SURF algorithm with color and global characteristics

Correspondence point matching is one of the necessary jobs in computer vision and it is not easy to find correspondence point in various environments, like environments with changing scale, rotation, view point and illumination. SURF algorithm which based on local geometric feature works faster than SIFT while maintaining matching performance. However, SURF algorithm can only use gray level information rather than color information. Moreover, this algorithm is difficult to perform matching on similar local patterns because only local information is used. Therefore, an extended SURF algorithm which can make use of robust color information and global content is proposed. Experiments of the proposed algorithm are made in case of illumination variation and similar local pattern matching, and its superiority in speed and accuracy compared to SURF and SIFT algorithm is proved.