Based On Spectral Graph Theory To The Shape Of The Expression And Recognition Research

Through the shape identification of entity is an important means of computer vision recognition, this method in scientific research and social practice in the field has been widely used,target recognition, data retrieval, medical diagnosis and so on. People have been trying to be capable of making intelligent machines with the human identification and judgment ability. Shape is the first to be get of the information of vision, at the same time shape is also important characteristics of entity, Due to the shape of the entity identification is an important ability of intelligent machine.The60years since the last century, With the development of computer technology, many scholars of shape analysis and recognition has done a lot of research, and many achievements have been obtained, for example, based on the entity profile numerical methods circumference, Area, compactness (perimeter square/area), rectangle, centrifugal rate, shape signature and so on. Based on the non numerical method of contour chain code, the syntactic analysis,boundary estimation, scale space, boundary decomposition an so on. Based on the regional numerical method of moment, shape of the matrix and vector and so on. Based on the regional non numerical method, convex hull, the medial axis transform and so on.In this paper, we explored and discussed analysis of shape recognition method. Put forward a based on spectral graph shape description, to achieve the shape analysis recognition method. Compared with previous, shape descriptors is simple, easy to identify. Through the meshing shape generate triangulation network composed of delaunay triangle primitives, get in shape corresponding to the map and the matrix norm using the matrix of graph, formed on the shape measurement, the last classified these metrics, in order to achieve the shape analysis and recognition.This method of translation, scaling, rotation has the shape of invariance, on the shape of boundary disturbance has strong anti interference ability, and because Delaunay sensitivity, sensitive on the change of shape, be able to identify the shape of the subtle difference.