Adaptive vertex quantization for mesh compression

Abstract

With the increasing adoption of 3D CAD and pervasive 3D visualization in PLM environments, efficient 3D compression is highly desired to reduce the size of mesh models for fast transmission and efficient storage purposes. Among existing vertex compression schemes, vertex quantization uses integers to represent floating point vertex coordinates to achieve smaller model size. However, the current uniform quantization approach can be further refined since vertices bear various visualization significance to viewers and justify various compression strength. Inspired by earlier research on geometric simplification, this paper presents the development of an adaptive quantization scheme that differentiates vertices based on their visual significance and performs quantization at different bit length. The initial experimental results indicate that the proposed approach enhances the performance of previous uniform quantization approach while maintaining good visualization quality. The application of this approach in role-based visualization is also discussed in the concluding part of the paper.