RGB and Optimal Waveband Image Fusion for Real-Time Underwater Clear Image Acquisition
Xudong Sun, Zhu Yuan, Xianping Fu
Abstract
Images captured underwater generally degrade due to scattering and absorption. Although various breakthroughs have been made recently in the general area of underwater image enhancement and restoration, many state-of-the-art methods are still challenged in scene adaptability or real-time processing by the limitations of the RGB camera. The concept of optimal waveband is derived from remote sensing interpretation, which is the most appropriate imaging wavelength for species discrimination in a particular scene. Inspired by that, this paper proposes a novel underwater imaging method based on RGB and optimal band image fusion. Specifically, it consists of three main components: a band selection (BS) algorithm based on multi-criteria decision-making (MCDM) is proposed to select the most appropriate wavelength for underwater observation, an underwater stereo imaging system with color and optimal waveband (USICO) for collecting images in these two different modalities, and an iteration-free fusion model based on multi-scale decomposition (MSD) to fast generate the enhanced underwater images. Different from the conventional enhancement or restoration methods, the proposed system combines the advantages of the complementary data and enhances the image quality in imaging phase, avoiding the undesired image corruption caused by post-processing. The effectiveness of our imaging system is proved on a self-collected RGB-optimal waveband image pair dataset. Experimental results indicate that, compared with the underwater images captured by RGB camera, the images acquired by the USICO system have richer details and more prominent objects.