Study On Critical Technologies In 2D-graphic Search System

The method for geometrical shape similarity matching among 2D graphics is the core of the graphic retrieval system. The performance of the retrieval system depends on the efficiency of the algorithm. However, there is not any perfect algorithm used widely in the field of graphic retrieval. In this paper, according to the inherent features of 2D graphics, we try to discuss respectively the similarity matching algorithm from the perspective of angles and vertexes. The angle-based approaches transform these interior angles into formula of vectors and judge the similarity between 2D graphics from the cosine value of the angle between two vectors. The method above can sort the matching results. The vertex-based approaches normalize the coordinate of the polygons before judging their similarity, that is, to normalize their coordinate origins, coordinate ratios and coordinate rotations. This approach judges the similarity of polygons by comparing the number of coincident vertexes. At the same time, the algorithm presented here supports defining the coincidence degree of vertexes by resetting the comparing accuracy. Besides, we present here an algorithm of polygon simplification and an algorithm of vertex expansion both which are used in the retrieval model. In this paper, each algorithm has a detailed process and a related experimental result.