Abstract: In terms of the problem of symmetry flip,it is always one of the difficulties in shape matching,in which feature point selection,symmetry point detection and initial matching have a great influence on the final matching result.In view of this problem,this paper proposes a 3D symmetry shape matching algorithm based on HKS.Besides,a new method using HKS as a tool is proposed considering the shortage of geodesic in symmetry shape detection on the basis of previous work.Firstly,it combines the HKS sampling and the farthest point sampling and fusing the sampling,so as to obtain the characteristic points with stable distribution and representativeness.Then,it uses the HKS to classify the characteristic points to realize the accurate detection of symmetrical points.After that,it uses the one-point matching algorithm based on HKS to carry out the initial matching and improve the accuracy.Finally,it uses the LS_MDS algorithm to embed the shape into the Euclidean space and project it into 2D.According to the normal direction of the vertex in the projection and the curvature information on the grid,the direction of the shape is calibrated,and the symmetric matching is finally completed.In the TOSCA database,it carries out the numerical experiments,in which the results of the comparison between the methods in this paper and the existing classical methods show that the method not only improves the matching accuracy,but also is suitable for the relatively special symmetrical confusion problems,such as symmetry shape matching in gorillas.