Color Image Enhancement Based on the Improved Retinex
Hongqing Hu, Guangqiang Ni · 2010
Considering the limitations of traditional centre / surround Retinex algorithm, an improved Retinex algorithm is proposed in this paper. The improved algorithm uses bilateral filtering on the each pixel and its neighborhood instead of employing traditional Gaussian filter, which makes the estimated background luminance maintain a good discontinuity in light mutation. Therefore, the proposed approach can reduce the halo artifacts effectively. In addition, to model the local adaptation characteristics of human visual system, a local contrast enhance function is introduced to adjust the intensity of each pixel adaptively, and to enhance the local contrast of images effectively. Based on this improved Retinex algorithm, a novel method for color image enhancement is proposed finally. Experimental results demonstrate the effectively of the proposed method.