SSIFT: An Improved SIFT Descriptor for Chinese Character Recognition in Complex Images

This article analyzes SIFT algorithm which has a strong and distinctive description capability based on local features. In considering the structure information inside Chinese characters, a novel feature vectors named SSIFT is proposed, which combines with SIFT local vectors and global shape context vectors. The experimental results indicate that SSIFT could show the distinction of local structure, and distinguish the similar local parts of various characters. In to dealing with different changes situations, such as rotation, scale changes, influence by noise and complex background, the proposed descriptor is superior to SIFT.