Underwater color image enhancement using improved multi-scale retinex and histogram linear quantification
Qingwu Li, Wenqing Zhu, Cao Mei · 2017
The quality of underwater color image is degraded due to scatting and attenuation of the illumination in the underwater environment. In order to compensate for visual degradation, an underwater color image enhancement method by imitating human visual mechanism of processing color images is proposed. Firstly, the incomplete Beta function is applied to underwater color images to enhance the brightness of the image. At the same time, the improved multi-scale homomorphic filter with a weighted sum of three different scale homomorphic filters, instead of Gaussian filter in the Retinex algorithm, is employed to estimate the illumination image so as to remove image haze. Finally, the histogram for each channel of the RGB color space is analyzed, and the histogram linear quantification algorithm is applied to achieve color balance. The performance of the proposed bionic model is evaluated both subjectively and objectively. Experimental results demonstrate that the algorithm proposed in this paper can improve the definition and balance the color of the underwater image effectively.