Abstract: This paper proposes a new algorithm for stereo matching which utilizes the wavelet theory developed recently. The original stereo images are decomposed into a family of bandpass versions which have good spatialfrequency Iocalization using Hardy elementary wavelethnction.The algorithm uses the phases of multichannel bandpass versions amatching primitives,so the problem of computing disparity can be for-mulated in terms of phase matching. The system is a coarse-fine multichannel matching strategy, and the matching constraints are combined with the phase matching. Experimental results show that the proposed algorithm is rather simple, noiseproof and can produce a dense disparity field efficiently.