Because of the presence of the difference in grey level and the distortion between object image and reference image, traditional matching method will be degraded. We need more useful characters. In this paper, a new image matching method based on singular value decomposition (SVD) is proposed. Firstly, object image is decomposed, its SVD value and SVD eigenvector are obtained, and then they are applied as characters of the object image in matching process. Because of the excellent characters of SVD, the experimental results show that this image matching method could bear image distortion to some extent and is very promising.