基于语义分割的三维模型版权保护算法OA北大核心CSTPCD

To solve the problem of reduced model accuracy caused by traditional digital watermarking techniques,and to protect local copyright,a three-dimensional model copyright protection algorithm based on semantic segmentation was proposed.In this algorithm,PointNet was first used for semantic segmentation of 3D model,and the model was divided into sub-blocks according to segmentation results.The copyright information was calculated in sub-blocks to protect local copyright.Then,a copyright information generation algorithm based on vertex norm and surface curvature was proposed.Since the copyright information was generated by means of coverless steganography,the method in this paper can protect the model as a whole and its sub-parts without modifying the model itself.Experimental results show that the proposed algorithm is robust to other kinds of attacks when it is completely resistant to similar transformation attacks.