Research On Small Seal Style In Qin And Han Eave Tiles Characters Recognition Method Based On Content

Since the last century the research of graphic and image processing has been rapidly developed. Based on achievement from this field, the technology of character recognition also has been improved greatly. For the character we used now, recognition technology can meet our requirements no matter the character are in handwriting form or printing form. But for ancient character, there exist few recognition methods. The challenges include fickle fonts and random stroke order. Under engineering research center’s important project of Chinese Ministry of Education—the design and realization of computer system in classification of eaves tiles and recognition of Small seal style in Qin and Han dynasties, we use computer aided technology to study the Small seal style that are carved on eaves tiles. We design and realize the entire process to recognize the Small seal style on the basis of the specific layout structure of eaves tiles in Qin and Han period, and aim to inherit, protect and develop these historical and cultural heritages in a digital way.The innovation points of our work are as follows:1. According to the specific layout structure of eaves tiles, a region elimination algorithm and a character partitioning algorithm are suggested to extract every Small seal style font from eaves tiles correctly.2. During feature extraction, a new method that combines SIFT with mathematical morphology is put forward to avoid difficulties such as similar feature and missing feature, which is a good foundation for character recognition.3. To improve low efficiency of conventional classifier, a multilayer SVM classifier based on binary tree structure is used, which greatly improve the efficiency of character classification and recognition.