatives in y and xy direction, reproducible orientation before a window, which is scale which is referred to as the Lxx, and Lyy Lxy. This approximation dependent is constructed in which a 64-dimensional vector is is accomplished using Weighted Box Filter. Determinant of extracted. The important thing is that all calculations for the the Hessian using Gaussian approximation can be obtained by descriptor are based on the measuring of relative to the scale Eq. (5) that is detected in order to achieve scale invariant results.

-122- ISBN: 978-1-4799-5302-8 Yogyakarta, 7-8 October 2014 ICITEE 2014

To achieve image rotation invariant each detected interest We use SURF detectors algorithm to extract the image point is assigned a reproducible orientation. To determine the features. This process will yield a set of key points as a hand orientation the responses of Haar wavelets of 4ơ are calculated image features. Feature extraction will be applied to the input for a set of pixels within radius 6ơ of detected point, where ơ image of the hand as well as the database image. Further, the refers to a scale where the point is detected. The specific set of key point features of a hand image will be compared with the pixels is determined by sampling those from within the circle image from database. The comparison based on the distance by using the size step of ơ. between each key point location from hand image captured from camera and the counterpart image in database. The The responses are weighted with a Gaussian centered at the minimum distance needs to be determined first, than if the interest point. After weighting of responses is represented as a distance less or equal than two times minimum distance, this point in vector space, with x- responses along the abscissa and key points can be considered match. If there a number of key y-responses along the ordinate. The dominant orientation is points and position are match, then the hand gesture image is chosen by rotating a segment of a circle covering an angle of identical to the hand gestures image in database. Thus, it can be π /3 around the origin. At each position, the x and y- response determined that the hand gestures images captured by the within the segment of aggregated and used to form the new camera represents a particular alphabet. The steps of image vector. Vector orientation will take's longest point of interest, features extraction of hand gesture image and the pattern as shown in Fig. 5. matching process with gestures image on the database based on the similarity of key point is shown in Fig. 7.

Hand Gesture Images Image ROI Database

Obtain Feature Extraction to Get Key Obtain Feature Extraction to Get Key Points Points

Compute Key Points Matching with Fig. 5 Orientation assignment: With the window slide around the SURF Algorithm origin, the component of the responses will be aggregated to yield a vector (shown in blue). The largest Vector determines the dominant Find Min and Max Get Next Image from Distance of Matching orientation. [10] Database Key Points To extract the SURF descriptor, the first step is to construct No a square window around the interest point. This window Is the Last image in No Key Points Distance <= 2x contains the pixels, which will form entries in the descriptor Database ? Min distance ? vector and is of size 20σ, again where σ refers to the detected scale. The descriptor window is divided into 4×4 regular Yes Yes subregions. Within each of these subregions Haar wavelets of Input Image is Mathc wit size 2σ are calculated for 25 regularly distributed sample Database Image points. If we refer to the x and y wavelet responses by dx and dy respectively then for these 25 sample points (i.e. each Stop Write Alphabet subregion) we collect, (5) Therefore, each subregion contributes four values to the Fig. 7. Key-points extraction and image matching. descriptor vector leading to an overall vector of length 4×4×4 = III. EXPERIMENT RESULT 64, as shown in Fig. 6. The resulting SURF descriptor is invariant to rotation, scale, brightness and, after reduction to First, we will present the evaluation result of each step, and unit length, contrast. then discus the result of the whole system. The evaluation is accomplished using Intel I7 processor with 1.7 GHz machine and the image sequence is captured using webcam with 20 frames per second. A. Skin Detection and Background Removal The aim of this process is to get hand image and remove the remaining (the image background). To increase the robustness again the variation of the environment illumination, the RGB image captured from webcam will first converted to HSV color space using Eq. (1) as shown in Fig. 7(a) and 7(b). Then use Figure 6: Descriptor Components. The green square bounds one of the Eq. (2) to get the hand image and eliminates the background. 16 subregions and blue circles represent the sample points at which we Fig. 8 shows the result of skin detection and segmented hand compute the wavelet responses. As illustrated the x and y responses image. are calculated relative to the dominant orientation. [10]

-123- ICITEE 2014 Yogyakarta, 7-8 October 2014 ISBN: 978-1-4799-5302-8

(a) RGB Image (b) HSV Image (a) Letter A (b) Letter W Fig. 10. The hand gesture matching result for letter A and W.

