自然图象单像素灰度值的统计特性研究

本文针对自然图象中单像素灰度值的统计规律进行实验和分析，发现对数正态分布可以对之进行粗略、简单的建模。此外，本文还给出了判断所给定的数据序列是否为统计自相似数据的有效判断方法，并在此基础上断定自然图象具有明显的自相似特性，从而得到针对单像素灰度值且较对数正态分布更精确的建模。最后，本文还给出了对自相似参数日进行估计的两种有效策略。

Research on the Statistical Characteristics of the Unipixel Gray Value of Natural Images

This paper does some experiments and analysis on the single pixel gray value of natural images. We find that lognormal distribution can be used for the coarse and simple model of single pixel gray. Moreover, we propose an efficient method to judge whether the given data sequence is statistically self-similar. Using the method we get a conclusion that the natural images have distinct self-similarity,thus a better mathematical model than the lognormal model is obtained. Finally we give two efficient methods to estimate the self-similar parameter H.