Single underwater image enhancement based on the reconstruction from gradients

Since the light is absorbed, reflected and scattered during the transmission process, underwater images are degraded suffering from color casts and low contrast. In the paper, an effective enhancement method for single underwater image is proposed based on the reconstruction technique from gradients. The method firstly corrects color by a simple and effective white-balancing approach after the compensation of red attenuation. And then an improved DCP method is used to estimate the transmission indicating the absorption and reflection of light, based on a combination filter because of edge preservation and high efficiency. Last, the enhanced gradients are obtained based on the estimated transmission, and details enhancement is accomplished by image reconstruction from the enhanced gradients on the luminance layer in Lab color space. Experimental results show that our method achieves better visual results in qualitative and quantitative evaluation, which can effectively recover the color and improve the image contrast even in very dense regions. Additionally, running time shows that our method is competitive and can be applied for real-time tasks.