Thermal infrared image super-resolution technology successfully solves the problems of low resolution and blurred texture details in infrared images. However, the problem of background thermal noise and streak interference in thermal infrared images has not been effectively solved. Therefore, in this paper, we analyze and model the generation of background thermal noise and streak interference, and propose a real-world super-resolution algorithm based on generative adversarial network with multi-structure fusion. We first statistically analyze the imaging principle and dataset of the thermal imager to better model the phenomenon of background thermal noise and streak interference present in thermal infrared images. Meanwhile, in order to better recover the details, we use grayed-out visible images to guide the network training and propose a novel generator with multi-structural fusion. In the generator, we design a dynamic dense-attention module that dynamically assigns weights to the attention branch and the densely connected branch to take full advantage of both branches. Compared to other state-of-the-art methods, our proposed method exhibits excellent visual effects, effectively eliminating the effects of noise and streaks while enhancing image texture information.

Fully convolutional networks (FCN) such as UNet and DeepLabv3+ are highly competitive when being applied in the detection of earthquake-damaged buildings in very high-resolution (VHR) remote sensing images. However, existing methods show some drawbacks, including incomplete extraction of different sizes of buildings and inaccurate boundary prediction. It is attributed to a deficiency in the global context-aware and inaccurate correlation mining in the spatial context as well as failure to consider the relative positional relationship between pixels and boundaries. Hence, a detection method for earthquake-damaged buildings based on the object contextual representations (OCR) and boundary enhanced loss (BE loss) was proposed. At first, the OCR module was separately embedded into high-level feature extractions of the two networks DeepLabv3+ and UNet in order to enhance the feature representation; in addition, a novel loss function, that is, BE loss, was designed according to the distance between the pixels and boundaries to force the networks to pay more attention to the learning of the boundary pixels. Finally, two improved networks (including OB-DeepLabv3+ and OB-UNet) were established according to the two strategies. To verify the performance of the proposed method, two benchmark datasets (including YSH and HTI) for detecting earthquake-damaged buildings were constructed according to the post-earthquake images in China and Haiti in 2010, respectively. The experimental results show that both the embedment of the OCR module and application of BE loss contribute to significantly increasing the detection accuracy of earthquake-damaged buildings and the two proposed networks are feasible and effective.

High-resolution remote sensing (HRRS) images, when used for building detection, play a key role in urban planning and other fields. Compared with the deep learning methods, the method based on morphological attribute profiles (MAPs) exhibits good performance in the absence of massive annotated samples. MAPs have been proven to have a strong ability for extracting detailed characterizations of buildings with multiple attributes and scales. So far, a great deal of attention has been paid to this application. Nevertheless, the constraints of rational selection of attribute scales and evidence conflicts between attributes should be overcome, so as to establish reliable unsupervised detection models. To this end, this research proposes a joint optimization and fusion building detection method for MAPs. In the pre-processing step, the set of candidate building objects are extracted by image segmentation and a set of discriminant rules. Second, the differential profiles of MAPs are screened by using a genetic algorithm and a cross-probability adaptive selection strategy is proposed; on this basis, an unsupervised decision fusion framework is established by constructing a novel statistics-space building index (SSBI). Finally, the automated detection of buildings is realized. We show that the proposed method is significantly better than the state-of-the-art methods on HRRS images with different groups of different regions and different sensors, and overall accuracy (OA) of our proposed method is more than 91.9%.

Nucleus segmentation plays a crucial role in tissue pathology image analysis. Despite significant progress in cell nucleus image segmentation algorithms based on fully supervised learning, the large number and small size of cell nuclei pose a considerable challenge in terms of the substantial workload required for label annotation. This difficulty in acquiring datasets has become exceptionally challenging. This paper proposes a novel weakly supervised nucleus segmentation method that only requires point annotations of the nuclei. The technique is an encoder–decoder network which enhances the weakly supervised nucleus segmentation performance (EnNuSegNet). Firstly, we introduce the Feature Preservation Module (FPM) in both encoder and decoder, which preserves more low-level features from the shallow layers of the network during the early stages of training while enhancing the network’s expressive capability. Secondly, we incorporate a Scale-Aware Module (SAM) in the bottleneck part of the network to improve the model’s perception of cell nuclei at different scales. Lastly, we propose a training strategy for nucleus edge regression (NER), which guides the model to optimize the segmented edges during training, effectively compensating for the loss of nucleus edge information and achieving higher-quality nucleus segmentation. Experimental results on two publicly available datasets demonstrate that our proposed method outperforms state-of-the-art approaches, with improvements of 2.02%, 1.41%, and 1.59% in terms of F1 score, Dice coefficient, and Average Jaccard Index (AJI), respectively. This indicates the effectiveness of our method in improving segmentation performance.

In recent, knowledge distillation approaches in semantic segmentation have approved high performance over many datasets. However they can't effectively utilize edge information and global semantic information across various images that are significant for segmentation. This paper proposes a novel Cross-Image Boundary Privileged Knowledge Distillation, which can abstract edge information from whole dataset images and transfer structured pixel-to-edge knowledge to the student network. Specifically, we employ multi loss functions to optimize, the significant two parts are pixel-to-region and pixel-to-edge loss, which can improve the segmentation performance through experimental results over CamVid datasets.

Object detection, a fundamental research direction in the field of computer vision, serves as the basis for various complex visual tasks. However, detection performance for small targets remains significantly inferior compared to regular-sized targets. In conventional tiny object detection methods, the uniform distribution strategy adopted for bounding box regression during prediction processes tends to induce instability, particularly when handling small targets. To address this critical issue, this paper proposes Row-Column Decoupled Loss to enhance the stability of the prediction process and improve the accuracy of detection results. The proposed approach innovatively transforms the traditional regression paradigm by incorporating probabilistic modeling that better aligns with the spatial characteristics of small objects. Firstly, to address the challenge where the absence of overlap between predicted and ground-truth bounding boxes prevents distance computation in small object scenarios, this work introduces a probabilistic modeling framework based on Binomial distribution to characterize the spatial distribution of bounding boxes. This approach effectively captures the probabilistic attributes of pixel distributions within target regions. Secondly, we design a dual-channel evaluation mechanism, established a similarity evaluation strategy that better aligns with the target natural characteristics. Finally, Row-Column Decoupled Loss: A probability-Based Geometric Similarity Framework is proposed to optimize feature learning for minuscule targets. Experimental validation on the AI-TOD dataset demonstrates the efficacy of our method across multiple mainstream object detection frameworks and state-of-the-art small object detectors. Detectors equipped with Row-Column Decoupled Loss exhibit significant performance enhancements over baseline models, achieving a peak value of 60.5 in $AP_{0.5} $ and delivering a maximum improvement margin of 16.2 in $AP_{0.5} $ compared to baselines.