Due to light scatter and absorption in waterbody, underwater imaging can be easily impaired with low contrast visual distortion. The resulting images are often unable meet the quality requirements of human perception computer processing. Therefore, Underwater Image Enhancement (UIE) has been attracting extensive research efforts. Although deep learning demonstrated its great success many vision tasks, huge amounts parameters computations not conducive UIE resource-limited scenarios. In this...

Underwater imaging is usually degraded by low contrast or color cast due to light absorption and scattering. This fact limits the accuracy of object recognition in underwater marine environments. In this article, address issue, we propose a CURE-Net that progressively improves images coarse-to-fine way. To be specific, our consists three cascaded subnetworks. The first two subnetworks use attention gate fusion learn multiscale contextual information, whereas third subnetwork preserves fine...

Images collected in low-light environments usually suffer from multiple, non-uniform distributed distortions, including local dark, dim light, backlit and so on. In this paper, we propose a Stage-Transformer-Guided Network (STGNet) that effectively handles region-specific distributions enhance diverse images. Specifically, our STGNet adopts multi-stage way to progressively learn hierarchical features benefit the robustness of model. At each stage, design an efficient transformer with...

Snow removal aims to locate snow areas and recover clean images without repairing traces. Unlike the regularity semitransparency of rain, with various patterns degradations seriously occludes background. As a result, state-of-the-art methods usually retains large parameter size. In this paper, we propose lightweight but high-efficient network called Laplace Mask Query Transformer (LMQFormer). Firstly, present Laplace-VQVAE generate coarse mask as prior knowledge snow. Instead using in...

Recently, many compression algorithms are applied to decrease the cost of video storage and transmission. This will introduce undesirable artifacts, which severely degrade visual quality. Therefore, Video Compression Artifacts Removal (VCAR) aims at reconstructing a high-quality from its corrupted version compression. Generally, this task is considered as vision-related instead media-related problem. In research, quality has been significantly improved while computational complexity bitrate...

Single-image desnowing aims at depressing snowflake noises while preserving a clean background. Existing methods usually mask the locations of and remove them in RGB color space. In this paper, we rethink problem by investigating impacts space selection. Theoretical analysis experiments reveal that feature exhibit different distributions spaces. particular, these are barely seen Hue channel, which inspires us to recover global structure texture information background from channel. More...

The limited bandwidth of underwater acoustic channels poses a challenge to the efficiency multimedia information transmission. To improve efficiency, system aims transmit less data while maintaining image utility at receiving end. Although assessing within compressed is essential, current methods exhibit limitations in addressing utility-driven quality assessment. Therefore, this paper introduces Distillation-based Compacted Information Quality assessment metric (DCIQ) for utility-oriented...

As the expectation for higher quality of life increases, consumers have demands food. Food authentication is technical means ensuring food what it says is. A popular approach to based on spectroscopy, which has been widely used identifying and quantifying chemical components an object. This non-destructive effective but expensive. paper presents a computer vision-based sensor system authentication, i.e., differentiating organic from non-organic apples. consists low-cost hardware pattern...

Low-light imaging task aims to approximate low-light scenes as perceived by human eyes. Existing methods usually pursue higher brightness, resulting in unrealistic exposure. Inspired Human Vision System (HVS), where rods perceive more lights while cones colors, we propose a Degradation Rectify Model (LDRM) with color-monochrome cameras solve this problem. First, use low-ISO color camera and high-ISO monochrome for under short-exposure of less than 0.1s. Short-exposure could avoid motion...

Single image desnowing is an important and challenge task for lots of computer vision applications, such as visual tracking video surveillance. Although existing deep learning-based methods have achieved promising results, most them rely on the local features neglect global relationship information between regions. Therefore, inevitably leading to over-smooth or detail loss results. To solve this issue, we design a UNet-based end-to-end architecture desnowing. Specially, better characterize...

This paper presents a low-cost sensor system for food authentication based on computer vision and pattern recognition. The uses smartphone to record video of sample under gradient coloured illumination transforms the into data vector analysis. is evaluated tasks detecting olive oil adulteration milk fat content, achieving 100% test accuracies. Since built in without any additional hardware, it has potential serve as simple effective solution authentication.

Rain streaks usually result in severe image visual degradation and foreground occlusion, affecting the quality of computer tasks outdoor scenes. Currently, mainstream methods single-image deraining are based on data-driven. However, deep learning network could be imperfect, with limited power for global information from rain all over map. In order to solve this problem, we proposed a novel Hierarchical Distillation Network (HD-Net). network, Feature Extraction Block (HFEB) can fully utilize...