检索结果-内蒙古大学图书馆

Gradient-Aligned convolution neural network

pattern recognition 2022年 122卷 108354-108354页

作者： Hao, You Hu, Ping Li, Shirui Udupa, Jayaram K. Tong, Yubing Li, Hua Chinese Acad Sci Inst Comp Technol Key Lab Intelligent Informat Proc Beijing 100190 Peoples R China Univ Penn Dept Radiol Med Image Proc Grp Philadelphia PA 19104 USA Microsoft Inc Beijing 100190 Peoples R China Baidu Inc Beijing 100085 Peoples R China Univ Chinese Acad Sci Beijing 100049 Peoples R China

Although Convolution Neural Networks (CNN) have achieved great success in many applications of computer vision in recent years, rotation invariance is still a difficult problem for CNN. Especially for some images, the content can appear in the image at any angle of rotation, such as medical images, microscopic images, remote sensing images and astronomical images. In this paper, we propose a novel convolution operation, called Gradient-Aligned Convolution (GAConv), which can help CNN achieve rotation invariance by replacing vanilla convolutions in CNN. GAConv is implemented with a prior pixel-level gradient alignment operation before regular convolution. With GAConv, Gradient-Aligned CNN (GACNN) can achieve rotation invariance without any data augmentation, feature-map augmentation, and filter enrichment. In GACNN, rotation invariance does not learn from the training set, but bases on the network model. Different from the vanilla CNN, GACNN will output invariant results for all rotated versions of an object, no matter whether the network is trained or not. This means that we only need to train the network with one canonical version of the object and all other rotated versions of this object should be recognized with the same accuracy. Classification experiments have been conducted to evaluate GACNN compared with some rotation invariant approaches. GACNN achieved the best results on the 360 degrees rotated test set of MNIST-rotation, Plankton-sub-rotation, and Galaxy Zoo 2. (C) 2021 Elsevier Ltd. All rights reserved.

关键词： Gradient alignment Rotation equivariant convolution Rotation invariant neural network

来源：评论

学校读者我要写书评

暂无评论

Quantifying Mangrove Dynamics using Sentinel-1 Multitemporal Data and Random Forest Algorithm 8

Quantifying Mangrove Dynamics using Sentinel-1 Multitemporal...

引用

8th Asia-Pacific Conference on Synthetic Aperture Radar, APSAR 2023

作者： Simarmata, Nirmawana Wikantika, Ketut Darmawan, Soni Harto, Agung Budi Nadzir, Zulfikar Adlan Sari, Dewi Nawang Geodesy and Geomatics Engineering Center for Remote Sensing Institut Teknologi Bandung Institut Teknologi Sumatera Bandung Indonesia Geodesy and Geomatics Engineering Center for Remote Sensing Institut Teknologi Bandung Bandung Indonesia Institute of Geodesy and Geoinformation University of Bonn Geomatics Engineering Institut Teknologi Sumatera Lampung Indonesia Geomatics Engineering Institut Teknologi Sumatera Lampung Indonesia

ISBN: (纸本)9798350393590

Mangroves are significant coastal ecosystems that are susceptible to changes in the environment. In this study, the distribution and size of mangrove expansions were predicted using multitemporal Sentinel-1 synthetic aperture radar (SAR) data and the random forest (RF) method. The data used are multitemporal Sentinel-1 SAR images of 2020, 2021, and 2022 on the coast of South Lampung Regency. We performed pre-processing on these data, including filtration and radiometric correction, to ensure good quality before use in the analysis. SAR features extracted from the multitemporal imagery provide valuable information for identifying and mapping mangroves. The results of Sentinel image extraction explain that vertical-vertical (VV) polarization in 2020 has the lowest value in July with a value of -19.62dB and the highest value in December with a value of -16.48dB;in 2021, the highest value is in November with a value of -16.37dB and the lowest in August with a value of - 19.01dB. In 2022, the lowest value is in February with a value of - 19.87dB and the highest is in May with a value of -16.44dB. Backscatter values in vertical-horizontal (VH) polarization have a relatively similar pattern. These backscatter values are input parameters for RF classification. The classification results are divided into Built-up Area, Road, Water body, sea, mangrove, vegetation, and agriculture classes. The highest accuracy evaluation was obtained in 2021 with a kappa value of 0.93 and an overall accuracy value of 93.3%. The usage of multitemporal Sentinel-1 SAR data and algorithms is significantly aided by this work. ©2023 IEEE.

