基于渐进插值的Catmull-Clark双正交细分小波及其应用

发布时间：2018-05-07 17:25

本文选题：渐进插值 + Catmull-Clark小波　；参考：《计算机辅助设计与图形学学报》2017年06期

【摘要】：针对lifting双正交Catmull-Clark细分小波在数据压缩、噪声滤波和低分辨率模型稳定性等方面的不足,提出基于渐进插值的Catmull-Clark双正交细分小波算法.对于任意拓扑的四边形网格,Catmull-Clark细分的极限曲面渐进插值于原有网格控制点,并且相邻2次细分之间网格改变程度很小,导致大量小波系数值趋于0,非常适于用零树编码提高3D网格的压缩性能;同时,该小波变换具有局部正交性和对位计算等特点,可显著减少内存占用量和计算复杂度.实验结果表明,与同类算法相比较,该算法在压缩效率、噪声滤波和低分辨率模型曲面的稳定性等方面均有明显提高,其中压缩编码Bits/vertex值减小14%,重构模型PSNR值增大5%,编解码耗时分别减少6%和9%.
[Abstract]:Aiming at the shortcomings of lifting biorthogonal Catmull-Clark subdivision wavelet in data compression, noise filtering and low resolution model stability, a Catmull-Clark biorthogonal subdivision wavelet algorithm based on progressive interpolation is proposed. For arbitrary topology quadrilateral mesh Catmull-Clark subdivision, the limit surface is interpolated to the original grid control point gradually, and the grid change degree between adjacent subdivision is very small. As a result, a large number of wavelet coefficients tend to be zero, which is very suitable for improving the compression performance of 3D meshes by using zero-tree coding. At the same time, the wavelet transform has the characteristics of local orthogonality and para-bit computation, which can significantly reduce the memory footprint and computational complexity. The experimental results show that compared with the similar algorithms, the proposed algorithm can significantly improve the compression efficiency, noise filtering and stability of low-resolution model surfaces. The compression coding Bits/vertex value is reduced by 14%, the reconstruction model PSNR value increases by 5%, and the coding and decoding time is reduced by 6% and 9% respectively.
【作者单位】：清华大学计算机科学与技术系;中国人民解放军总医院计算机应用与管理科;中国卫星海上测控部试验技术部;
【基金】：教育部博士点基金(20130002110002)
【分类号】：O157.5

【相似文献】