IDENTIFIKASI PERPINDAHAN POSISI UAV BERBASIS PEMROSESAN CITRA DIGITAL

  • Ariel Valentino Fakultas Teknik Elektro / Universitas Surabaya
Abstract Views: 94 times
PDF - FULL TEXT Downloads: 83 times
Keywords: Visual odometry, cross-correlation, Harris-Stephens corner detection, intensity macthcing

Abstract

Abstrak – Penentuan posisi pada pesawat tanpa awak atau UAV sangat penting untuk menjalankan sistem kendali pesawat tersebut. Tetapi sensor GPS yang digunakan pada UAV tidak selalu dapat memberikan informasi posisi sensor tersebut. Alasan tersebut mendasari penelitian ini untuk dikembangkan topik terkait visual odometry pada daerah yang sensor GPS tidak dapat lakukan. Fokus dari penelitian ini adalah mengembangkan algoritma yang memiliki kompleksitas rendah. Algoritma tersebut akan dikaji dengan berdasarkan tingkat akurasi, kompleksitas notasi big-O, dan waktu eksekusi. Algoritma dikatakan berhasil dengan harapan mampu mengidentifikasi perpindahan posisi dengan akurasi lebih dari 80%. Adapun algoritma yang dikembangkan pada penelitian ini ini yaitu: (1)cross-correlation, (2)Harris-Stephen corner detection, dan (3)penyesuaian intensitas. Keandalan algoritma akan diuji dengan melakukan beberapa variasi pengujian. Berdasarkan pengujian yang telah dilakukan, algoritma terbaik untuk mengidentifikasi pergerakan dihasilkan oleh algoritma Harris-Stephen corner detection. Algoritma ini dapat mengidentifikasi pergerakan dengan tingkat akurasi di atas 80% untuk berbagai variasi gerakan, ketinggian, dan variasi jumlah objek pada gambar.

Kata kunci: Visual odometry, cross-correlation, Harris-Stephens corner detection, penyesuaian intensitas

Abstract – The position estimation of unmanned aerial vehicle (UAV) is very important in order to control it autonomously. But, the GPS sensor used on UAVs does not always provide information about its position. Because of that, this final project aims to develop visual odometry in unreachable GPS signal areas. The focus of this research is to develop algorithm that have a low complexity. The algorithm will be assessed based on accuracy level, big-O notation, and execution time. The algorithm will identify as successful if it is able to identify an accuracy in position displacement of more than 80%. The algorithm that is developed in the final project i.e. (1)cross-correlation, (2)Harris-Stephen corner detection, and (3)intensity matching. The reliability of the algorithm will be tested by conducting several test variations. Based on the algorithm implemented in the testing, the best algorithm to identify movement change is Harris-Stephen corner detection algorithm. That algorithm can identify movement with an accuracy of above 80% for a wide variety of movements, heights, and amounts of objects in the image.

Keywords: Visual odometry, cross-correlation, Harris-Stephens corner detection, intensity macthcing

Downloads

Download data is not yet available.

References

[1] Tomas K. and et. al., “A Simple Visual Navigation system for an UAV”, in: 9 th IEEE International Multi-Conference on System, Signal, and Devices (SSD), 2012.

[2] N.D. Pah dan Henry H., “The Development of Image-based Algorithm to Identify Altitude Change of a Quadcopter”, in: the 7 th International Conference on Information Technology and Electrical Engineering (ICITEE), 2015.

[3] N. D. Pah, “Image-based Distance Change Identification by Segment Correlation”, in: Proceedings of Second International Conference on Electrical System, Technology and Information (ICESTI), 2015

[4] Chaira T. and et. al., “Low Computational-complexity Algorithms for Vision-aided Inertial Navigation of Micro Aerial Vehicles”, Readings in Sciencedirect: Robotics and Autonomous Systems, vol 69, page 80-97, 2015.

[5] Yung-Yu C., lecture topic: Digital Visual Effects, Readings in Spring 2006
Published
2019-09-01