遗传退火算法的模型相似性计算方法

Method of calculating model similarity based on genetic annealing algorithm

为了检索最相似的CAD模型,本文结合遗传算法的全局搜索能力和模拟退火算法的局部搜索能力,提出了基于遗传退火算法的模型相似性度量方法。利用面的边数差异来计算源模型面与目标模型面之间的形状相似性。结合面的形状相似性和面的邻接关系来计算面的结构相似性。以面的形状相似性和结构相似性为基础,构造2个模型的整体相似度矩阵。利用遗传退火算法对该矩阵进行搜索,得到2个模型之间的最优面匹配序列。以最优面匹配序列为基础,计算2个模型的相似性。实验结果表明:相对于模拟退火算法,本文所提出方法使13.33%的模型的排序效果有所改善。该方法能够更准确地度量2个模型之间的差异。

To retrieve the most identical CAD model,this study proposes a model similarity calculation method based on a genetic annealing algorithm,which combines the global searching ability of a genetic algorithm and the local searching ability of a simulated annealing algorithm.The difference of the faces′edge numbers is applied to compute the shape similarity between source and target faces.The shape similarity and adjacency relation of the faces are combined to calculate the structural similarity.Based on the shape and structural similarities of faces,a global similarity matrix of two models is constructed.The genetic annealing algorithm is used to find an optimal matching sequence of faces between the two models.Based on this optimal matching sequence of faces,the similarity of two models is calculated.Experimental results show that compared with simulated annealing algorithm,the proposed method improves the ranking effect of 13.33%of models,which proves that it can measure the difference of two models accurately.