Research On Panoramic Texture Data Compression Algorithm For Resource-constrained Terminal

3D scene visualization technology is widely used in the fields of digital earth,digital tourism and 3D GIS.Texture rendering is the key to the scene reality in 3D scene visualization.The facade of building,sky,earth and so on need to use textures to increase the fidelity of the scene.It limits the development of 3D scene visualization,as excessive quantities of texture data are needed,which means network traffic and power consumption on low power terminal and low memory terminal such as mobile phone and IPad.Mip-Map and texture rendering of inverse procedural modeling are two of the most popular methods to reduce the amount of texture data.When the distance between the viewpoint and the scene changes in Mip-Map,Low resolution of texture and high resolution of texture are mutual exchanged,and the different resolution texture of the facade of building are transmitted and loaded multiple times,which increases the amount of traffic from the server to the resource-constrained terminal.Mip-Map is suitable for graph-based rendering because of depth information.Since the users need to reconstruct the cell texture into the original texture by texture rendering of inverse procedural modeling,the amount of calculation of rendering in terminal is increased.Based on the traditional 3D scene modeling method,this paper uses the panoramic modeling method with less 3D geometric model data to compress the texture image.This paper presents a panoramic texture data compression algorithm based on information entropy for resource-constrained terminal to realize the multilevel of details among same layers texture techniques used in cubic texture rendering.According to the different of information selection strategy of the human visual system when facing vary areas,we use texture with lower resolution to reappear low concerned area,so as to reducing the texture data and the resource consumption of the terminal.The scene area is distinguished by information entropy of texture,we use multi-resolution to reappear different area so that achieve multilevel of details among the same layer.We put forward iterated bi-cubic interpolation algorithm to eliminate the block effect of which image scaling rate is less than 0.5.The experimental results are based on the compression quality,real-time,mean square error(MSE)and peak signal-to-noise ratio(PSNR)and feature similarity(FSIMc).Using our method the texture data needed is reduced by 30% and time of rendering panorama is reduced by 30% with no visual perception change was found when texture rendering.