检索结果-内蒙古大学图书馆

Encoding discriminative representation for point cloud semantic segmentation

ELECTRONICS LETTERS 2021年第6期57卷 258-260页

作者： Zhao, Yuehua Ma, Jie Wang, Qian Ye, Mao Wu, Lin Hebei Univ Technol Sch Elect & Informat Engn Tianjin Peoples R China

The main challenge for semantic segmentation based on deep learning is how to encode effective representation due to the nature of point cloud. This letter proposes a plug-and-play module to abstract information from local, multi-scale context, and interdependency to make up for the deficiency. Given an intermediate feature, our module first performs a multi-scale context layer to adaptive density variance and gets rich contextual information. Then, our module infers attention masks along two dimensions, channel and pointwise in parallel. The attention masks are performed on the input features for further cultivation. Through this module, we can refine features from different perspectives. Experiments conducted on the Stanford 3D Indoor Semantics dataset demonstrate the efficacy of the proposed method. The results show that our method can achieve state-of-the-art performance.

关键词： Optical, image and video signal processing Mobile robots Computational geometry computer vision and image processing techniques

来源：评论

学校读者我要写书评

暂无评论

Absolute localisation in confined spaces using deep geometric features

引用

ELECTRONICS LETTERS 2021年第16期57卷 621-623页

作者： Brogaard, Rune Y. Ravn, Ole Boukas, Evangelos Tech Univ Denmark Dept Elect Engn Lyngby Denmark

When operating in dark and confined spaces, the capacity of the robots to localise in an absolute reference frame is of utmost importance. This letter presents an absolute localisation system, using deep 3D landmarks, for known confined spaces. The system estimates the robot's relative localisation using visual inertial odometry. Local deep 3D landmarks are extracted from the robot's view. Similar 3D landmarks are, also, extracted from the prior map which are then registered with the local landmarks to provide absolute localisation via an extended Kalman filter. To the best of knowledge, deep 3D feature registration has not been used before for absolute localisation. The proposed localisation system is tested within a representative application area-i.e. a structured, confined space-and the results indicate greater accuracy and lower processing time when compared to mainstream 3D registration approaches.

关键词： Optical, image and video signal processing Filtering methods in signal processing Mobile robots computer vision and image processing techniques

来源：评论

学校读者我要写书评

暂无评论

Autonomous trail-following unmanned aerial vehicle system based on resource partitioning of single hardware platform

引用

ELECTRONICS LETTERS 2021年第6期57卷 245-248页

作者： Lim, Yoojin Kim, Kyungil Shin, Jinah Lim, Chaedeok Elect & Telecommun Res Inst Daejeon South Korea

As deep neural networks are spreading to almost all fields, flight systems in the unmanned aerial vehicle (UAV) domain are undergoing various transitions to intelligent systems. Among these transitions-in a bid to reduce flight risk-is the active research domain of autonomous navigation for intelligent UAVs. The autonomous trail-following flight system that this letter introduces can safely consolidate flight control and mission control within the latest commercial hardware platform. The resource usage and degradation of pass-through delay in vision-based convolutional neural network workloads show that virtualisation overhead is not significantly negative, and the overall performance of the introduced system is acceptable. Real-time cooperation is also verified as achievable-in that the workloads incur minimal communication delay-between the controls. Finally, the actual field test analysis demonstrates the applicability of our autonomous UAV system, whereby our system controls the UAV to follow the centre of a set trail.

关键词： Optical, image and video signal processing Aerospace control Mobile robots Telerobotics computer vision and image processing techniques Control engineering computing Aerospace engineering computing Neural nets

来源：评论

学校读者我要写书评

暂无评论

Network slimming for compressed-sensing cardiac cine MRI

引用

ELECTRONICS LETTERS 2021年第7期57卷 297-299页

作者： Park, Seong Jae Lim, Chae Guk Ahn, Chang-Beom Kwangwoon Univ Dept Elect Engn Seoul South Korea

The Unet is attempted to be slimmed for compressed-sensing cardiac cine magnetic resonance imaging. Despite the excellent performance of the U-net, its heavy structure and long training time restrict its applications in an environment with limited computational resources. We slimmed the U-net by changing the multiple convolutions to single convolution and the transpose convolution to upsampling and by adopting fewer layers, without performance degradation. The number of trainable weights and training time of the slimmed network was reduced by 87.9% and 48.1%, respectively. The proposed network showed improved performance with a 1.45% reduction in the normalised mean square error.

