A Underwater Image Enhanced Algorithm combined MSR and Channel Energy

Abstract

Vision is the most important source of information for Unmanned Underwater Vehicle(UUV), only a clear image can make it possible for a UUV to operate automatically, Due to strong absorption and light scattering for water, the image of underwater imaging has a serious degradation, whether in distance or in quality. Based on classical Retinex theory, this paper proposed an underwater image enhanced algorithm. For different channel of underwater image has apparently different energy, especially green channel has overwhelming power than others, we added a step to energy balance of the RGB three channel in classical Multi-Scale Retinex (MSR). Experiment data show that, compared to the traditional three-channel simple addition, our improvement can get more natural recovered underwater image.