改进的多重分形图像奇异性分析算法

为了准确地研究图像奇异性以及各部分的属性及特征,采用一种基于亚像素边缘测度的多重分形算法,该算法根据方形孔径采样定理计算亚像素位置的梯度面密度函数值和图像任意子集(半径可以达到亚像素精度)的边缘测度,进而利用多重分形理论将实际图像分割成一系列具有不同奇异性指数的分形集合.并利用含有不同信息含量的分形集合重建原图像算法,实现了图像从纹理到边缘各层面内容的精确划分.对该算法进行了理论分析和实验验证,得到3×3亚像素方法提取的边缘信息重构原图像,其峰值信噪比达到14.76dB.结果表明,重建图像峰值信噪比主要依赖于所提取的边缘信息质量以及重构系数比,提取的各层面信息与人类的视觉系统所捕获的重要信息相吻合.

An improved multifractal algorithm for analyzing image singularity

In this paper, a new multifractal algorithm is proposed based on sub-pixel edge measure. The method can precisely divide the image from texture to edge for image singularity analysis. The edge-measure of random subsets is obtained by computing the greylevel gradient distribution law on the position of sub-pixel. Utilized the multifractal frame, the image can be segmented into a series fractal sets of the different singularity exponents, the edge is the most singular multifractal subset. At the same time, the simple reconstruction algorithm is presented by using the different information content of multifractal subset. The Peak Signal-to-Noise Ratio of the reconstruction image depend on the extracted image edge quality. The reconstruction PSNR is 14.76 dB from the edge extracted by 3?3 sub-pixels method and the results are more identical with the human visual reception.