检索结果-内蒙古大学图书馆

arXiv 2024年

作者： Shi, Zican Hu, Jing Ren, Jie Ye, Hengkang Yuan, Xuyang Ouyang, Yan He, Jia Ji, Bo Guo, Junyu Huazhong University of Science and Technology Wuhan China National Key Laboratory of Science and Technology on Multi-Spectral Information Processing Wuhan China

The introduction of Feature Pyramid Network (FPN) has significantly improved object detection performance. However, substantial challenges remain in detecting tiny objects, as their features occupy only a very small proportion of the feature maps. Although FPN integrates multi-scale features, it does not directly enhance or enrich the features of tiny objects. Furthermore, FPN lacks spatial perception ability. To address these issues, we propose a novel High Frequency and Spatial Perception Feature Pyramid Network (HS-FPN) with two innovative modules. First, we designed a high frequency perception module (HFP) that generates high frequency responses through high pass filters. These high frequency responses are used as mask weights from both spatial and channel perspectives to enrich and highlight the features of tiny objects in the original feature maps. Second, we developed a spatial dependency perception module (SDP) to capture the spatial dependencies that FPN lacks. Our experiments demonstrate that detectors based on HS-FPN exhibit competitive advantages over state-of-the-art models on the AI-TOD dataset for tiny object detection. Copyright © 2024, The Authors. All rights reserved.

关键词： High pass filters

来源：评论

学校读者我要写书评

暂无评论

Convergence acceleration for Lucas-Kanade algorithm via discarding Strongly Concave/Convex Areas

引用

Journal of information and Computational Science 2014年第2期11卷 433-447页

作者： Qin, Yueming Cao, Zhiguo Zhuo, Wen Yu, Zhenghong National Key Laboratory of Science and Technology on Multi-spectral Information Processing Wuhan 430074 China School of Automation Huazhong University of Science and Technology Wuhan 430074 China

Image alignment is a useful technique in many important research fields of computer vision. Among all the image alignment algorithms, Lucas-Kanade algorithm is one of the most widely used algorithms. During the past 30 years, a wide variety of extensions have been made to the original formulation and an unabridged Lucas-Kanade algorithm series has been formed. In this paper, we propose a novel method which can accelerate the convergence of Lucas-Kanade algorithm series. Based on the theory of strongly concave/convex function, we define the Strongly Concave/Convex Areas (SC/CA) of images. Through discarding the SC/CA, the iteration number of Lucas-Kanade algorithm series can be reduced without decreasing the precision. Finally, we present various experimental results that validate our method. These results also show that the proposed method is robust with the parameter used in detecting the SC/CA. Copyright © 2014 Binary information Press.

关键词： Iterative methods

来源：评论

学校读者我要写书评

暂无评论

Fast marine route planning for UAV using improved sparse A * algorithm

Fast marine route planning for UAV using improved sparse A *...

引用

International Conference on Genetic and Evolutionary Computing

作者： Yang, Xin Ding, Ming-Yue Zhou, Cheng-Ping Huazhong Unvi. of Sci. and Tech. Image Processing and Intelligence Control Key Laboratory of Education Ministry of China Wuhan 430074 Hubei China School of Life Science and Technology Huazhong Unvi. of Sci. and Tech. Image Processing and Intelligence Control Key Laboratory of Education Ministry of China Wuhan 430074 Hubei China Huazhong Unvi. of Sci. and Tech. State Key Laboratory for Multi-spectral Information Processing Technologies Wuhan 430074 Hubei China

ISBN: (纸本)9780769542812

This paper focuses on route planning, especially for unmanned aircrafts in marine environment. Firstly, new heuristic information is adopted such as threat-zone, turn maneuver and forbid-zone based on voyage heuristic information. Then, the cost function is normalized to obtain more flexible and reasonable routes. Finally, an improved sparse A* search algorithm is employed to enhance the planning efficiency and reduce the planning time. Experiment results showed that the improved algorithm for aircraft in maritime environment could find a combinational optimum route quickly, which detoured threat-zones, with fewer turn maneuver, totally avoiding forbid-zones, and shorter voyage. © 2010 IEEE.

关键词： Cost functions

来源：评论

学校读者我要写书评

暂无评论

A novel 3D terrain matching algorithm based on image laser radar

A novel 3D terrain matching algorithm based on image laser r...

