Underwater images captured by optical cameras can be degraded by light attenuation and scattering, which leads to deteriorated visual image quality. The technique of underwater image enhancement plays an important role in a wide range of subsequent applications such as image segmentation and object detection. To address this issue, we propose an underwater image enhancement framework which consists of an adaptive color restoration module and a haze-line based dehazing module. First, we employ an adaptive color restoration method to compensate the deteriorated color channels and restore the colors. The color restoration module consists of three steps: background light estimation, color recognition, and color compensation. The background light estimation determines the image is blueish or greenish, and the compensation is applied in red-green or red-blue channels. Second, the haze-line technique is employed to remove the haze and enhance the image details. Experimental results show that the proposed method can restore the color and remove the haze at the same time, and it also outperforms several state-of-the-art methods on three publicly available datasets. Moreover, experiments on an underwater object detection dataset show that the proposed underwater image enhancement method is able to improve the accuracy of the subsequent underwater object detection framework.
Absorption and scattering in aqueous media would attenuate light and make imaging difficult. Therefore, an artificial light source is usually utilized to assist imaging in the deep ocean. However, the artificial light source typically alters the light conditions to a large extent, resulting in the non-uniform illumination of images. To solve this problem, we propose a non-uniform illumination correction algorithm based on a fully convolutional network for underwater images. The proposed algorithm model the original image as the addition of the ideal image and a non-uniform light layer. We replace the traditional pooling layer with dilated convolution to expand the receptive field and achieve higher accuracy in non-uniform illumination recognition. To improve the perception ability of the network effectively, the original image and parameters which pre-trained on the ImageNet are concentrated. The concentrated information is used as input to the network. Due to the color shift and blurred details of the underwater image, we design the novel loss function, which includes three parts of feature loss, smooth loss, and adversarial loss. Moreover, we built a dataset of the underwater image with non-uniform illumination. Experiments show that our method performs better in subjective assessment and objective assessment than some traditional methods.
The aim of multi-focus image fusion technology is to acquire an image of every scene all focused on the same visual point at different focal settings. To achieve this goal, we propose an improved multi-focus image fusion algorithm based on a Gaussian curvature filter and synthetic focusing degree criterion. First, in order to realize the salient feature extraction, a Gaussian curvature filter is applied to obtain the most sharpness regions. Then we obtain a coarse fusion map by composing a synthetic focusing degree criterion, which is a combination of the spatial frequency and the local variance of image. The coarse fusion map is further processed by morphological filters and median filters to acquire an optimized fusion map. Finally, the fusion image is obtained via a weighted fusion operation. Experimental results demonstrate that our proposed algorithm can be competitive with, or even outperform, many existing fusion methods on both qualitative and quantitative measures.
Imaging through the wavy air–water surface suffers from severe geometric distortions, which are caused by the light refraction effect that affects the normal operations of underwater exploration equipment such as the autonomous underwater vehicle (AUV). In this paper, we propose a deep learning-based framework, namely the self-attention generative adversarial network (SAGAN), to remove the geometric distortions and restore the distorted image captured through the water–air surface. First, a K-means-based image pre-selection method is employed to acquire a less distorted image that preserves much useful information from an image sequence. Second, an improved generative adversarial network (GAN) is trained to translate the distorted image into the non-distorted image. During this process, the attention mechanism and the weighted training objective are adopted in our GAN framework to get the high-quality restored results of distorted underwater images. The network is able to restore the colors and fine details in the distorted images by combining the three objective losses, i.e., the content loss, the adversarial loss, and the perceptual loss. Experimental results show that our proposed method outperforms other state-of-the-art methods on the validation set and our sea trial set.