一种新的NSCT域图像增强算法  

A NEW NSCT DOMAIN IMAGE ENHANCEMENT ALGORITHM

作　　者：全永奇 邓家先[1] 寇计萍 霍荣[1] 

机构地区：[1]海南大学信息科学技术学院,海南海口570228

出　　处：《计算机应用与软件》2016年第3期206-209,221,共5页Computer Applications and Software

基　　金：海南省自然科学基金项目(613155);海南省科技兴海专项资金项目(XH201311)

摘　　要：针对实际应用中所采集的部分图像对比度低、边缘细节模糊的问题,提出一种基于非下采样Contourlet变换NSCT(Nonsubsampled Contourlet Transform)的多尺度Retinex与非线性增益函数相结合的图像增强算法。使用改进的多尺度Retinex算法对低频子带系数进行处理,以提升图像的灰度动态范围并改善图像的亮度均匀性;采用非线性增益函数和贝叶斯萎缩阈值相结合的方法对各个高频子带系数进行处理,在提升图像纹理细节的同时抑制噪声。实验结果表明:该算法能够有效提升图像对比度和清晰度,增强图像细节信息,有效改善视觉效果。Part of the images collected in actual application has low contrast degree and blurry edge details. Aiming at this problem,we proposed an image enhancement algorithm which combines the NSCT-based multiscale Retinex algorithm with nonlinear gain function. It uses the improved multiscale Retinex algorithm to process the low-frequency sub-band coefficients so as to improve the dynamic range of image greyscale and to meliorate the brightness uniformity of image. The method of nonlinear gain function and Bayesian shrinkage threshold combination is adopted to process the coefficients of each high-frequency sub-band,which suppresses image noise while enhancing image texture details. Experimental result showed that the algorithm could effectively enhance the contrast degree and definition of the image,boost up image details information,these led to noticeable improvement in visual quality.

关 键 词：图像增强 非下采样CONTOURLET变换 多尺度Retinex法 非线性增益函数 

分 类 号：TP751.1[自动化与计算机技术—检测技术与自动化装置]