图象扫描波动性度量及Hilbert扫描矩阵的快速生成

Measuring the Undulation of the Image Scanning and a Fast Hilbert Scanning Matrix Constructing Algorithm

在图象扫描技术中 ,首要的问题是如何尽量减小生成数据的波动性 .为此在深入研究各种扫描技术的基础上 ,定义了一种描述扫描方式产生数据波动性的度量指标 ,基于图象存在局部连续性 ,证明了Hilbert分形曲线扫描优于其他传统扫描 ,表明其生成数据的最佳连续特性 ,并通过实验分析了图象扫描方式对DCT变换编码效果的影响 .最后 ,给出了一个构造Hilbert扫描矩阵的快速算法 ,以方便该方法的进一步研究及应用 .

The image transformation coding is classified into two categories:one-dimension transformation coding and two-dimension transformation coding. One-dimension transformation coding need little computation, proceeds fast, but its effect is dependent on the image scanning techniques. The central problem of the image scanning techniques is how to decrease the undulation of the generated data. In this paper, based on the deep researches, and objective quantity exponent is defined to measure the undulation of the scanning techniques. By it, with the help of computer experiments, it is proved that the Hilbert scanning technique is superior to the other scanning techniques and the data it generates is the most continuous based on the hypothesis that the images are continuous in the small range. The effect to the DCT coding of the scanning techniques is analyzed by computer experiments. Finally, a fast recurrent algorithm to construct the Hilbert scanning matrixes is given out for easy to use this scanning technique to code the images and to go on more researches.