基于离散余弦变换的自适应图像增强

提出一种自适应的频域增强算法.对图像进行离散余弦变换(DCT),按照图像空间内容和频域系数的对应关系,对低频系数(DC)采用-αrooting算法进行调整,压缩图像的动态范围,增强图像照度分量中明区和暗区的细节.把高频系数(AC)分成4部分,按照人的视觉特性进行多尺度非线性调整,增强图像反射分量中大块边缘细节,抑制弱噪声细节,使图像的对比度得到有效的提升.同时由子块图像频域系数幅值分布特征确定低频系数调整函数的控制参数,消除了图像产生的块效应.实验结果表明增强后的图像清晰,层次分明,视觉效果得到了明显的改善.

An adaptive image enhancement algorithm based on discrete cosine transform

An adaptive image enhancement algorithm in the spectral domain is proposed.The image is translated to the spectral domain with the discrete cosine transform (DCT) algorithm.According to the corresponding relations between the spatial image and the spectral coefficient,the α-rooting algorithm is used to adjust the low frequency coefficients,which compresses image dynamics and enhances image details in the bright or dark regions.The high frequency coefficients are divided into the four parts.Every part coefficient is adjusted with multi-scale nonlinear operator based on human visual perception,which enhances the big bulk edge details,restrains the feeble noise and improves the image contrast.Moreover using the energy distribution of the DCT coefficients which is defined directly in each DCT block of an image to confirm the controlled parameter of low frequency coefficients,which is effective to eliminate the block artifacts.The experimental results show that the enhanced image is sharp and distinct,which has an obvious improvement about visual effect.