Research On Multi Feature Fusion Retrieval Technology Of Cultural Relics

Heritage is a symbol of our national culture.The preservation of cultural relics are not only the protection of precious material cultural heritages,but also the inheritance of our forefathers' wisdom.As we all know,many cultural relics in China are damaged more or less in the excavation,and after the hand stitching,it is easy to encounter twice destruction.Therefore,the technology of fictitious culture relics fragments reconstruction aided by computer has gradually become a hot issue in the research field.In face of a large number of digital relics,how to retrieve the splice-needed fragments efficiently and accurately has become a hot research topic for the cultural relic workers.In this paper,multiple feature fusion is used to retrieve the fragments of cultural relics.The main works of this paper are as follows:(1)Aiming at the characteristics that there many concave and convex points on the fracture surface of the cultural relics and the totally opposite adjacent stitching fragments and the concave and convex points of the fracture surfaces,a new modified model detail feature extraction algorithm has been put forward.The algorithm is based on model size of fan beam geometry segmentation to extract the peak point calculation model,each segmented region and the saddle point,the peak point of decision tree algorithm and the saddle point to reduce the computation order,feature preserving feature points more significant regional characteristics of fracture surface vector.The experiment shows that the extracted model details can effectively express the feature of the fracture surface.(2)Aiming at the fusion feature algorithm,algorithm evaluation information based on fuzzy set membership function was came up with.It is used to optimize the total value coefficient of fusion feature to make the weight coefficient assigned to the model feature more reasonable.The experiment shows that the optimized fusion feature weight coefficient can improve the retrieval accuracy effectively.(3)aiming at the problem of searching efficiency of existing 3D cultural relics,an improved 3D model retrieval algorithm based on multi-feature fusion is proposed,and a simple 3D model retrieval system is designed by using this algorithm.First,the modified PCA(principal component analysis)is used to normalize the model.Secondly,the fusion(1)characteristics,details of fracture surface extracted from the global search and good effect of D2 distance histogram feature,extraction algorithm of the internal structure of retrieval feature extraction based on 3D radon transform has good effect,combined with the sample classification information entropy and fuzzy set membership function,the distribution of each feature weight coefficient.Finally,the algorithm is applied to more than 600 artifacts models and more than 200 models from Princeton model library.The experimental results show that the accuracy and efficiency of the algorithm are improved,and the universality of the algorithm is good.