关键词： Synthetic aperture radar

来源：评论

学校读者我要写书评

暂无评论

End-to-End Network for Early Crop Planting Area Prediction

End-to-End Network for Early Crop Planting Area Prediction

引用

pattern recognition and Artificial Intelligence (PRAI), International Conference on

作者： Kedi Lu Zhong Ma Zhao He Pengcheng Huo Haochen Zhang Xi'an Microelectronics Technology Institute Xi'an China

ISBN: (数字)9798350350890

ISBN: (纸本)9798350350906

Early prediction of crop planting areas is essential for agricultural planning and management. Traditional two-stage calculation methods face limitations due to remote sensing imagery classification and are prone to significant area predicting errors due to ‘error accumulation.’ This paper introduces an innovative end-to-end network designed for accurate, large-scale early-stage crop area prediction. Leveraging multi-subimages processing and ‘trusted pixel’ technology, our approach overcomes the technical challenges of conventional methods, by extracting features from multi-year land cover data, our model directly predicts crop planting areas. Experiments on estimating corn and soybean areas in the USA show a relative error below 4%. Our research lays a foundation for early-stage crop monitoring, supports sustainable agricultural practices, and provides actionable insights for agricultural decision-making and resource allocation.

关键词： Smoothing methods Crops Land surface Predictive models Feature extraction Planning Resource management remote sensing Monitoring Meteorology

来源：评论

学校读者我要写书评

暂无评论

Video Fog Detection Based on Dynamic Texture Analysis

Video Fog Detection Based on Dynamic Texture Analysis

引用

2023 IEEE International Symposium on Broadband Multimedia Systems and Broadcasting, BMSB 2023

作者： Wang, Fei Wang, Manyu Li, Han Yang, Zongkai Song, Youbin Chen, Zhihua Beihang University School of Eletronics and Information Engineering Beijing China Beijing Institute of Remote Sensing Equipment Beijing China Chang'an University School of Information Engineering Xi'an China Scinence and Technology on Special System Simulation Laboratory Beijing China Southwest Technology and Engineering Research Institute Chongqing China

ISBN: (纸本)9798350321524

The accurate recognition of outdoor weather has a very important application value in weather prediction, disaster warning, automatic driving and other fields. Fog, rain, snow and other bad weather pose a serious threat to driving safety, which is the focus of outdoor weather recognition. At present, video surveillance system has been widely used in highway surveillance system, and fog detection based on video image has received extensive attention. This paper will study fog detection technology based on dynamic texture *** paper uses MATLAB as the simulation platform to realize the fog detection based on optical flow method. First of all, considering that the fog area in the video image will change in shape and concentration over time, appropriate anti-interference methods including median filtering are selected to complete the preprocessing;Secondly, according to the characteristics of fog, such as diffusion, the method of feature calculation and motion analysis based on optical flow is studied;Finally, the corresponding motion rules and analysis methods are established to detect and recognize the foggy video regions. The smoke video is processed in this paper, and the results show that the fog area can be accurately detected and the detection effect is good. The experimental results show that the processing method in this paper has a good effect, and has a high application value in video fog detection. © 2023 IEEE.

关键词： Textures

来源：评论

学校读者我要写书评

暂无评论

Fine-Grained recognition for remote sensing images Using Feature Fusion and Reconstruction

SSRN

引用

SSRN 2023年

作者： Wang, Guolin Ji, Jian Liu, Lize Xu, Chengwei School of Computer Science and Technology Xidian University Shaanxi Xi’an710071 China

Object detection and fine-grained recognition of remote sensing images are important computer vision tasks, with the goal to identify the spatial position and category of the object through processing and analysis of the image, and further subdividing it into specific subcategories. Although some research has attempted to solve the problem of object detection in remote sensing image fields, fine-grained recognition tasks have been neglected. The main challenges for fine-grained recognition of aerial objects are: (1) complex backgrounds and densely distributed small objects;(2) arbitrary and diverse orientations of objects;(3) the class imbalance of training samples;(4) small object scales and various orientations in remote sensing images, which can lead to misaligned features and insufficient local detailed texture features, making fine-grained recognition of aerial objects more difficult. To address these challenges, we propose an end-to-end two-stage object detection and fine-grained recognition network. The backbone network uses Cascaded Multi-scale Feature Extraction networks(CMFE), and then optimizes fine-grained recognition by aligning features and reconstructing features to improve accuracy. Experimental results on three challenging aerial datasets, DOTA, HRSC2016, and FAIR1M, demonstrate the effectiveness of our proposed method. © 2023, The Authors. All rights reserved.

