图象的模型及其统计特性

Models and Statistical Characteristics of Images

本文对图象的统计特性及其适用模型作了分析。讨论了以前常用的平稳图象模型的缺陷,对实际图象的统计特性作了分析,认为它是非平稳的,不满足各态历经性,同时在空间上是高度相关的;讨论了相应的随机参数统计模型和描述性统计模型。指出作用于整幅图象的“全局”图象复原算法比每次只独立计算单个象元的“点”图象复原算法更为优越。

The statistical characteristics and models of images are decribed in detail in this paper. The defects of the models which were in common use previously are discussed. The statistical characteristics of the models of some realistic image are analysed, with the results showing that they are nonstationary, nonergodic and high correlated spatially. The corresponding random parameter and descriptive statistical models are discussed and it is pointed out that the image restoration algorithms which operate on the global image (global estimator) will be superior to the point restoration algorithms which estimate 1 pixel at a time (point estimator).