Location algorithm for seal imprints on Chinese bank-checks based on region growing

Abstract

The paper focuses on location of seal imprints on Chinese bank-checks based on region growing. Region growing method can be applied to searching and locating connection region in an image. A seal imprint, however, is generally composed of various connection regions which are unconnected to each other. In order to locate the seal imprint, these connection regions must be fused together. In the paper, an algorithm for locating seal imprints on Chinese bankchecks based on region growing is proposed, of which a fusion criterion for connection regions in a seal imprint is put forth based on the image feature of Chinese bankcheck, and a center-rays model is proposed to find the topological relationship between connection regions, for which externally-tangent rectangle of region is used as the mark of location of region. The location experiment is achieved with a false-acceptance rate of 7.1% and a false-rejection rate of 0% on Chinese bankcheck.

References

[1] LeeSeong Whan, KimJin H.. Pattern Recognition, 1989, 22: 653-653

[2] HoriuchiTakahiko, RiouxMarc, HarukiRyoji. Pattern Recognition, 1996, 29: 719-719

[3] ChangShih-hsu, ChenFang-hsuan, HsuWen-hsing. Pattern Recognition Letters, 1999, 20: 1545-1545

[4] Xin-junYang, Zhao-qiWang, Wei-yiLiu, GuoguangMu. Journal of Optoelectronics · Laser, 2001, 12: 1177-1177

[5] DanaH., Ballard, ChristopherM.. Brown,Computer Vision, 1982, Englewood Cliffs, New Jersey, Prentice-Hall, INC.

[6] WangLiqiang, NiXuxiang, LuZukang. Chinese Optics Letters, 2003, 1: 520-520

[7] ChengS.-C.. IEE Proceedings-Vision, Image and Signal Processing, 2003, 150: 270-270

[8] TremeauAlain, BorelNathalie. Pattern Recognition, 1997, 30: 1191-1191