关键词： image and video coding Biomedical magnetic resonance imaging and spectroscopy Interpolation and function approximation (numerical analysis) computer vision and image processing techniques Medical magnetic resonance imaging and spectroscopy Patient diagnostic methods and instrumentation Biology and medical computing Neural nets

来源：评论

学校读者我要写书评

暂无评论

A new range-only measurement-based glass line feature extraction method

引用

ELECTRONICS LETTERS 2021年第21期57卷 804-806页

作者： Lee, Seung Hwan Oh, Jung Hyun An, Ye Chan Kumoh Natl Inst Technol Sch Elect Engn Gumi Si Gyeongsangbuk D South Korea Sungkyunkwan Univ Dept Elect & Comp Engn Suwon Gyeonggi Do South Korea Kwangwoon Univ Dept Robot Seoul South Korea

This letter addresses a new range-only measurement-based glass line feature extraction method, which allows a low-cost light detection and ranging (LiDAR) sensor to detect glass-like objects without intensity and multi-echo data. The principle that the emitted LiDAR ray incidence angle with respect to a glass surface is close to 0 degrees is concerned. The geometric relationship between the sensor and glass objects is also considered for successful glass feature extraction. The extracted features are accumulated and reformulated for the generation of glass line features using the RANdom SAmple Consensus algorithm (RANSAC). Two experiments with different robot paths were conducted in a glass environment. The results showed that the proposed method accurately represented glass objects without any false-positive glass feature.

关键词： Metrological applications of lasers Patient diagnostic methods and instrumentation Instrumentation and techniques for geophysical, hydrospheric and lower atmosphere research Metrological applications of lasers Optical, image and video signal processing Radar equipment, systems and applications Optical radar computer vision and image processing techniques Biology and medical computing

来源：评论

学校读者我要写书评

暂无评论

CMOS image sensor for wide dynamic range feature extraction in machine vision

引用

ELECTRONICS LETTERS 2021年第5期57卷 206-208页

作者： Kim, Hyeon-June Kangwon Natl Univ Dept Elect Informat Commun Engn Gangwon South Korea

This letter presents a wide dynamic range (WDR) feature extraction (FE) readout scheme for machine vision applications using CMOS image sensors (CISs). The proposed scheme with the proposed pixel structure has two operating modes, the normal and WDR modes. In the normal operating mode, the proposed CIS captures a normal image with high sensitivity. In addition, as a unique function, a bi-level image is obtained for real-time FE even if a pixel is saturated in strong illumination conditions. Thus, compared to typical CISs for machine vison, the proposed CIS can reveal object features that are blocked by light in real time. In the WDR operating mode, the proposed CIS produces a WDR image with its corresponding bi-level image. A prototype CIS was fabricated using a standard 0.35-mu m 2P4M CMOS process with a 320 x 240 format (QVGA) with 10-mu m pitch pixels. At 60 fps, the measured power consumption was 5.98 mW at 3.3 V for pixel readout and 2.8 V for readout circuitry. The dynamic range of 73.1 dB was achieved in the WDR operating mode.

关键词： image recognition image sensors computer vision and image processing techniques

来源：评论

学校读者我要写书评

暂无评论

Will people like your illustration? Popularity assessment of illustrations

引用

ELECTRONICS LETTERS 2021年第2期57卷 58-61页

作者： Yang, Jiachen Men, Yao Zhao, Yang Tianjin Univ Sch Elect & Informat Engn Tianjin Peoples R China

Illustrations frequently appear in advertisements, films, books, etc. Research into the popularity of illustrations shows practicality and necessity already, which plays an important role in the improvement of illustrations and advertising decisions. In this letter, an Illustration Popularity Assessment Network (IPANet) is designed, inspired by the illustration creation process. Atrous convolutional layers are densely concatenated to extract image composition information and Gram matrix is used to measure the image style. We also introduce a scoring model based on 'views' and 'likes' to score the popularity of collected illustrations. The IPANet is proved to be effective according to experimental results on the scored illustrations. As far as we know, we are the first to explore the popularity of illustrations and practice successfully.

