Study Of Seal Identification Method

In the oriental countries, such as China, Japan and Korea, seal identification plays an very important role. With the development of the computer technique, difficulties of forging seal decline. Therefore economic crimes caused by forgery seals are hard to be forbidden. The purpose of seal identification is to provide a system that can identify tested seal images to be genuine or forgery, and it also can offer an automatic, efficient and accurate documentation authority identification way to avoid crimes. The correct verification rate of seals is affected by many factors, such as the technical expertise, current emotion of the operators and operational environment. So the manual seal verification has become a bottleneck of information management system, therefore an automatic method for seal identification is urgently expected. Based on several mature method in domestic and overseas, Genuine and forgery seal identification technique are studied systemic in this paper. The paper analyses the complex characteristic of seal image with stamp condition, and bring forward a pick-up red seal image method that based on HSI color model. Then processing seal image with Gray Scale and binary Scale. Using the methods of seal image enhance to improve process quality. Finally the good binary seal image can be obtained. Based on the good binary seal image can be obtained, the paper adopt two method of seal image identification. First, the paper adopts method based on circle projection and matrix changeless seal image identification. The circle projection method pick-up reflex seal image morphology and statistics domains as characteristic vector to identify seal image. The matrix changeless method base on pick up 7 target characteristic such as mass, shape center, distortion degree, kurtosis etc to identify seal image. Second, the paper adopts rotary matching method to identify seal image. The method is to calculate the shape center and rotary angle of seal image to exactly achieve image matching, And then difference match the two image, at last using Hough Transform calculate the number of beeline in difference matching image to get the discriminating result. Combining the two methods we can get the good result of correctness of 91%, so basically meet the aim of this research work.