In this paper, an algorithm for image matching which utilizes the wavelet transform was proposed. A binary wavelet radicle was selected and the multiresolution zero-crossing of binary wavelet transform was used as matching primitive. This algorithm used more structure features of the image. The matching strategy from coarse-to-fine and strict constraints were used to reduce the searching area. The quality control after matching was carried out also. Experiment shows that the algorithm is reliable and stable in operation, and has high accuracy.