New Approach to Underwater Image Dehazing using Dark Channel Prior

Abstract

Images captured in dense, hazy, foggy atmospheric conditions or in a surrounding filled with impurities, are exposed to deterioration of the captured image which lowers the contrast, color changes, and the object features observed are difficult to recognie by normal human vision and also by some outdoor computer vision systems. This paper analyses the enhancement in the visibility of a single degraded image. The single image is processed to give two or more images of different characteristics and features. The information from these images is used to generate a solitary image with more accurate information of the scene than the original images. To maintain the important features and information of the image for good visibility regions, various parameters are used as filters using dark channel prior technique. The resultant images are then combined and weighed to reduce the unwanted attributes. The resultant image eventually is improved as compared to the resource images.