三维几何数据压缩算法的累计误差消除方法
An Approach for Eliminating Accumulative Error for 3D Geometry Compression Algorithms
-
摘要: 已有的三维几何数据压缩算法中几何信息普遍采用连续量化和预测编码策略,存在较严重的累计误差,这一误差是由预测点的值在编码前后不一致带来的.针对此问题提出一种解决方案,借助一组具有值不变性的参照平面代替原有算法的预测点,该方案能有效地消除预测编码策略中产生累计误差的上述根源.理论和实验表明:该方法不仅在相同压缩比和速度下可进一步提高压缩算法精度,而且和多种几何压缩算法兼容.Abstract: Most of the existing 3D geometry compression algorithms employ two strategies: successive quantization and predictive coding, and induce accumulative error. The accumulative error is mainly caused by the inconsistency of values at predictive vertexes in predictive coding. In this work, we propose a novel approach to eliminate the accumulative error. The new approach employs a set of reference planes instead of predictive vertexes in coding. The values of the reference planes maintain consistency in coding, so the accumulative error can be eliminated efficiently. Both theory and experiments show that this approach can achieve higher precision and less distortion without reducing the compression ratio and processing speed. The approach is also compatible with most of the existing compression algorithms.
下载: