Modified Effective Histogram Equalization method for Night Time Color image enhancement with Energy Curve

By modifying brightness, contrast, sharpness, and color balance, color photographs can be made more visually appealing. By highlighting significant characteristics and reducing noise or distortion, the main aim of enhancement is to make the image more aesthetically visible, lucid, and interpretable. Techniques vary from basic brightness and contrast tweaks to sophisticated algorithms. Improved quality in low-light video is vital for distinguishing individuals and activities in security and surveillance. Challenges like noise amplification and over-enhancement can create unnatural images with exaggerated features. This paper addresses these issues by introducing the Exposure-Based Sub Histogram Equalization (ESIHE) method that uses an Energy Curve to enhance low-exposure or night color images effectively, which is similar to a histogram and based on the spatial contextual information of an image. To enhance outcomes, the suggested approach, ESIHE_Energy, combines an Energy Curve and Exposure-based Sub-image Histogram Equalization with spatial contextual information. The evaluation of the proposed approach was conducted on multiple datasets consisting of night-time color images. Its performance was benchmarked against several established methods, including Histogram Equalization (HE), Brightness Preserving Bi-Histogram Equalization (BBHE), Contrast Limited Adaptive Histogram Equalization (CLAHE), Dynamic Stochastic Histogram Equalization (DSHE), Recursive ESIHE (R-ESIHE), Recursive Symmetric ESIHE (RS-ESIHE), and ESIHE. Various image quality metrics, such as Peak PSNR, MSE, Entropy, Structural Similarity Index (SSIM), and Feature Similarity Index (FSIM), were utilized for comparison. The method yielded an average PSNR of 16.336, surpassing the majority of the other techniques. Notably, the proposed ESIHE method integrated with the Energy Curve delivered the best results, achieving a PSNR of 29.057, an MSE of 33.982, an SSIM of 0.969, and an FSIM of 0.917678. These results emphasize the significance of leveraging spatial contextual information to significantly improve image quality