检索结果-内蒙古大学图书馆

IEEE International Symposium on Geoscience and Remote Sensing (IGARSS)

作者： Shunli Tian Xiangrong Zhang Guanchun Wang Xiao Han Puhua Chen Xina Cheng Key Laboratory of Intelligent Perception and Image Understanding of Ministry of Education Xidian University Xi’an China

Hyperspectral image change detection (HSI-CD) can accurately identify changing regions by capturing subtle spectral differences and has become a research hotspot in the field of remote sensing (RS). Convolutional neural networks (CNNs) have excellent local context modeling capabilities and have been proven to be powerful feature extractors in HSI-CD. However, due to its inherent network structure limitation, CNN cannot well mine and represent the sequential properties of spectral features, especially the medium and long-term dependencies. In contrast, transformer-based network architecture shows a strong ability to model long-distance dependencies, which can fully mine and extract global features, but exhibits weak performance in extracting local information. To this end, we propose HSI-CD network based on adaptive contrastive learning (CTACL). Specifically, we first propose a parallel network of CNNs and transformers to mine local and global temporal-spatial-spectral features of HSI, respectively. Second, we propose adaptive contrastive learning to pre-train the network to learn the latent features of a large amount of unlabeled data and better mine and utilize local and global information. Experimental results on the farmland dataset show that the proposed method performs well.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Domain Adversarial Debiased Self-Training for Hyperspectral image Classification

Domain Adversarial Debiased Self-Training for Hyperspectral ...

引用

IEEE International Symposium on Geoscience and Remote Sensing (IGARSS)

作者： Tianshu Zhang Jie Feng Ziyu Zhou Xiangrong Zhang Licheng Jiao Key Laboratory of Intelligent Perception and Image Understanding of Ministry of Education Xidian University Xi’an China

Unsupervised domain adaptation (UDA) has been widely used in hyperspectral image (HSI) classification. Domain adversarial learning methods and self-training methods are two major UDA methods. Most existing methods use these two methods independently, which limit their capacity to knowledge transferability and robustness of classification. Therefore, a novel domain adversarial debiased self-training (DADST) is proposed to combine these two methods, where domain adversarial learning reduces the domain discrepancy and debiased self-training updates model via a self-paced curriculum policy. To this end, we first apply debiased self-training into HSI classification. Then domain adversarial learning module is combined to align joint distribution between the source domain and the target domain. Experimental results on two cross-scene HSI datasets demonstrate that the proposed DADST method outperforms other domain adaptation approaches.

关键词：

来源：评论

学校读者我要写书评

暂无评论

Global-Local Representation Coupling Network for Remote Sensing image Change Detection

Global-Local Representation Coupling Network for Remote Sens...

引用

IEEE International Symposium on Geoscience and Remote Sensing (IGARSS)

作者： Fanghan Yang Xiangrong Zhang Peng Zhu Zhenhang Weng Puhua Chen Key Laboratory of Intelligent Perception and Image Understanding of Ministry of Education Xidian University Xi’an China

Change detection is one of the important tasks in remote sensing image processing, and the powerful feature extraction ability of convolutional networks has achieved some success in change detection. However, the problem of the limited field size of pure convolutional networks makes the change detection accuracy of high-resolution remote sensing images limited. The introduction of transformers can link the concept of long-distance in space and time. Therefore, in order to maximize the respective advantages of transformers and CNNs, we propose a new parallel architecture. To accomplish the above goal, we propose a new network, which consists of a local detail branch and transformer global spatial-temporal feature branch and a feature fusion module. The experimental results reach the current sota level.

