非均匀并行图像压缩方法研究

Parallel Algorithms For Image Coding Based on Unbalance Segmentation

给出一种基于非均匀分割的并行图象压缩方法。根据灰度值的局部变化把图像分割成不等长的矩形块（block），子块内各点值用矩形顶点值线性插值近似代替。改进了传统图象分割方法[1～3]，给出了一种在不降低压缩比条件下的并行分割压缩方法，提高了压缩速度；同时，还给出一种确定分割阀值的新法－累积2乘误差方法，使生成图像的信噪比提高了1～2dB。对于一类图像，该方法压缩效率优于JPEG方法．

A parallel algorithms has been given for image coding based on unbalance segmentation. Image has been splited into rectangle(sub-block) that is not equal. The greys value of points in sub-block is bordered with vertex interpolation. We improved classical segmentation methods, gived a parallel algorithms and a new methods to assign spliting condition, have got a more speed in image compression and a better quality of reconstructural.