Adaptive Image Enhancement via Local Processing in Double Channel

In order to adaptively process various kinds of degraded images and effectively recover their details without manual intervention, an image enhancement algorithm is proposed that locally processes images in double channel. Firstly, in HSV color space, images are analyzed locally to calculate the appropriate enhancement functions in two channels, and then to be enhanced using these functions; then, enhancement results in two channels are weightedly merged using mixed Gaussian functions into enhanced intensity components; finally, a color compensation method is employed that considers intensity transformation to calculate compensation function and correct the color distortion resulting from the enhancement. Furthermore, a fast sliding window technique is applied to significantly reduce the computational cost. Experimental results show that the proposed algorithm can flexibly process images with different characteristics such as underexposed, overexposed, backlight, and misty images, as well as complex images of all the mentioned characteristics, the performance and adaptive ability are much better than other methods.