An Improved Retinex Algorithm for Underwater Image Enhancement Based on HSV Model

In underwater navigation and map construction, it is necessary to process and use the underwater image, but there are many impurities in the water and the water has strong light absorption and scattering effect. As a result, there are often some problems in the underwater imaging of the camera, such as color deviation, low contrast, dark brightness, serious noise and so on. These problems will directly lead to large errors in underwater mapping, so it is very important to enhance and update the image in real time. In this paper, an improved Retinex algorithm based on HSV(Hue-Saturation-Value) space is proposed to improve the above problems. Firstly, the RGB (Red-Green-Blue) model of the underwater image improved by Retinex is transformed into HSV model, and the high pass component in the converted V component is enhanced by NSST (Non-subsampled shearlet transform) to enhance the value component. Finally, the image enhancement is completed by inverse transform. Through experimental simulation, the improved Retinex algorithm proposed in this paper is compared with the original three Retinex algorithms, and it is proved that this algorithm has certain practicability and superiority in underwater image enhancement.