关键词： Object detection

来源：评论

学校读者我要写书评

暂无评论

Noise in Digital image processing: A Review Study 3

Noise in Digital Image Processing: A Review Study

引用

3rd Information Technology to Enhance e-Learning and Other Application, IT-ELA 2022

作者： Al Mudhafar, Rusul A. El Abbadi, Nidhal K. University of Kufa Faculty of Computer Science and Mathematics Computer Science Department Iraq Faculty of Education Computer Science Department Najaf Iraq

ISBN: (纸本)9798350334968

We are awash with images. The quality of an image is greatly impacted by image noise, making it difficult to extract, analyze, recognize, and make quantitative measurements of features. During the image capture, coding, transmission, and processing stages, noise is always present in digital images. Robotics, education, biometrics, biomedical imaging, medical imaging, remote sensing, security, and surveillance are just a few fields where image processing needs to be without noise. Digital image noise removal is a highly challenging technique without understanding the kind of noise. Many techniques are suggested for noise removing and restoring the original image. The current paper provided an analysis of recent studies on the recognition and removal of noises published in recent years (2017-2022) to help determine the best strategy and provide resources for academics to comprehend the present methods and strategies in this area. Ultimately, new and more effective algorithms for identifying and removing noise from the images may be developed. This study aims to highlight the effect of noise on image processing, unfortunately, noise removal algorithms are still used inaccurately due to the lack of a mechanism to determine the type of noise in the image and the appropriate type of algorithm to remove this type of noise, so this study will provide information helps the researchers to address the problem of noise in the image. © 2022 IEEE.

关键词： image denoising

来源：评论

学校读者我要写书评

暂无评论

Optical remote sensing image salient object detection via bidirectional cross-attention and attention restoration

引用

pattern recognition 2025年 164卷

作者： Gu, Yubin Chen, Siting Sun, Xiaoshuai Ji, Jiayi Zhou, Yiyi Ji, Rongrong Xiamen Univ Sch Informat Minist Educ China Key Lab Multimedia Trusted Percept & Efficient Com Xiamen 361005 Peoples R China

Salient object detection (SOD) in optical remote sensing images (ORSI) has garnered considerable attention in recent years. The inherent high complexity of scenes in ORSI poses significant challenges to SOD methods. Current models face two major limitations. First, they fail to fully exploit the relationships between features due to ignoring the bidirectional attention relationship and continuity among adjacent feature layers. Second, while non-local modules are designed to enhance global context understanding by modeling pixel-wise relationships, their traditional implementations suffer from attention vacuity, as they treat all spatial locations equally without focusing on the most informative regions. To overcome these limitations, we introduce a novel Bidirectional Cross-Attention and Attention Restoration Neural Network (BCAR-Net), comprising the Bidirectional Cross-Attention Module (BCAM) and the Attention Restoration Module (ARM). BCAM enhances the semantic representation of detail features in lower-level maps and improves the detail representation of semantic features in higher-level maps. This is achieved by computing cross-attention between two adjacent layers in a parallel bidirectional manner, playing a crucial role in spatial information representation. Additionally, ARM addresses attention vacuity through the Foreground-Background Decoupling (FBD) and Local Attention Vacuity Supplementation (LAVS) components. Specifically, FBD refines the segmentation of salient objects from their backgrounds, while LAVS remedies local object detection omissions. Experimental results demonstrate that our proposed model performs favorably and outperforms existing methods overall. Specifically, on the ORSSD and EORSSD benchmark datasets, our method outperforms the SOTA approaches by 10% and 5% in terms of MAE, respectively. The source codes and the outcomes can be accessed at https://***/ClimBin/BCARNet.

关键词： Salient object detection Optical remote sensing images Feature fusion Attention mechanism

来源：评论

学校读者我要写书评

暂无评论

DCT-Based Deep Learning of Polarimetric Scattering From a Dielectric Cylinder

引用

IEEE GEOSCIENCE AND remote sensing LETTERS 2022年 19卷 1页

作者： Wang, Xiaofeng Yang, Chao Zhang, Zheng Du, Yang Zhejiang Univ Coll Informat Sci & Elect Engn Hangzhou 310027 Zhejiang Peoples R China Northwest Inst Nucl Technol Xian 710024 Peoples R China

Electromagnetic wave interaction with cylindrical structures plays an important role in numerous research fields. For the efficient analysis of electromagnetic scattering by cylindrical components, we previously proposed a deep learning scheme which verified that with increased abstraction capability by deep neural network (DNN), it was able to handle the highly oscillatory scattering patterns to an acceptable degree. In this letter, we extend by focusing on more efficient data structural exploration by eliminating the redundancy inherent in the bistatic scattering pattern. Two schemes based on discrete cosine transform (DCT) compression are identified with increasing complexity from 2-D to 3-D DCT, realizing compression ratios of around 26 and 58, respectively, and correspondingly training is about 11 and 14 times more efficient than our previous work on equivalent graphics processing unit (GPU) requirements. Both compression-based schemes have shown good learning capabilities in terms of accuracy and generalization potential, as demonstrated by their predictions of the complicated polarimetric bistatic scattering patterns within and outside the entire dataset and by the extremely high R-2 values and low root mean square errors (RMSEs) in the scatter plots. The physical considerations of reciprocity are achieved automatically. It is expected that the proposed compression-based deep learning schemes be useful in the design of future physically based inversion algorithms.