关键词：

来源：评论

学校读者我要写书评

暂无评论

CSCT: Channel-Spatial Coherent Transformer for Remote Sensing image Super-Resolution

引用

IEEE Transactions on Geoscience and Remote Sensing 2025年 63卷

作者： Zhang, Kexin Li, Lingling Jiao, Licheng Liu, Xu Ma, Wenping Liu, Fang Yang, Shuyuan Xidian University Key Laboratory of Intelligent Perception and Image Understanding Ministry of Education of China International Research Center of Intelligent Perception and Computation School of Artificial Intelligence Xi'an710071 China

Remote sensing image super-resolution (RSISR) techniques are crucial in practice as an economical approach to enhancing the resolution of remote sensing images (RSIs). The scale of structural information and the richness of texture details in RSIs far exceed those in natural images. Therefore, accurately restoring and preserving edge and detail information are a critical challenge in the super-resolution (SR) process. Currently, convolutional neural network (CNN)-based methods primarily rely on local feature extraction, which fails to effectively capture and integrate global contextual information. Generative adversarial network (GAN)-based methods, while improving the visual quality, often suffer from artifacts and training instability, adversely affecting image quality. Moreover, these approaches struggle to accurately represent high-frequency features, leading to blurriness or distortion when reconstructing fine details and edges. To address these limitations, we introduce the channel-spatial coherent transformer (CSCT). The core of CSCT includes the channel-spatial coherent attention (CSCA) and the frequency-gated feed-forward network (FGFN), which work synergistically to enhance edge and detail preservation while significantly improving overall image clarity. CSCA efficiently aggregates channel and spatial information, while FGFN adaptively adjusts frequency information to enhance high-frequency details and suppress low-frequency noise. Moreover, this article leverages advanced data augmentation methods that markedly boost RSISR performance, offering new avenues for further exploration. The empirical analysis across several remote sensing SR benchmark datasets reveals that our approach excels in detail restoration, effectively reduces artifacts and noise, and significantly enhances the quality of SR images. © 2025 IEEE.

关键词： Remote sensing

来源：评论

学校读者我要写书评

暂无评论

Graph-based Adaptive Network With Spatial-Spectral Features for Hyperspectral Unmixing

引用

IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing 2025年 18卷 12865-12881页

作者： Dong, Hua Zhang, Xiaohua Zhang, Jinhua Meng, Hongyun Jiao, Licheng Xidian University Key Laboratory of Intelligent Perception and Image Understanding of Ministry of Education School of Artificial Intelligence Xi'an710126 China Xidian University School of Mathematics and Statistics Xi'an710126 China

Hyperspectral unmixing aims to extract basic material (endmember) spectra and estimate their corresponding fractions (abundances) from observed pixels in hyperspectral images (HSIs). Recently, blind unmixing methods based on Autoencoders (AEs) have gained significant attention due to their capability to simultaneously obtain endmembers and abundances, as well as their strong performance. Also, many AE-based unmixing methods have explored the use of spatial information in HSIs to enhance unmixing performance. However, most of these methods fail to effectively integrate both local and global spatial-spectral features and lack adaptive selection and constraints for the extracted features. To address this issue, this paper proposes an adaptive spatial-spectral unmixing method based on Graph. In the method, HSIs are treated as data on manifold structures, with superpixels serving as graph nodes to construct a global graph-structured data. We then use Graph Attention Network (GAT) to learn superpixel-level global spatial-spectral feature representations from this graph-structured data. Since the spatial-spectral feature learned by GAT is at the superpixel level, it is impossible to unmix each pixel using only this feature. Thus, we integrate a Convolutional Neural Network (CNN) to learn local discriminative spatial-spectral features. Attention mechanisms are further employed to select the features most relevant to unmixing. Additionally, An Adaptive Abundance Constraint Module is proposed to dynamically enforce constraints on the extracted features. Experiments conducted on synthetic and real hyperspectral images demonstrate the efficiency and superior performance of our proposed method compared to several existing unmixing approaches. © 2008-2012 IEEE.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

SSRN

引用

SSRN 2025年

作者： He, Jianghai Shang, Ronghua Wu, Ting Wang, Chi Li, Yangyang Key Laboratory of Intelligent Perception and Image Understanding of Ministry of Education School of Artificial Intelligence Xidian University Shaanxi Province Xi’an710071 China

