结合人眼视觉特性的红外图像增强新技术  

New technology of infrared image contrast enhancement based on human visual properties

作　　者：于天河[1,2] 戴景民[1] 

机构地区：[1]哈尔滨工业大学电气工程及自动化学院自动化测试与控制系,黑龙江哈尔滨150001 [2]哈尔滨理工大学计算中心,黑龙江哈尔滨150080

出　　处：《红外与激光工程》2008年第6期951-954,共4页Infrared and Laser Engineering

基　　金：黑龙江省自然科学基金资助项目(F200818)

摘　　要：红外图像具有边缘模糊和对比度较低的缺点,不适合人眼观察,所以要对其进行增强。但是现有的增强方法没有考虑人眼的视觉特性,视觉效果不好。提出基于小波的多分辨分析方法和Retinex图像增强算法相结合的红外图像增强方法,对红外图像不同的高频细节进行有针对性增强,同时用Retinex算法把人眼的视觉特性融入其中,能够使得增强的红外图像光照均匀,亮度适中,更适合人眼观察。算法既增强了图像的细节,又增强了图像的对比度,实验证明:该方法解决了红外图像低对比度和细节模糊的问题。The infrared image has disadvantages of low contrast and blurry edge, is unsuitable to observe through eyes, must be enhanced. The existing enhancement algorithms of infrared image bring problems that the noise is excessively enhanced and the details are distorted, therefore the visual effect of an enhanced image will be depressed. In this paper, the infrared image enhancement method combined wavelet multiscale with Retinex enhancement algorithm was proposed. The human visual properties was considered and the enhanced image was adapted to the human＇s view system. Firstly, the infrared image was decomposed into multi-level wavelet to obtain the multi-level wavelet and scaling coefficients. Secondly, every level of wavelet coefficients was enhanced by different nonlinear algorithms, and the scaling coefficients were processed by Multiscale Retinex. The proposed algorithm can improve the contrast and detail of the infrared image. Experiments show that the algorithm can solve the problems of low contrast and fuzzy detail.

关 键 词：小波变换 图像对比度增强 多分辨分析 人眼视觉特性 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]