具有细节补偿和色彩恢复的多尺度Retinex色调映射算法

Multi-Scale Retinex with Color Restoration and Detail Compensation

针对带色彩恢复的多尺度Retinex色调映射算法(MSRCR)在分离图像光照信息时未保留部分细节信息,导致结果图像出现细节模糊和颜色失真的问题,提出了一种具有细节补偿和色彩恢复的多尺度Retinex色调映射算法(MSRCD).该算法利用Retinex理论的基本原理将高动态范围图像分为反射层和光照层,先使用双边滤波从图像光照层中提取出细节信息进行补偿,然后从图像的反射层中分离出基本层信息并进行自适应调整,压缩其动态范围,最后通过色彩校正还原图像颜色.实验结果表明,与MSRCR算法及基于双边滤波的算法相比,MSRCD算法的处理结果保留了更丰富的细节信息,色彩逼近于真实场景且避免了光晕的产生.

An algorithm is proposed to solve the problem that the results of multi-scale Retinex with color restoration（MSRCR） are prone to have fuzzy detail and color distortion when it is used to separate lightness components.The proposed algorithm utilizes Retinex theory to divide the luminance into the components of lightness and reflectance.Then,the boundary of image is compensated by adding the detailed information that is extracted from lightness using bilateral filtering.The base-layer is extracted from the components of reflectance,and then is adaptively adjusted so as to compress the dynamic range of the high dynamic range image.Furthermore,the method can prevent color distortion and make the target images achieve better visual effect through color correction.Experimental results show that the proposed algorithm has better effect on dynamic range compression than both MSRCR and bilateral filtering methods have,and prevents halo artifacts,and that not only details but also color can be reproduced by the algorithm in visual point of view.