Underwater Image Restoration using Scene Depth Estimation Technique

Abstract

Underwater image restoration is a difficult task due to the shortage of reference image and changing underwater environment. The images are distorted by absorption and scattering eff ect of light. In deep water, light undergoes wavelength-dependent attenuation. Thus, red color light that has a large wavelength attenuates more than blue light and any other color. As depth increases, light attenuates most of the red content and image appears in bluish-green color. These images also have low contrast, color cast and hazy appearance. Existing methods may need specialized hardware or it may be based on multiple images of the same scene. Thus they cannot be used in real-time or video acquisition task. So, it's better to form an effective method for color enhancement and image restoration of images. Here depth map estimation along with image blurriness is proposed. As light travels deeper into the water, the image gets blurred and this is used to obtain depth map. The backlight is also obtained. These factors are substituted in the IFM (Image Formation Model) to restore the image.