Design And Implementation Of The Seal Localization And Registration System

The seal is authoritative. Seals have been applied to various fields. Currently,folding the seal multi-angle artificial remains the main method of identification seals. With the aid of the rich experience, the test result is subjective. The case of forging seals to obtain high profits is increasing. Due to lack of effective identification seal, the criminals can obtain money by commercial fraud. Artificial recognition seals may bring huge losses to the society. In order to solve this problem, finding a fast and accurate seal recognition method using the computer is an urgent problem need to be solved.Localization seals and registration seals are considered as the key steps in the whole process of seal authenticity identification. How to locate and register seals effectively becomes a serious problem. Therefore, this paper proposes a seal localization and registration method. The seal localization and registration system which is prepared for the seal authenticity identification is designed and implemented.First of all, this paper analyzes the characteristics of seal test images. The seal is overlapped with various colors of words and the image background is very complex. To address these concerns, using the advantage of HSI color space model,this paper adopts the HSI threshold control method to extract red pixels. Then the RGB image is transformed to the gray image.Secondly, the paper proposes a method of locating different shape seals to find the position of the seal in the test image. Rectangular seals testing method is put forward according to the characteristics of the digital linear. The line segment is the minimum unit of linears. Based on the above properties, the rectangle is detected.For ellipse seals, this paper puts forward a kind of detection method based on curve arc segment. The outline is divided into several arc segments. Random sampling points are taken into the elliptic general equation to calculate the parameter.Circular seal test method is similar to the ellipse seal test method.Thirdly, based on the Harris corner feature, this paper puts forward a method to register the reserved seal and the sample seal. Rough registration makes that the reserved seal and the sample seal are aligned on the spatial location. On this basis,traversal searching within the small angle scope is executed to find the best position so that it completes the accurate registration.Finally, this paper proposes quantifying registration results through the positive difference and the negative difference. By calculating the difference image,the sample seals can be divided into real seals and suspicious seal preliminary.In conclusion, a strategy of seal localization and registration method is proposed in this paper. The design and implementation of the seal localization and registration system is completed. The test results show that the system function is correct. It lay a good foundation for the seal identification, showing that this appraising theory has practical value.