Penelitian ini menyajikan penggunaan metode Phase Only Correlation (POC) pada motion estimation dengan teknik full search menggunakan Graphical Processing Unit (GPU). Dengan fungsi POC, seseorang dapat melakukan estimasi translasi motion antara dua blok citra referensi dan citra yang diproses. Full Search berbasis POC adalah algoritma yang membutuhkan waktu proses lama. Hal ini menyebabkan sistem yang dicoba pada penelitian ini memproses fungsi POC pada Graphical Processing Unit (GPU) yang memiliki kelebihan dalam menyelesaikan perhitungan bilangan floating point dibandingkan CPU. Evaluasi dilakukan dengan menghitung kecepatan waktu proses menggunakan GPU pada video resolusi tinggi dengan resolusi hingga 1280x720 pixel. Hasil pengujian menunjukkan bahwa metode yang diselesaikan menggunakan GPU memiliki percepatan hingga hampir dua kali lipat pada ukuran blok POC 256 x 256 daripada menggunakan CPU.
This research presents a method using Phase Only Correlation (POC) on the motion estimation with full search technique using the Graphical Processing Unit (GPU). With POC function, someone can estimate the translational motion between two blocks of the reference image and the processed image. POC based Full Search is an algorithm that takes long time process. This leads the system that is used in this research to process the POC function on Graphical Processing Unit (GPU) which has advantages in solving the floating point calculations than the CPU. Evaluation is conducted by calculating the speed of processing time using a GPU on a high-resolution video with resolutions up to 1280x720 pixels. The test results show that the method that is solved using GPU has an acceleration up to nearly twice the size of the POC block 256 x 256 instead of using the CPU.
[1]
T. Higuchi,et al.
A Sub-Pixel Correspondence Search Technique for Computer Vision Applications
,
2004
.
[2]
Coarse-to-fine Least Squares Motion Estimator
,
1993
.
[3]
Simon Stegmaier,et al.
Exploiting frame-to-frame coherence for accelerating high-quality volume raycasting on graphics hardware
,
2005,
VIS 05. IEEE Visualization, 2005..
[4]
Yakup Genc,et al.
GPU-based Video Feature Tracking And Matching
,
2006
.
[5]
John Watkinson.
The Engineer’s Guide to Motion Compensation
,
1997
.
[6]
T. Higuchi,et al.
High-Accuracy Subpixel Image Registration Based on Phase-Only Correlation
,
2003,
IEICE Trans. Fundam. Electron. Commun. Comput. Sci..
[7]
Takafumi Aoki,et al.
Robust Motion Estimation for Video Sequences Based on Phase-Only Correlation
,
2004
.
[8]
C. D. Kuglin,et al.
The phase correlation image alignment method
,
1975
.
[9]
Eero P. Simoncelli.
Coarse-to-fine estimation of visual motion
,
1993
.