关键词： Other topics in statistics Graphics techniques computer vision and image processing techniques

来源：评论

学校读者我要写书评

暂无评论

UDA-Net: Densely attention network for underwater image enhancement

引用

IET image processing 2021年第3期15卷 774-785页

作者： Li, Yang Chen, Rong Dalian Maritime Univ Coll Informat Sci & Technol Dalian 116026 Peoples R China

Underwater imaging usually suffers from negative impacts due to the absorption and scattering effects in water. Underwater images thus have unfavourable visual quality to support the work in such environment. This paper addresses the problem of image improvement for single underwater image. The core idea lies in a new enhancement model based on deep learning architecture, in which a feature-level attention model is developed. This model is a multi-scale grid convolutional neural network that can facilitate fusing different types of information during representation learning. According to this information combination, a synergistic pooling mechanism is proposed to extract the channel-wise attention maps to derive the locally weighted features. Therefore, this model can adaptively focus on the feature regions corresponding to degraded patches in one underwater image and improve these patches consistently. Comprehensive experiments are conducted on benchmark and natural underwater images, and it can be demonstrated that this model is effective.

关键词： Optical, image and video signal processing computer vision and image processing techniques Neural nets

来源：评论

学校读者我要写书评

暂无评论

Generative domain adaptation for chest X-ray image analysis

引用

IET image processing 2021年第13期15卷 3118-3129页

作者： Yin, Baocai Liu, Wenchao Fu, Zhonghua Zhang, Jing Liu, Cong Wang, Zengfu Univ Sci & Technol China Hefei Peoples R China iFLYTEK Res Hefei Peoples R China Xian iFLYTEK Hyper Brain Informat Technol Co Xian Peoples R China Univ Sydney Darlington NSW Australia

Chest X-ray images taken under different conditions follow different distributions, preventing the models trained on a domain from generalising well on the other domain. In this paper, a generative domain adaptation (GDA) method is proposed to address this issue and facilitate the learning process for downstream analysis. GDA adapts different domains to a virtual common one where images are aligned at the appearance level. To this end, a domain shared generator is used to transform the input images and two competitive discriminators are used to adversarially supervise the transforming process. The domain discriminator drives the generator to narrow the domain gap while the fidelity discriminator forces the generator to keep the inherent information. Moreover, a specific classification or detection network is attached to the generator to supervise it in a task-oriented manner. Experiment results on a large-scale dataset containing 46k chest X-ray images demonstrate that GDA outperforms representative domain adaptation methods by a large margin for both disease classification and lesion detection as well as provides useful transformed images to assist experts for diagnosis.

关键词： Optical, image and video signal processing computer vision and image processing techniques Patient diagnostic methods and instrumentation Biology and medical computing

来源：评论

学校读者我要写书评

暂无评论

A deep sequence-to-sequence method for accurate long landing prediction based on flight data

引用

IET INTELLIGENT TRANSPORT SYSTEMS 2021年第8期15卷 1028-1042页

作者： Kang, Zongwei Shang, Jiaxing Feng, Yong Zheng, Linjiang Wang, Qixing Sun, Hong Qiang, Baohua Liu, Zhen Chongqing Univ Coll Comp Sci 174 Shazhengjie Chongqing 400044 Peoples R China Chongqing Univ Minist Educ Key Lab Dependable Serv Comp Cyber Phys Soc Chongqing Peoples R China Civil Aviat Flight Univ China Flight Technol & Flight Safety Res Base Guanghan Peoples R China Guilin Univ Elect Technol Guangxi Key Lab Optoelect Informat Proc Guilin Peoples R China Guilin Univ Elect Technol Guangxi Key Lab Trusted Software Guilin Peoples R China Chongqing Adtech Sci & Technol Co Ltd Chongqing Peoples R China

In civil aviation industry, runway overrun is a typical landing safety incident concerned by both airlines and authorities. Among various contributing factors to the runway overrun incident, long landing plays an important role. However, existing studies for long landing prediction mainly depend on classic machine learning methods and handcrafted features. As a result, they usually require much expert knowledge and provide unsatisfactory results. To address these problems, this paper proposes an innovative deep sequence-to-sequence model which utilizes QAR (Quick Access Recorder) data for accurate long landing pre- diction. Specifically, to cope with the high heterogeneity of QAR dataset, a data pre-processing procedure is first proposed which includes data cleaning, interpolation and normalization steps. Second, to avoid the noises incurred by too many QAR parameters and relieve the reliance on expert experience, the GBDT (gradient boosting decision trees) model is employed to choose the most relevant parameters as features. Then a CNN-LSTM and TG-attention encoder-decoder architecture is proposed to accurately predict future aircraft ground speed and radio height sequences, based on which the touchdown distance can be finally calculated. Experimental results on a large QAR dataset with 44,176 A321 flights validate effectiveness of the proposed method.

关键词： computer vision and image processing techniques Data handling techniques Aerospace engineering computing Aerospace industry Other topics in statistics Combinatorial mathematics

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：