3D Model retrieval based on fuzzy weighted shape distributions

Shape distribution is considered as a kind of art state 3D model retrieval algorithm due to its simplicity, robustness, and not need model pretreatment. Its disadvantage is that the retrieval precision is not high enough. Despite the introduction of five kinds of shape functions, each of which can not sufficiently expresse the 3D models. This paper uses the D2 distribution, the improved D1 distribution and the total surface area method to retrieve 3D models respectively. Then give weights to each method after unitary them. Set fuzzy rules are set to decide the fuzzy weights of three methods according to the mean proportion and variance proportion, Experimental results show that this algorithm can improve the retrieval results significantly.