This paper presents an efficient multiscale binocular correspondence algorithm based on dyadic wavelet transform．The stereo images are first decomposed on the dyadic wavelet bases．By analyzing the transform coefficients，the authors get an efficien(method to remove noises and extract edges．Some basic constraints for matching，e．g．the one which requires that the edge directions be nearly the same，are naturally expressed by the coefficients．More importantly，they can deduce four useful constraints for selecting the potential matches which are called the smooth component gradient constraints and the smooth component Laplacian constraints．They have strong disambiguation power and greatly remove those potential matches that would cause mismatches，and thus greatly im prove the accuracy and speed for matching．A multiscale coarse to fine control strategy is included to increase robustness．