Three-dimensional model similarity based on improved shape distribution algorithm
Xionghui Zhou · Computer Integrated Manufacturing Systems · 2007
To eliminate poor distinguish ability of traditional shape distribution algorithm,two improvements based on traditional D2 distance function were put forward: Firstly,classify sample point pairs into different categories to represent model's convex and concave features distribution according to the geometric property of the line connecting them.Secondly,introduce the angle information formed by the normal vectors of adjacent area so as to distinguish polyhedral box shaped parts,revolving parts and free-form surfaces.Case study proved that the improved shape distribution algorithm had better distinguishing ability and precision.