A Noise Eliminating and Point Matching Algorithm for Seal Image Verification

Seal is widely used in oriental countries. Seal verification is performed by matching an input seal image with a reference seal image that was registered beforehand. This paper proposed a noise eliminating and point matching algorithm for seal image verification. The proposed algorithm includes two procedures: seal detection and seal verification. In seal detection procedure, background text is eliminated by comparing color component intensity; then red region is detected; finally, multi-iterations of 4-neighbors noise eliminating algorithm are performed to degrade noise. In seal verification procedure, the center of a seal is obtained by the projections in x-axis and y-axis; then a mask is defined to reserve text in the seal, and the text in the seal is reserved by mask operation; finally, the input seal image is rotated to get maximum matching similarity level. The experiment results showed that the algorithm is with low error rate and is promising for seal verification.