Photometry Data Coding for Three-Dimensional Mesh Models Using Connectivity and Geometry Information

Abstract

In this paper, we propose new predictive coding schemes for photometry data of three-dimensional (3-D) mesh models as per-vertex binding. We exploit geometry and connectivity information to enhance coding efficiency of the color and normal vector data. For color coding, we predict the color of the current vertex by a weighted sum of colors of adjacent vertices considering angles between the current vertex and the adjacent vertices. For normal vector coding, we generate an optimal plane using distance equalization to predict the normal vector of the current vertex. Experimental results show that the proposed coding schemes provide improved coding performance over previous works for various 3-D mesh models.