Towards Qualitative Advancement of Underwater Machine Vision with Generative Adversarial Networks
share
Underwater machine vision attracts more attention, but the terrible quality prohibits it from a wide range of applications. There exist many algorithms to solve this problem, but the real-time and adaptive method is deficient. In this paper, based on filtering and Generative Adversarial Networks (GANs), two approaches are proposed for the aforementioned issue, i.e., filtering-based restoration scheme (FRS) and GAN-based restoration scheme (GAN-RS). Distinct from the previous, FRS restores underwater vision in the frequency domain, which is composed of parameter search, filtering, and enhancement. Improving the image quality further, the GAN-RS can adaptively clear underwater machine vision in real time without any pretreatment. In particular, the information in Lab color space and the dark channel is developed as the loss function, namely underwater index loss and dark channel prior loss. More specifically, learning from the underwater index, the discriminator is equipped with a carefully crafted underwater branch to predict the underwater probability of an image. Multi-stage loss strategy, then, is designed to guarantee the effective training of GANs. Through extensive comparisons of image quality and applications, the superiority of the proposed approaches is evidenced. As a result, the GAN-RS achieves a quite high processing speed and makes progress on color correction, contrast stretch, and detail emergence of the changeable underwater scenes.
READ FULL TEXT