Underwater images suffer from color casts and low illumination due to the scattering and absorption of light as it propagates in water. These problems can interfere with underwater vision tasks, such as recognition and detection. We propose an adaptive learning attention network for underwater image enhancement based on supervised learning, named LANet, to solve these degradation issues. First, a multiscale fusion module is proposed to combine different spatial information. Second, we design a novel parallel attention module(PAM) to focus on the illuminated features and more significant color information coupled with the pixel and channel attention. Then, an adaptive learning module(ALM) can retain the shallow information and adaptively learn important feature information. Further, we utilize a multinomial loss function that is formed by mean absolute error and perceptual loss. Finally, we introduce an asynchronous training mode to promote the network’s performance of multinomial loss function. Qualitative analysis and quantitative evaluations show the excellent performance of our method on different underwater datasets. The code is available at: https://github.com/LiuShiBen/LANet .

Palm vein recognition is a high-security biometric. Outside the NIR capture box and contactless palm vein recognition are more popular but challenging. The users feel comfortable outside the NIR capture box but face more optical blurring brought by visible light. Contactless capture gestures solve the hygienic problem but face the image rotation, position translation, and scale transformation which makes classification difficult especially in large-scale databases. To address these problems, we develop a wavelet denoising ResNet, which consists of two models: the wavelet denoising (WD) model and the squeeze-and-excitation ResNet18 (SER) model. The WD model focuses on removing noise from skin scattering and optical blurring from palm vein images. The WD model enhances the low-frequency feature into a deep learning feature by residual learning technology. This strategy increases the weight of an effective handcrafted feature in the deep learning network. The SER model overcomes rotation, position translation, and scale transformation by selectively emphasizing classification features and weakening less useful features. To train and verify the network, an inside box palm vein image database and an outside box palm vein image database are set up. The Tongji contactless palm vein image database was also employed in the experiments. The validity and superiority of our network are verified in a series of experiments.

Since light is scattered and absorbed by water, underwater images have inherent degradation (e.g. hazing, color shift), consequently impeding the development of remotely operated vehicles. Towards this end, we propose a novel method, referred to as Best o f Bo th Worlds (Boths). With parameters of only 0.0064M, Boths can be considered a super lightweight neural network for underwater image enhancement. On the whole, it has three levels: Structure and Detail Features; Pixel and Channel Dimensions; High and Low Frequency Information. Each of these three levels represents ”Best of Both Worlds”. Initially, by interacting with structure and detail features, Boths can focus on these two aspects at the same time. Further, our network can simultaneously consider channel and pixel dimensions through 3D attention learning, which is more similar to human visual perception. Lastly, the proposed model can focus on high and low frequency information, through a novel loss function based on the wavelet transforms. Upon subsequent analysis and evaluation, Boths has shown superior performance compared with state-of-the-art methods. Our models and datasets are publicly available at: https://github.com/perseveranceLX/Boths.

With the growing demand for information security, finger-vein recognition has become widespread. However, the robustness of the recognition process becomes a major problem. When identifying unseen categories with traditional finger-vein recognition systems, a few issues remain, such as recognition interference and low efficiency. This paper proposes a Deep Generalized Label Finger-Vein (DGLFV) model to extract feature maps and achieve high-accuracy recognition. The largest rectangular finger-vein region is extracted through image semantic segmentation and the advanced bidirectional traversing and center diffusion method for the known categories. Then we generalize all the unseen categories actively as Class $C+1$ to reduce interference from unregistered users. Furthermore, an adaptive threshold acquisition algorithm is proposed for Label Receiver Operating Characteristic (LROC), so that the procedures of classification, recognition, and verification are unified. Apart from Shandong University Homologous Multi-modal Traits (SDUMLA-HMT), we have conducted additional experiments on our self-built database, Finger Veins of Signal and Information Processing Laboratory (FV-SIPL). The recognition accuracy of the approach proposed in this paper has reached 99.25% and 99.08% testing on FV-SIPL and SDUMLA-HMT, with a low error rate at 1.481% and 2.228% and little time consumption of 0.157s for a single image, which is better than most state-of-the-art finger-vein recognition methods.