基于视觉特性的真实影像再现技术进展及展望  

New Progress and Prospect of Realistic Image Rendition Based on Visual Characteristics

作　　者：倪国强[1] 肖蔓君[1] 胡宏清[1] 陈思颖[1] 黄光华[1] 

机构地区：[1]北京理工大学信息科学技术学院光电工程系,北京100081

出　　处：《中国激光》2007年第4期451-460,共10页Chinese Journal of Lasers

基　　金：国家自然科学基金(60377038)资助项目

摘　　要：视觉特性研究的成果,融入到真实影像再现技术,可提供更适于人类观察的高品质图像,有利于机器视觉的研究。首先简单总结当前主要的视觉特性研究成果,包括视觉空间对抗、光谱对抗、视觉适应、视觉皮层初级处理等。综述了基于视觉特性的真实影像再现技术的研究动态和新进展,并根据算法的侧重点不同,分析了以保持颜色恒定性为主要目标的Retinex系列算法、各种动态范围压缩算法等。其中也涉及到已开展的部分工作,从原理分析、实验效果对比等角度,对研究成果与国外优秀算法做出评价。介绍了国外在软/硬件平台实现上的最新研究成果。简述进一步开展的工作及研究方向,并就其发展方向做出展望。Visual characteristics have been successfully applied to the study of realistic image rendition, which may provide higher quality images for human observation and machine vision. At first, the latest main researches on visual characteristics are briefly summarized, including visual spatial opponent, spectral opponent, visual adaptation, and visual cortex primary processing and so on. The development of realistic image rendition techniques based on visual characteristics is summarized in this paper. Since various algorithms have different emphases, the Retinex family methods primarily focusing on keeping color constancy and some other methods aiming at getting high dynamic range compression are analyzed in detail. Our study on this field is also mentioned and compared with some excellent algorithms according to theory structure and experiment results. Related representative software and hardware realization techniques are introduced, and our next research direction and the prospect on this field are given.

关 键 词：图像处理 真实影像再现 视觉感知 颜色恒定性 明度 动态范围压缩 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]