Robust watermarking approach for 3D triangular mesh using self organization map

Abstract

The objective of this paper is to explore innovative ways to insert the maximum amount of secret information into 3D mesh models without causing perceptual distortion and also make it difficult for the attacker to guess where the watermark was inserted. It is based on clustering 3D vertices into appropriate or inappropriate candidates for watermark insertion using Self Organization Map (SOM). Two methods were used to embed the watermark into 3D model. The first method is a statistical approach that modified the distribution of vertex norms to hide watermark information into host 3D model while the second method is a mixed insertion of watermark bits into host model using vertex norm distribution and mesh vertices at the same time. The robustness of proposed techniques is evaluated experimentally by simulating attacks such as mesh smoothing, noise addition and mesh cropping.