Low-Light Image Enhancement with Retinex Theory Optimized by Improved Whale Optimization Algorithm

To solve the problems of insufficient contrast, blurred details, and color distortion in low-light image Enhancement, this paper proposes a Low-Light Image Enhancement with Retinex Theory Optimized by Improved Whale Optimization Algorithm(WOA). The method innovatively employs Cauchy chaos initialization to enhance the population diversity and global search capability of the WOA population, while introducing adaptive weight adjustment and Gaussian perturbation strategies to improve the local search precision of WOA. By dynamically searching the optimal combination of Retinex parameters—including Gaussian kernel standard deviations (\(\:\sigma\:\) ₁ ,\(\:\sigma\:\) ₂ ,\(\:\sigma\:\) ₃ ), brightness coefficient, bias (b), and color factor—through the improved WOA, a multi-metric joint optimization model is constructed for intelligent tuning of low-light enhancement parameters. Experimental validation on the LOL dataset demonstrates that compared to several traditional image enhancement algorithms (automatic white balance, adaptive contrast enhancement, histogram equalization) and the deep learning-based ZeroDCE method, the proposed approach significantly improves image clarity and contrast while effectively enhancing key metrics such as information entropy (IE), average gradient (AG), and perceptual quality indices (NIQE and BRISQUE). Ablation studies further quantify the contributions of individual improvements: Gaussian perturbation accounts for approximately 79.2% of performance enhancement, adaptive weight adjustment contributes 19.6%, while Cauchy chaotic initialization shows minimal impact (1.2%) in the current parameter configuration. The proposed WOA-Retinex method effectively balances natural appearance preservation with detail enhancement in low-light imaging scenarios.