Color image enhancement based on retinex theory with guided filter
Shi Tang, Mingjie Dong, Jinlei Ma, Zhiqiang Zhou, Changqing Li · 2017
Color image enhancement is widely used in digital image processing. Retinex performs well in color image enhancement, however, traditional Gaussian filter-based retinex algorithms exist some problems such as halo artifacts and detail loss. To solve these problems, we propose an improved retinex image enhancement algorithm based on the guided filter, which is processed in IHS color space. We replace Gaussian filter with the guided filter to get the detail information in different fine scales to better enhance different bands of high-frequency information. Then, we also extract a certain amount of low-frequency information through the decomposition with guided filter in the log domain, while the retinex method based on Gaussian filter only extracts the high-information to enhance the image. Next, we enhance the high-frequency information of the image and combine the enhanced high-frequency information and low-frequency information to get the combined image. Finally, we stretch the combined image to enhance the contrast of the image. In this way, we get the result image with enhanced details and contrast. Compared with some existing retinex methods in image enhancement, our algorithm can avoid the halo artifacts and detail loss.