Low-illumination images are usually taken in non-uniform environmental light, such as extremely dark or bright light or artificial light. The enhancement results achieved by existing techniques are prone to halo artifacts, color un-naturalness, and information loss. To address these problems, we present a physical lighting model that describes the degradation of poor illumination images, in which the environmental light is a point-wise variable and changes with the local light source. As long as the parameters in the model are properly estimated, the low-illumination images can be directly recovered by solving the model. First, the initial environmental light can be considered as the incident component according to the Retinex theory and esti-mated via a Gaussian surrounding function. Second, the en-vironmental light and light-scattering attenuation rate are iter-atively adjusted with the information loss constraint. Last, to re-strain the halo and block effects, the two parameters are refined by the weighted guide filter. The experimental results indicate that the proposed algorithm can improve the appearance of low-illumination images that are captured in different scenes, reveal the details in textured regions with few halo effects, in-crease the richness of the visible edges, retain color consistency and reproduce the color quality and naturalness.

Outdoor images captured during sand–dust weather condition typically yield poor contrast and colour shift. A novel method for single sand–dust images restoration is introduced in this paper, which relies on the atmospheric scattering model and information loss constraint. To compensate the colour shift and achieve proper luminance, the proposed atmospheric light is changing with the content of the local scenes, which is initially estimated on the basis of the general scattering model and the grey-world assumption. Then, the initial atmospheric light is updated and the coarse transmission is estimated under the information loss constraint. Next, the fast guide filter is exploited in the post-refinement process to inhibit the halo artefacts. Comparison experiments demonstrate that the proposed algorithm is straightforward and efficient, the contrast and colour shift of different kinds of sand-dust images can be well compensated, especially, nice colour fidelity and proper luminance can be maintained.

Visually sensitive regions in the scene are thought to be important for scene categorization. In this paper, we propose to utilize the important visually sensitive information represented by deep features for scene categorization. Specifically, the context relationship between the objects and the surroundings is fully utilized as the main basis for judging the content of the scene, and combining with the deep convolution neural networks (CNNs), a scene categorization model based on deep visually sensitive features is constructed. First, the saliency regions of the scene images are marked according to the context-based saliency detection algorithm. Then, the original images and the corresponding visually sensitive region detection images are superimposed to obtain the visually sensitive region enhancement images. Second, the deep convolution features of the original images, the visually sensitive region detection images, and the visually sensitive region enhancement images are extracted through the deep CNNs pre-trained on the large-scale scene dataset Places. Finally, considering that the deep features extracted by different layers of the convolution network have different capabilities of discrimination, the fusion features are generated from multiple convolution layers to construct visually sensitive CNN model (VS-CNN). In order to verify the effectiveness of the proposed model, the experiments are conducted on the five standard scene datasets, i.e., LabelMe, UIUC-Sports, Scene-15, MIT67, and SUN. The experimental results show that the proposed model is effective and has good adaptability. Especially, our categorization performance is superior to many state-of the-art methods for a complex indoor scene.

Existing stereo matching networks with high accuracy generally demand high computational overhead, making them expensive to deploy and unsuitable for many real-time applications, especially on source-constrained hardware platforms. Some recent works have tried to build lightweight stereo matching networks, but their accuracy and inference speed is far from satisfactory when deployed on source-constrained GPUs. This paper aims to build a lightweight stereo matching network with high accuracy. First, we claim that a reasonable disparity upsampling strategy is crucial for lightweight stereo matching to achieve high accuracy, whose importance is underestimated in existing works. Accordingly, an efficient and effective disparity upsampling strategy named Spatial Adaptive Disparity Shuffle (SADS) is proposed to meet the upsampling requirements of disparities within diverse regions. Second, a novel Channel-Disparity-Mixed Attention (CDMA) mechanism is proposed to regularize the high-resolution 4D compact cost volume. The attention values in CDMA are adaptive across channels, pixels and disparity candidates, making CDMA suitable for regularizing the compact cost volume. Based on the proposed methods, a lightweight stereo matching network (SADSNet) is designed that achieves real-time performance with high accuracy on source-limited hardware platforms (e.g., the NVIDIA Jetson TX2). In addition, SADSNet can be easily scaled up or down to suit different application scenarios according to available computational power. Extensive experimental results on multiple datasets show its clear superiority over SOTA lightweight stereo models and even many large accuracy-oriented models.