摘要
讨论玻耳兹曼关系式的状态数、熵及其常数与光学成像灰度级划分的谱关系.将热力学原理应用于光学成像过程的微观分析,提出实际图像、理想图像及其玻耳兹曼熵谱概念.认为实际图像灰度级处于平衡态,理想图像则是非平衡态的灰度级分布所构成.从平衡态到非平衡态转化是1个非自发过程,需要外部施加作用,利用平衡态和非平衡态之间的熵差分布实现图像分割.实验与分析表明图像分割的实质是在外力作用下实现从实际图像对理想图像的逼近.
This paper debates upon pedigree correlation of the microstate number, entropy and constant in Bohzmann equation with the corresponding optical imaging gray levels, applies principle of thermodynamics to microcosmic analysis of optical imaging process, brings forward some concepts of Bohzmann entropy pedigree, reality and ideality images, etc. , and believes that reality image gray levels are in equilibrium state, but ideality image is constituted of non-equilibrium gray levels. The transform from equilibrium to non-equilibrium states is a non-spontaneous process, needing to be given an exterior action. So it makes use of entropy difference distribution between the equilibrium and non-equilibrium states to realize the image segmentation. The experiments and analysis show that the essentials of image segmentation are approach to ideality image from reality image by means of the exterior action.
出处
《南京信息工程大学学报（自然科学版）》
CAS
2009年第2期106-111,共6页
Journal of Nanjing University of Information Science & Technology（Natural Science Edition）
基金
国家863计划(2007AA701510)
博士后科学基金(20080431336)
陕西省自然科学基金(2007D23)
关键词
灰度谱
光学成像
玻耳兹曼熵谱
图像分割
可分解马尔柯夫网
gray levels
optical imaging
Bohzmann entropy pedigree
image segmentation
decomposable Markov network