From the results of the observation, the hand gestures pattern matching process is going pretty well but still not optimal yet. Fig. 10 shows the matching results between key (c) Background Removal (d) Binary Image point features of hand gesture image from the database with the Fig 8. The image conversion result from RGB to HSV, and image input from webcam. The lines that connect the left and background removal proses and the corresponding binary image. right image show the existence of matching key point feature. A line with certain colors will connect every feature of a key point with the same distance. If the number of lines has B. Contour Detection and ROI Finding exceeded the minimum requirement (in this case is 15), then it can be considered that both hand gesture images are identical. The binary image resulted from background removing and segmentation stage will then be used to search the contour. In this experiments we exclude ‘J’ and 'Z' alphabet, Practically, the contour searching will produce more than one because they contains dynamic elements, i.e. its need hand contour due to the imperfect segmentation, for example, there movement to performs that signs. Two users show the sign are other similar objects by hand but smaller, as shown in Fig. language pose for alphabets that will captured by webcam as a 9(a). In order to ensure that only one contour, that is hand system input. Each alphabet’s sign will be repeated 9 and 3 contour is processed, it is need to specify the largest contour times by two users respectively. Table 1 shows the results for area as shown in Fig. 9(b). Thus the contour with the largest the 24 characters who the image signs are already stored in the area only will be saved, and ignore the other contours. Finally, database. Practically the input sample images have differences based on hand contour can be draw a boundary box to with the image Stores in database in terms of image size and determine the region of interest (ROI) as depicted in Fig 9(c). orientation because it is very difficult to make the same. The experiment shows that system can recognize all of alphabet, except J and Z with the average accuracy around 63 %.

TABLE I. THE IMAGE MATCHING RESULTS OF INDONESIAN SIGN FOR 24 ALPHABETICAL. Alphabets Recognized False Unrecognized Recognized A 7 4 1 (a) (b) (c) B 8 2 2 Fig 9. The step result from contour detection, drawing a boundary box C 9 1 2 and determine the region of interest image. (a) Contour detection D 9 2 1 result, (b) Hand contour, (c) Boundary box. E 7 3 2 F 9 2 1 G 7 3 2 C. Feature Extraction and Gesture Recognition H 7 4 1 Feature extraction of hand image is implemented using I 8 2 2 K 9 1 1 SURF algorithm to extract key points. Both hand image L 9 3 0 resulted from previous stage as well as from database will then M 6 4 2 processed to extract their key points features. Starting with the N 7 3 2 detection of key point features of each image then followed O 8 2 2 P 7 3 2 with key point extraction and matching based on the minimum Q 6 2 4 distance of every key points location on the hand image and R 6 4 2 their counterpart in database. If the number of matching key S 7 3 2 point meets the minimum allowable limits, then considered T 5 4 3 U 9 2 1 that the both images are identical and the alphabet can then be V 8 3 1 determined. W 9 1 2 X 7 3 2 Y 8 2 2 Average (%) 63 22 15

