Self-tuning underwater image fusion method based on dark channel prior

Underwater images always suffer from low visibility and color distortion because of exponential attenuation along with scattering caused by the water property. In order to get a precise vision of the underwater environment, an underwater image fusion method is proposed in this paper. We use dark channel prior theory and histogram equalization to get a set of optimal images from the input images, and use the image fusion algorithm to preserve the best part of each image by specific weight assessment matrix. Downhill simplex algorithm is implemented to produce the optimal parameters of underwater dark channel prior model automatically from the initial parameters setting by optimizing the quality criterion based on image entropy. Meanwhile, this optimal procedure can ensure us to get the contrast enhanced image in different underwater environment simultaneously. Comparison experiments are carried out to prove the effects of this method. The experimental results show that our method can enhance the visibility and recover image color to get a better vision of underwater scenes when compared with other state-of-the-art methods.