DEEP LEARNING-BASED LOW LIGHT IMAGE ENHANCEMENT FOR IMPROVED VISIBILITY

Abstract

Low light conditions pose significant challenges for image capture and processing, leading to degraded image quality with reduced visibility and increased noise. Traditional low light image enhancement methods typically involve hand-crafted image processing techniques, such as histogram equalization, contrast stretching, and noise reduction filters. While these methods may provide some improvement, they often fail to produce visually pleasing and natural-looking results. The lack of adaptability and limited ability to learn complex patterns from data makes traditional approaches less effective in handling various low light scenarios. The need for an advanced low light image enhancement technique arises from the widespread application of imaging devices in low light conditions. Industries such as surveillance, automotive, and photography heavily rely on cameras to capture images in challenging lighting situations. By enhancing the visibility and overall quality of low light images, the accuracy and reliability of image-based systems can be significantly improved. Therefore, an intelligent approach that can learn and adapt from data becomes essential to tackle the limitations of traditional methods. In recent years, deep learning has shown remarkable potential in various computer vision tasks, including image enhancement. This project aims to explore and propose a deep learning-based approach to address the issue of low light image enhancement for improved visibility. The deep learning-based approach overcomes the limitations of traditional techniques by automatically capturing intricate patterns and features in low light images. This adaptability allows the model to generalize well across various low light scenarios, leading to visually appealing and realistic enhancements.