-124- ISBN: 978-1-4799-5302-8 Yogyakarta, 7-8 October 2014 ICITEE 2014

D. Discussion low, which is about 63% with computation time around 1.6 From the results of the observation, process shows that the second. system is capable to recognize hand gestures sign but still not B. Future Works optimal. The low recognizing rate is also influenced by the size of image that relatively small at 320x240 pixels. The Based on the limitations of the system, for future works it is consideration of image size selection due to the duration of recommended to optimized and simplified the algorithm computing process that is more than 1.6 seconds. Because the especially key point’s features extraction and matching to algorithm must compare the key points feature of the input increase the capability in identification and matching hand image with every key point’s features images in the database, gesture images with different scale as well as to reduce in this case we have 24 images in the database that represent computation time. The performance of the algorithm need to the entire alphabet except J and Z. Further, some of the increase to handle the variation of image gesture input alphabet’s sign have the form or pose that nearly equal (e.g. A, orientation. It is need to increase image size as well to improve E, M, N, S, and T) so that the key points feature are also identification performance while maintain the computation relatively similar that lead to false recognized result. time. Furthermore, the influence of the hand gestures orientation and position toward the cameras also effect on the performance REFERENCES of key point features matching process. As we know, it is very difficult in practice to maintain the input image of hand gesture pose and orientation to be the same as the one stored in [1] J. L. Hernandez-Rebollar, N. Kyriakopoulos and R. W. Linderman, "A database, even though done by the same person. From the New Instrumented Approach For Translating American Sign Language experiments, the differences in orientation with less than 15 Into Sound and Text," Proceedings of the Sixth IEEE International Conference on Automatic Face and Gesture Recognition, 2004. degrees of the input hand gestures relatively with hand gestures [2] D. Stein, P. Dreuw, H. Ney, S. Morrissey and A. Way, "Hand in Hand on the database can still be recognized by the system, as shown Automatic Sign Language to English Translation," Proceedings of The in Fig. 11. The number of lines connecting two hand images 11th Conference on Theoretical and Methodological Issues in Machine from the database and image inputs as shown in Fig. 11 (a) and Translation, 2007. (b) are still fulfilled the minimum requirement, so it is still [3] N. Tanibata, N. Shimada and Y. Shirai, "Extraction of Hand Features for considered identical. While in Fig. 11(c), is less than the Recognition of Sign Language Words," in International Conference on minimum requirement and is considered not identical. Vision Interface, 2002. [4] D. J. Rios-Soria, S. E. Schaeffer and S. E. Garza-Villarreal, "Hand- gesture recognition using computer-vision techniques," in WSCG 2013: communication papers proceedings: 21st International Conference in Central Europe on Computer Graphics, Visualization and Computer Vision in co-operation with EUROGRAPHICS Association, 2013. [5] J. Singha and K. Das, "Hand Gesture Recognition Based on Karhunen- Loeve Transform," in Mobile & Embedded Technology International Conference (MECON), 2013. (a) (b) (c) Fig. 11 The orientation difference. [6] K. K. Safaya and P. J.W.Bakal, "Real Time Based Bare Hand Gesture Recognition," IPASJ International Journal of Information Technology (IIJIT), vol. 1, no. 2, 2013. One of the restriction is the computing time to recognize the pattern of a hand gesture sign input which takes more than [7] P. Chaudhary and H. S. Ryait, "Neural Network Based Static Sign Gesture," International Journal of Innovative Research in Computer and 1.6 seconds. Therefore, this system can only use to recognize Communication Engineering, vol. 2, no. 2, pp. 3066-3072, 2014. Indonesia alphabet sign language, with turnover between [8] J. R. Pansare, S. H. Gawande and M. Ingle, "Real-Time Static Hand alphabetic gestures with one another for more than 2 seconds. Gesture Recognition for American Sign Language (ASL) in Complex However, to recognize Indonesia sign language other than Background," Journal of Signal and Information Processing, vol. 3, pp. alphabet (e.g. word) in real time still needs improvement on the 364-367, 2012. algorithm of the key points extraction and comparison of hand [9] S. Pramada, D. Saylee, N. Pranita, N. Samiksha and M. S. Vaidya, gestures, due to reduce the computation time to less than 1/15 "Intelligent Sign Language Recognition Using Image Processing," IOSR seconds, or speed of image capture at least 15 FPS (frames per Journal of Engineering (IOSRJEN), vol. 3, no. 2, pp. 45-51, 2013. second). [10] C. Evans, "Notes on the OpenSURF Library," University of Bristol, Department of Computer Vision, 2009. IV. CONCLUSSION [11] H. Bay, T. Tuytelaars and L. V. Gool, "SURF: Speeded Up Robust 9th European Conference on Computer Vision A. Conclussion Features," in , 2006. [12] M. A. Rahman, Ahsan-Ui-Ambia and M. Aktaruzzaman, "Recognition Based on the evaluation and discussion above can be Static Hand Gesture of Alphbetin ASL," IJCIT, vol. 2, no. 01, 2011. concluded that the prototype of sign language recognition system can be implemented and work in real time. The system is capable to eliminate the background pretty well so it can recognize hand gesture sign without using glove or other additional device. The system performance in recognizing an alphabet sign language in the complex background is relatively

-125-