改写该论文提出的算法主要用SIFT11方法提取关键点然后做匹配对匹配成功的关键点做聚类如果一个簇中有不少于3个的关键点则认为两个簇匹配。

The algorithm proposed in this paper primarily employs the SIFT[11] method to extract key points, followed by matching and clustering of the matched key points. If a cluster contains no less than three key points, it is considered a match between two clusters.