基于人眼视觉的红外图像增强算法研究

发布时间：2018-04-22 02:26

本文选题：红外图像增强 + Retinex　；参考：《激光与红外》2017年01期

【摘要】：红外图像具有低对比度、噪声大、动态范围大以及视觉效果差等特点。传统的图像增强算法具有各自的局限性,处理后的视觉效果并不理想。为了改善红外图像的视觉效果,增强图像中目标的识别力,研究了基于人眼视觉特性的Retinex算法,结合红外图像的成像机制,将适用于可见光图像的Retinex增强算法应用于红外图像,取得了良好的视觉效果。在此基础上,对算法进行自适应改进,提出了AMSR算法。采用几种传统的图像增强算法和基于Retinex的算法对具有典型特性的红外图像进行增强处理,并计算客观评价指标。通过对实验结果的主观评价和客观指标分析,验证了AMSR增强算法对红外图像具有适用性和优越性。本文的研究工作对红外目标探测和识别具有重要的意义。
[Abstract]:Infrared images are characterized by low contrast, large noise, large dynamic range and poor visual effect. Traditional image enhancement algorithms have their own limitations, and the visual effect after processing is not ideal. In order to improve the visual effect of infrared image and enhance the recognition power of target in the image, the Retinex algorithm based on human visual characteristics is studied. Combining with the imaging mechanism of infrared image, the Retinex enhancement algorithm suitable for visible image is applied to infrared image. Good visual effect has been obtained. On this basis, the algorithm is improved adaptively and the AMSR algorithm is proposed. Several traditional image enhancement algorithms and algorithms based on Retinex are used to enhance infrared images with typical characteristics, and objective evaluation indexes are calculated. Through the subjective evaluation and objective index analysis of the experimental results, the applicability and superiority of the AMSR enhancement algorithm to infrared images are verified. The research work in this paper is of great significance to infrared target detection and recognition.
【作者单位】：中国科学院上海技术物理研究所;中国科学院大学;
【基金】：上海物证重点实验室基金(No.2011xcwzk04) 国家“十二五”国防预研项目(No.41101050501)资助
【分类号】：TP391.41

【参考文献】