像素点的邻域分形近似及其分形编码

Fractal Approximation of Pixel Dots in Its Neighborhood and the Fractal Coding

提出了一种新的图像块的分形近似方法，该方法借助于像素点的邻近点以达到对图像块的最小均方分形近似．这实际上是寻找仿射变换Ｔ，使图像块各像素点的邻近点压缩到自身时，均方差最小．量化后的仿射变换Ｔ构成了图像块的分形块编码．对每一图像块，仅仅解二阶线性方程组．用该分形块编码对图像编码。

A new method for the fractal approximation of an image block is presented. The neighboring points of the pixel dots are used to realize the least mean square fractal approximation of the image block. This is in effect to find an affine transformation T in order to map all the neighboring points of each pixel in the image block to be compressed into itself and thus the mean square deviation is minimized. Quantized affine transformation T constitutes the fractal block coding of the image block. For each image block, the computation involves the solving of a system of second order linear equations. The fractal block coding is applied to the image encoding and high compression ratio and real time operation can be achieved.