Underwater Image Enhancement With a Deep Residual Framework

Abstract

Owing to refraction, absorption, and scattering of light by suspended particles in water, raw underwater images have low contrast, blurred details, and color distortion. These characteristics can significantly interfere with visual tasks, such as segmentation and tracking. This paper proposes an underwater image enhancement solution through a deep residual framework. First, the cycle-consistent adversarial networks (CycleGAN) is employed to generate synthetic underwater images as training data for convolution neural network models. Second, the very-deep super-resolution reconstruction model (VDSR) is introduced to underwater resolution applications; with it, the Underwater Resnet model is proposed, which is a residual learning model for underwater image enhancement tasks. Furthermore, the loss function and training mode are improved. A multi-term loss function is formed with mean squared error loss and a proposed edge difference loss. An asynchronous training mode is also proposed to improve the performance of the multi-term loss function. Finally, the impact of batch normalization is discussed. According to the underwater image enhancement experiments and a comparative analysis, the color correction and detail enhancement performance of the proposed methods are superior to that of previous deep learning models and traditional methods. INDEX TERMSAsynchronous training, edge difference loss, residual learning, underwater image enhancement.

Fast Underwater Image Enhancement for Improved Visual Perception

“…On the other hand, several single image enhancement models based on deep adversarial [15,43,28] and residual learning [29] have reported inspiring results of late. However, they typically use only synthetically distorted images for paired training, which often limit their generalization performance.…”

“…Several models based on deep Convolutional Neural Networks (CNNs) and Generative Adversarial Networks (GANs) provide state-of-theart performance [21,10,25,47] in learning to enhance perceptual image quality from a large collection of paired or unpaired data. For underwater imagery, in particular, a number of GAN-based models [15,43] and CNN-based residual models [29] report inspiring progress for automatic color enhancement, dehazing, and contrast adjustment. However, there is significant room for improvement as learning perceptual enhancement for underwater imagery is a more challenging ill-posed problem (than terrestrial imagery).…”

An Improved Enhancement Algorithm Based on CNN Applicable for Weak Contrast Images

“…The image verifies the feasibility of the network and that it achieves good visual results. Then, quantitative methods are used to demonstrate the advantages of the proposed algorithm through comparing the Peak Signal to Noise Ratio (PSNR) and Structural Similarity Index (SSIM) of images with those of other two kinds of widely used algorithms: one is traditional image enhancement algorithms based on parameters, such as the CLAHE (Contrast-Limited Adaptive Histogram Equalization) algorithm [15], the SSR (Single-Scale Retinex) algorithm [19], the ABC (Artificial Bee Colony) algorithm [20] and the DOCS (Distance Oriented Cuckoo Search) algorithm [21], and the other is deep learning based image enhancement algorithms, for instance the CAEN (Convolutional Auto-encoder Network) algorithm [24], the DCNN (Deep Convolutional Neural Networks) algorithm [25], the DRF (Deep Residual Framework )algorithm [26] and DHN (Deep Hybrid Network) algorithm [27]. In addition, an ablation experiment on the convolution kernel size is conduct to indicate the advantage of proposed network in this paper.…”

“…Kuang et al proposed a deep learning method for single infrared image enhancement, and the conditional generative adversarial networks were incorporated into the optimization framework to avoid the background noise being amplified [25]. Liu et al proposed an underwater image enhancement solution through a deep residual framework to solve the problems of low contrast, blurred details and color distortion of the original underwater images [26]. Ren et al…”

Single Underwater Image Restoration Based on Adaptive Transmission Fusion

“…Recently, Cao et al [25] presented the use of a convolutional neural network to estimate background light and a multiscale deep network to determine the transmission map. Liu et al [26] proposed an Underwater Resnet (UResnet) based on the very-deep super-resolution reconstruction (VDSR) model for underwater image enhancement.…”