Non-uniform illumination image enhancement via symmetric brightness mapping and virtual multi-exposure fusion

Aiming at the problems of low contrast,unclear details and poor visualization in non-uniform illumination images due to over-dark or over-bright local regions, an image enhancement method via symmetric brightness mapping and virtual multi-exposure fusion is proposed. A color space conversion is used to retain the hue and saturation components and separate the brightness component of the input image for enhancement processing. According to the camera response model,the principle of image information entropy and average gradient maximization is adopted to estimate the optimal exposure ratio. A pair of symmetrical brightness mapping functions are designed to generate virtually corresponding images with the enhanced exposure and reduced exposure,forming an image sequence with different exposures together with the original brightness component. Then,a multi-exposure fusion method with detail enhancement is applied to the image sequence to reconstruct the enhanced result. Experimental results indicate that the average values of evaluation indices such as image information entropy,average gradient,image contrast, and color consistency of the proposed method are 7. 644,9. 209,450. 683 and 0. 962 on seven public datasets respectively,which are superior to those of the contrast method and achieve enhanced results with high dynamic range,strong contrast,clear details and good visualization.