An improved Contour and Gradient based Bionic Retinex Model for Image Enhancement

Low-light color images are often captured in real life due to environmental, equipment and other factors. The role of low- light color image enhancement is to enhance such images to improve their visual effects. Retinex theory decomposes an image into illumination component and reflection component. This paper proposed an improved contour and gradient based bionic Retinex model. Contour is detected using a method that simulates the human visual system and decomposes the image gradient into an illumination gradient and a reflectance gradient. The new Retinex model is built based on lighting and texture gradients for Retinex decomposition and image enhancement. Forward and inverse gamma transforms are applied to the image as the initial illumination and reflection components of the Retinex model. Experiments show that the model can accurately decompose the image into illumination and reflection components and enhance the low-light image naturally. Compared with other methods on 2 low-light image datasets, the proposed model has the best subjective evaluation and gets ahead in evaluation metrics.