Trainable Nonlinear Reaction Diffusion (TNRD) based Low Light Image Enhancement

Abstract

In the present days, digitalised images are used in various applications. For these applications the images should have more quality. But, images captured in low light conditions are with low visibility and highly degraded. To improve the poor quality of the image, we proposed a novel technique named Low light image enhancement through Trainable Non-reactive Reaction Diffusion (TNRD). This technique is largely benefited from the training of the parameters and finally lead to the best reported performance on common test data sets for the tested applications. Our trained models preserve the structural simplicity of diffusion models and take only a small number of diffusion steps, thus they are highly efficient.