A PCA AND PERTURB BASED FRAGILE WATERMARKING SCHEME FOR 3D MODELS

Abstract

In this paper, a new fragile watermarking scheme that enables the verification of 3D models for detecting unauthorized alterations is presented. The principal component analysis (PCA) is adopted to produce the PCA coordinate system and make the system robust to similarity transformation attacks. Before watermark embedding, the coordinates of illegitimate intervals are slightly perturbed to ensure all intervals legitimate. In order to control distortion, a special quantization technique is proposed to watermark a legitimate interval. Experimental results show that 100% vertices of the cover model can be embedded and extracted in the proposed method. Moreover, the distortion of the stego model is insignificant and under controlled by quantization step setting. Finally, the proposed method is immune to the causality, convergence and embedding holes problems.