离散余弦变换域内的Retinex图像增强模型研究  

Retinex Model in Discrete Cosine Transform Domain for Image Enhancement

作　　者：张新龙[1] 申永[1] 张璇[1] 汪荣贵[1] 

机构地区：[1]合肥工业大学计算机与信息学院,安徽合肥230009

出　　处：《工程图学学报》2011年第6期33-42,共10页Journal of Engineering Graphics

基　　金：国家自然科学基金资助项目(60575023;60705015);安徽省自然科学基金资助项目(070412054)

摘　　要：针对现有离散余弦变换(DCT)域图像增强算法大多统一处理DCT系数,且不能有效抑制块状效应的问题,将Retinex思想引入DCT域,提出一种新增强算法:将DCT系数分为入射分量和反射分量,构造一种带强光抑制的函数,将入射分量映射到理想的动态范围;定义一种频带限制的对比度,实现反射分量的局部细节增强;使用子块分解方法抑制块状效应。实验结果表明,与现有Retinex及DCT域增强算法相比,该算法具有更好的细节增强及强光抑制效果,且能较好抑制块状效应。Existing image enhancement algorithms by using of Discrete Cosine Transform mostly treat the DCT coefficients uniformly,and can not inhibit artifacts.By introducing Retinex idea into DCT,a new enhancement algorithm is proposed.The algorithm divides DCT coefficients into illumination component and reflection component.The dynamic range of illumination component is mapped to an ideal range,using a function with strong light bating.The local details of reflection component are enhanced by defining a band limited contrast measure.The block artifacts will be suppressed using a block decomposing method.Compared to MSR and DCT enhancement algorithms,results verify this approach’s efficiency of details enhancement and strong light bating,also with artifacts suppressed.

关 键 词：计算机应用 图像增强 离散余弦变换 RETINEX 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]