基于视觉特性的真实影像再现技术进展及展望

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。

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

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.