Research On Technology Of 3D Mesh Model Compression

3D Geometry Compression is representing the connectivity, geometry and attribute data with fewer bits as possible based on itself character. Then arranging these bits sequencely and compressing it, which is convenient for 3D geometry model with less data and higher efficency to store, process and display real-time in PC and network transmission. The complexity of 3D geometrical data requires new processing theories and new compressing methods to be researched.We start from the algorithms of 3D mesh model compression, and specially, do some researches on the compression of topological and geometrical data in polygon meshes, streaming format of large indexing mesh models and single-rate compression of topological and geometrical of streaming meshes. The main works and creations are as the following:1,According to single-rate compression and multi-rate compression, lots of 3D mesh compression main algorithms are re-categorized and concluded.2,After deeply studying the Face Fixer method based on edge conquering, a method of single-rate connectivity compression for Polygon mesh model using half-edge data structure and multi-order adaptive range coder is put forward. The comparing experiments show that:Range coder is superior to arithmetic coder in stability, anti-disturbance and encode velocity. For larger model, which compressibility is more valuable in practical use, the adaptive range coder is better than same order arithmetic coder for higher compression ratio.3,We study and experiment the single-rate geometry compression based on the parallelogram rule for Polygon mesh model. The compression rates are 7.24bpv～25.04bpv at different quantization levels vary from 8bit to 16bit. The result of experimentation demonstrates that:with increasing precision the achieved compression ratio decreases. And the predictive compression does not scale linearly with different levels of precision.4,A detail approach to converting a standard indexing mesh to a streamable format is put forward based on researching for streaming meshes theories. Our experiments show that the layout becomes more compact in streaming format. What's more, the preprocess time is shortened and RAM requires less.5,Study and make experiments on the single-rate connectivity compression and geometry compression of streaming meshes. The connectivity compression test shows that increasing the reusing vertexes and avoiding using dynamic references can improve the compression rates. During the geometry compression experiment, we summarize the detail process of lossless compression without quantization and carry out it and the lossy compression under uniformity quantization based on the parallelogram rule respectively. The experiments show that:although lossless compression cannot achieve same compression rate as lossy compression, it is an effective completing approach without a-priori knowledge about the precision or bounding box of the data. Finally, the Hausdorff distance and the RMS distance are used to evaluate the geometric approximation error between original triangular mesh models and their decompressed models. The experimental data indicate that the similarities increase with the promoted quantization levels. The lossless decompressed models are loss-free in quality. The decompressed meshes can restore original meshes very well when quantization value more than 10bit.6,It is the first time to compress the large scale 3D terrain data with the streaming compression method. The experiments show that:the terrain data reduce about 86.6% after lossless streaming compression. The geometry data reduce about 75.1% under 8～16bit quantization when using lossy compression. And the streaming compression algorithm requires less preprocess time and temporary disk space than out of core compression.