In multi-person scenes, localizing character joints is a challenging task due to occlusion and complex human interactions. Especially in predicting invisible joints, deep networks cannot extract valid information from occluded inputs. Besides, Transformer-based algorithms with the highest accuracy are sensitive to image size, leading to rapidly increased computational costs with higher-resolution images. Moreover, existing algorithms often lack an evaluation feedback mechanism, resulting in unreliable outputs. To tackle these issues, this paper proposes the Strongly Correlated Nodes and Confidence Feedbacks based CNN and Transformer combined multi-person pose estimation (SCNCF). Firstly, to address the Transformer's sensitivity to input size, SCNCF constructs a CNN-based feature extraction network to screen valid inputs. It filters out a few image blocks containing specific joints as Transformer’s input, significantly reducing the computational load. Secondly, to better predict invisible joints, SCNCF learns node associations from the large pre-training dataset. By aligning strongly correlated nodes according to the similarity matrix, a robust pose model is established. The positions of the occluded joints can be directly derived from the model, thus minimizing accuracy loss from inaccurate poses. Lastly, to make up for the missing validation feedback mechanism, SCNCF introduces a confidence scoring method. The candidate's joints are initially scored according to their probabilities, and the scores decline if their probabilities are close to neighbors. Joints with excessively low scores are fed back into the network and treated as invisible. When reprocessing them, only strongly correlated nodes with sufficient scores are used. This avoids introducing excessive precision errors. Iteratively applying the procedure ensures that the network avoids generating low-quality solutions. On COCO-val 2017, CrowdPose, and OCHuman-val dataset, SCNCF shows a significant improveme

关键词： Iterative methods

来源：评论

学校读者我要写书评

暂无评论

Logits DeConfusion with CLIP for Few-Shot Learning

arXiv

引用

arXiv 2025年

作者： Li, Shuo Liu, Fang Hao, Zehua Wang, Xinyi Li, Lingling Liu, Xu Chen, Puhua Ma, Wenping Key Laboratory of Intelligent Perception and Image Understanding of Ministry of Education International Research Center for Intelligent Perception and Computation Joint International Research Laboratory of Intelligent Perception and Computation School of Artificial Intelligence Xidian University Xi’an710071 China

With its powerful visual-language alignment capability, CLIP performs well in zero-shot and few-shot learning tasks. However, we found in experiments that CLIP’s logits suffer from serious inter-class confusion problems in downstream tasks, and the ambiguity between categories seriously affects the accuracy. To address this challenge, we propose a novel method called Logits DeConfusion, which effectively learns and eliminates inter-class confusion in logits by combining our Multi-level Adapter Fusion (MAF) module with our Inter-Class Deconfusion (ICD) module. Our MAF extracts features from different levels and fuses them uniformly to enhance feature representation. Our ICD learnably eliminates inter-class confusion in logits with a residual structure. Experimental results show that our method can significantly improve the classification performance and alleviate the inter-class confusion problem. The code is available at https://***/LiShuo1001/LDC. © 2025, CC BY.

关键词： Zero-shot learning

来源：评论

学校读者我要写书评

暂无评论

Renormalized Connection for Scale-preferred Object Detection in Satellite imagery

arXiv

引用

arXiv 2024年

作者： Zhang, Fan Li, Lingling Jiao, Licheng Liu, Xu Liu, Fang Yang, Shuyuan Hou, Biao Key Laboratory of Intelligent Perception and Image Understanding of Ministry of Education International Research Center for Intelligent Perception and Computation School of Artificial Intelligence Xidian University Xi’an710071 China