关键词： Discrete cosine transforms Training Deep learning Three-dimensional displays Electromagnetic scattering Redundancy image reconstruction Compression cylindric component discrete cosine transform (DCT) deep neural network (DNN) polarimetric scattering

来源：评论

学校读者我要写书评

暂无评论

A YOLO-based Object Simplification Approach for Visual Prostheses 35

A YOLO-based Object Simplification Approach for Visual Prost...

引用

35th IEEE International Symposium on Computer-Based Medical Systems (CBMS)

作者： Elnabawy, Reham H. Abdennadher, Slim Hellwich, Olaf Eldawlatly, Seif German Univ Cairo Digital Media Engn & Technol Dept Cairo Egypt German Univ Cairo Comp Sci & Engn Dept Cairo Egypt German Int Univ Comp Sci Dept Cairo Egypt Tech Univ Berlin Comp Vis & Remote Sensing Berlin Germany Ain Shams Univ Comp & Syst Engn Dept Cairo Egypt

ISBN: (数字)9781665467704

ISBN: (纸本)9781665467704

Visual prostheses have been introduced to partially restore vision to the blind via visual pathway stimulation. Despite their success, some challenges have been reported by the implanted patients. One of those challenges is the difficulty of object recognition due to the low resolution of the images perceived through these devices. In this paper, a deep learning-based approach combined with image pre-processing is proposed to allow visual prostheses' users to recognize objects in a given scene. The approach simplifies the objects in the scene by displaying the objects in clip art form to enhance object recognition. These clip art images are generated by, first, identifying the objects in the scene using the You Only Look Once (YOLO) deep neural network. The clip art corresponding to each identified object is then retrieved via Google images. Three experiments were conducted to measure the success of the proposed approach using simulated prosthetic vision. Our results reveal a remarkable decrease in the recognition time, increase in the recognition accuracy and confidence level when using the clip art representation as opposed to using the actual images of the objects. These results demonstrate the utility of object simplification in enhancing the perception of images in prosthetic vision.

关键词： YOLO visual prostheses image pre-processing phosphene simulation clip art

来源：评论

学校读者我要写书评

暂无评论

An Adaptive Edge Detection Algorithm for Weed image Analysis

引用

Computer Systems Science & Engineering 2023年第12期47卷 3011-3031页

作者： Yousef Alhwaiti Muhammad Hameed Siddiqi Irshad Ahmad College of Computer and Information Sciences Jouf UniversitySakakaAljouf73211Saudi Arabia Department of Computer Science Islamia College(Chartered University)Peshawar25000Pakistan

Weeds are one of the utmost damaging agricultural annoyers that have a major influence on *** have the responsibility to get higher production costs due to the waste of crops and also have a major influence on the worldwide agricultural *** significance of such concern got motivation in the research community to explore the usage of technology for the detection of weeds at early stages that support farmers in agricultural *** weed methods have been proposed for these fields;however,these algorithms still have challenges as they were implemented against controlled ***,in this paper,a weed image analysis approach has been proposed for the system of weed *** this system,for preprocessing,a Homomorphic filter is exploited to diminish the environmental ***,for feature extraction,an adaptive feature extraction method is proposed that exploited edge *** proposed technique estimates the directions of the edges while accounting for non-maximum *** method has several benefits,including its ease of use and ability to extend to other types of ***,low-level details in the formof features are extracted to identify weeds,and additional techniques for detecting cultured weeds are utilized if *** the processing of weed images,certain edges may be verified as a footstep function,and our technique may outperform other operators such as gradient *** relevant details are extracted to generate a feature vector that is further given to a classifier for weed ***,the features have been used in logistic regression for weed *** model was assessed against logistic regression that accurately identified different kinds of weed images in naturalistic *** proposed approach attained weighted average recognition of 98.5%against the weed images ***,it is assumed that the proposed approach might help in the weed classification s

关键词： Weeds images classification enhancement logistic regression agricultural fields remote sensing

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：