Efficient disparity estimation algorithm based on spatial correlation
Jin-Woo Bae · Optical Engineering · 2003
As one of the disparity estimation algorithms for stereoscopic images, we propose the block-based adaptive field-dependent search (AFDS) algorithm to reduce encoding time and increase image quality at the same time. The AFDS algorithm is a fast disparity estimation algorithm that uses the homogenous direction property of the disparity vector and the high spatial correlation between casual neighboring blocks on the epipolar line. Performance of the disparity estimator is improved using a dynamic search range and decreasing the disparity compensation error that is caused by objects taken by a camera at a short distance. As a result of the simulation, we can reduce the computational load up to 78% but can improve the peak SNR (PSNR) up to only about 0.2 to 2.8 dB, compared with the block-matching algorithm.