欢迎访问《图学学报》 分享到：
图学学报

图学学报

• 图像处理与模式识别 • 上一篇    下一篇

离散余弦变换域内的Retinex图像增强模型研究

  

  • 出版日期:2011-12-30 发布日期:2015-08-12

Retinex Model in Discrete Cosine Transform Domain for Image Enhancement

  • Online:2011-12-30 Published:2015-08-12

PDF (PC)

183

被引次数

2

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。

关键词: 计算机应用, 图像增强, 离散余弦变换, Retinex

Abstract: 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.

Key words: computer application, image enhancement, discrete cosine transform, Retinex