Abstract: The image matching algorithm based on gray-scale is sensitive to illumination changes, large in computation, and its efficiency is very low. The matching algorithm based on feature is complex and relies heavily on the quality of feature extraction. This paper presents a new image matching algorithm. It is a fast matching algorithm based on Sobel edge features, the wavelet transform, recursive plan and multi-template. First, using Sobel operator to extract edge features; second, using wavelet transform to decompose the image in multi-scale; last, using the recursive plan and multi-template method proposed in this paper to match with the low-frequency part which includes the main information of the image. The algorithm is applied in the midcourse guidance system based on the ADSP-TS201. The experiment shows that this algorithm not only ensures the matching accuracy but also improves the system’s realtime.