Photorealistic Geometry Image Coding Based on Directional Lifting Wavelet Transform

Abstract

Nowadays computer graphics over the Internet has attracted a lot of attentions; however transmitting 3D model takes a long time. In this paper, a novel framework of geometry image coding is proposed which utilized the image compression standard to represent 3D model. The mesh data of 3D model is first converted into geometry image and normal map image, and then normal map image can be predicted by geometry image. Hence, only the geometry image and residual normal map image should be transmitted. Moreover, these images are compressed using directional lifting wavelet transform. According to the experiments, both objective and subjective performances of 3D model are enhanced, the photorealistic rendering is improved and the time of transmission is reduced. Most importantly, this proposed framework is more adapted to integrate into industry since it is based on improved JPEG2000 standard.