Aiming at the problem of uneven distribution of extracted feature points in traditional SIFT algorithm, the image uniform segmentation algorithm is selected to segment the image, which avoids the dense distribution of extracted feature points in the whole image, and improves the accuracy of the algorithm matching. The feature point matching in the traditional SIFT algorithm uses one by one calculation of the minimum Euclidean distance, which is computationally intensive and increases the time consumption of the algorithm, so the two-way approximate nearest neighbor algorithm is used for feature point matching and false matching elimination, which not only reduces the algorithm’s time consumption when searching for the feature point pairs in the high-dimensional vectors, but also eliminates the problem of false matching of the point pairs.