BRIGHTNESS CONTROLLED IMAGE ENHANCEMENT DIFFUSION MODEL

Abstract

Low-light image enhancement can be defined as the process of improving the visi bility and quality of images captured in poor lighting conditions. The goal is to make these images clearer and more aesthetically appealing by increasing luminance, reduc ing noise, and refining details. However, many present algorithms focus on increment ing the brightness uniformly and to a particular extent which limits the experience of a user. This model utilizes a diffusion model that is conditioned on an illumination embedding. This framework allows the model to improve images in an iterative man ner slowly refining the image quality by reducing noise and improving brightness. The illumination embedding serves as a control mechanism, enabling users to specify their desired brightness levels. This approach offers a high degree of adjustment which al lows for tailored enhancements according to choice of the user. The embedding of the SAM gives freedom to users to choose particular areas in an image for the de sired upgradation. This attribute improves user experience by providing an intuitive procedure for confined adjustments making the process efficient.