Satellite imagery, due to its long-range imaging, brings with it a variety of scale-preferred tasks, such as the detection of tiny/small objects, making the precise localization and detection of small objects of interest a challenging task. In this article, we design a Knowledge Discovery Network (KDN) to implement the renormalization group theory in terms of efficient feature extraction. Renormalized connection (RC) on the KDN enables "synergistic focusing" of multi-scale features. Based on our observations of KDN, we abstract a class of renormalized connections with different connection strengths, called n21C, and generalize it to FPN-based multi-branch detectors. In a series of FPN experiments on the scale-preferred tasks, we found that the "divide-and-conquer" idea of FPN severely hampers the detector's learning in the right direction due to the large number of large-scale negative samples and interference from background noise. Moreover, these negative samples cannot be eliminated by the focal loss function. The Renormalized Connections extends the multi-level feature's "divide-and-conquer" mechanism of the FPN-based detectors to a wide range of scale-preferred tasks, and enables synergistic effects of multi-level features on the specific learning goal. In addition, interference activations in two aspects are greatly reduced and the detector learns in a more correct direction. Extensive experiments of 17 well-designed detection architectures embedded with n21Cs on five different levels of scale-preferred tasks validate the effectiveness and efficiency of the Renormalized Connections. Especially the simplest linear form of RC - E421C performs well in all tasks and it satisfies the scaling property of renormalization group theory. All experiments can be trained and tested on a graphics card with 8GB of video memory, which greatly enhances the applicability of our methodology. We hope that our approach will transfer a large number of well-designed detectors from t

关键词： Satellite imagery

来源：评论

学校读者我要写书评

暂无评论

Multi-Scale Co-Attention Learning For SAR image Change Detection

Multi-Scale Co-Attention Learning For SAR Image Change Detec...

引用

IEEE International Symposium on Geoscience and Remote Sensing (IGARSS)

作者： Dan Zhang Xu Liu Licheng Jiao Fang Liu Joint International Research Laboratory of Intelligent Perception and Computation School of Artificial Intelligence Key Laboratory of Intelligent Perception and Image Understanding of Ministry of Education International Research Center for Intelligent Perception and Computation Xidian University Xi’an China

ISBN: (数字)9798350360325

ISBN: (纸本)9798350360332

Synthetic aperture radar (SAR)iamge change detection is an important and challenging task. Existing methods mainly extract features from the difference map and original images in a parallel structure, ignoring the correlations between features. In this paper, we propose a novel SAR image change detection method based onmulti-scale and co-attention learning, called the Multi-scale Co-attention Network (MCNet). Specifically, we extract features from the difference map and original images using co-attention mechanism, which establishes a relationship between the two kinds of features in the feature space while retaining key information from the original images. With this attention mechanism, the model pays more attention to the changed areas in SAR images, and the features contain richer information. Furthermore, we propose a multi-scale feature fusion method to combine high-level and low-level semantic information, improving the generality and robustness of the features. Finally, the performance of the proposed method has been validated for its effectiveness.

关键词： Correlation Attention mechanisms Semantics Geoscience and remote sensing Feature extraction Radar polarimetry Robustness

来源：评论

学校读者我要写书评

暂无评论

Combination Learning for Few-Shot Segmentation

SSRN

引用

SSRN 2024年

作者： Li, Shuo Liu, Fang Jiao, Licheng Li, Lingling Liu, Xu Chen, Puhua Key Laboratory of Intelligent Perception and Image Understanding of Ministry of Education China International Research Center for Intelligent Perception and Computation China Joint International Research Laboratory of Intelligent Perception and Computation School of Artificial Intelligent Xidian University China

Few-Shot Segmentation aims to segment objects with the same semantics in a query image based on a few annotated support images. Due to large intra-class differences, only a few annotated images on novel classes make it difficult for existing methods to obtain accurate boundaries and complete objects. With the development of general-purpose segmentation models, all objects in an image can be segmented into masks that usually have accurate boundaries. However, these masks have no semantics and most of them are over-segmented. In this work, we attempt to learn how to combine these over-segmented masks to predict results. To mine the masks generated by general-purpose models, we propose a novel Combination Learning (ComLearn) method for FSS. We first design a mask generation module to obtain over-segmented masks and their features. Then, we propose a score generation module to calculate the score that each query mask belongs to the target semantics. Furthermore, we design a mask quality module to calculate the score of each mask, thus adding a score loss function in addition to the prediction loss. Experiments show that our ComLearn has a strong ability to generalize from base classes to novel classes. © 2024, The Authors. All rights reserved.

关键词： Semantics

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：