一种基于相位基元的立体匹配方法

基于人类的视觉机理和小波变换的特性，提出了一种基于小波变换相位基元的多通道立体匹配算法．该算法利用Ｈａｒｄｙ小波基函数将原始像对表示成一系列具有良好空－频域定位性的带通分量，以多通道带通分量的相位信息作为匹配基元．从而将视差估计转化为相位估计问题．研究结果表明，该算法结构简单，稳定可靠，能快速有效地产生密集的视差场．

A Stereo Matching Algorithm Based on Phase Primitives

This paper is concerned with a new algorithm for stereo matching that utilizes the wavelet theory. The original stereo images are decomposed into a family of bandpass versions with good spatial frequency localization by using the Hardy elementary wavelet function. As the phases of multichannel bandpass versions are used, as matching primitives, the problem of computing disparity can be formulated in terms of phase matching. The system is a coarse to fine multichannel matching strategy, and the matching constraints are combined with the phase matching. Experimental results show that the algorithm proposed is fairly simple, noiseproof and can efficiently produce a dense disparity field.