引用

7th World Congress on Intelligent Control and Automation (WCICA 2008), vol.17

作者： Junbin Gong Hua Cheng Jie Ma Jinwen Tian State Key Laboratory for Multi-spectral Information Processing Technologies Institute for Pattern Recognition and Artificial Intelligence Huazhong University of Science and Technology Wuhan China

The imaging laser radar was an ideal imaging sensor to get the high precision 3D terrain for Terrain aided navigation (TAN). For the application, decreasing the mismatching rate was very important. In this paper, the measurement model of imaging laser radar was introduced and a novel 3D terrain matching algorithm with low mismatching rate was presented. According to the theory of function orthogonal decomposition, a terrain orthogonal decomposition feature vector (TODFV), composed of the orthogonal decomposition coefficients of the digital elevation map (DEM), was employed to represent the features of the terrain. Then TODFV was used to generate an interim matching position vector (IMPV). The IMPV were utilized to do real-time confidence analysis to improve the matching confidence, and were fused to give the final matching at the last. The algorithm could effectively decrease the mismatching rate of terrain matching algorithm, and improved the reliability of TAN. Experiments showed that the algorithm was effective and robust.

关键词： Terrain matching Feature vector Image laser radar

来源：评论

学校读者我要写书评

暂无评论

Fast ground target detection method in infrared image sequences

Fast ground target detection method in infrared image sequen...

引用

2012 3rd IEEE/IET International Conference on Audio, Language and Image processing, ICALIP 2012

作者： Zhu, Hu Deng, Lizhen Wang, Gaihua Zhou, Gang Bai, Xiaodong National Key Laboratory of Science and Technology on Multi-spectral Information Processing Institute for Pattern Recognition and Artificial Intelligence Huazhong University of Science and Technology 1037 Luoyu Road Wuhan 430074 China College of Electrical and Electronic Engineering Huazhong University of Science and Technology 1037 Luoyu Road Wuhan 430074 China

ISBN: (纸本)9781467301718

Automatic target detection (ATD) in infrared (IR) imagery is a fundamental and challenging task in computer vision. A fast automatic target detection method in IR image sequence is proposed in this paper. Since the position and scale of target change real-timely, we can predict the target position in real-time image by using the history position of target and flight parameters information of previous and current frames, and then estimate the scale of target depending on flight parameters and imaging parameters for getting the model with the appropriate scale. In order to make the template matching more robust for target rotation, the template matching method based on parametric template vector is used to recognize the position of target. The detection result is identified by using multiframe integration based on recognition information of history and currant frames. Some experimental results using real-world images with complicated background validate the effectiveness and robustness of the proposed method under rotation and scale variance condition. © 2012 IEEE.

关键词： Template matching

来源：评论

学校读者我要写书评

暂无评论

The Design and Implement of 3D Terrain Matching Processor for Image Laser Radar

The Design and Implement of 3D Terrain Matching Processor fo...

引用

Congress on Image and Signal processing, CISP

作者： Junbin Gong Yufei Sun Hongbo Xu Jinwen Tian State Key Laboratory for Multi spectral Information Processing Technologies Institute for Pattern Recognition and Artificial Intelligence Huazhong University of Science and Technology Wuhan China

The imaging laser radar is fine measure equipment for TAN with the ability to get the high precision 3D terrain. A 3D terrain matching processor was needed to be designed for the specifical application. In this paper, base on the specialty of the imaging laser radar, the3D terrain matching processor was designed, with scheme of DSP+FPGA calculating engine, multi-level memory system, flexible parallel architecture, and FPGA-embedded accelerators. A parallel correlating accelerator base on 2D modular systolic array and other structures were designed to improve the system performance in FPGA. In the end, with strictly abiding the rules of high-speed circuit designing, the detail physical implement was drawn out. The experiments showed that the system could work well in the harsh military environment with high real-time, high reliability, and flexible architecture.

关键词： Laser radar Optical design Radar imaging Radar equipment Radar measurements Process design Engines Parallel architectures Systolic arrays System performance

来源：评论

学校读者我要写书评

暂无评论

A self-adaptive unscented Kalman filtering for underwater gravity aided navigation

A self-adaptive unscented Kalman filtering for underwater gr...

引用

IEEE Symposium on Position Location and Navigation (PLANS)

作者： Lin Wu Jie Ma Jinwen Tian State Key Laboratory of Multi-spectral Information Processing Technologies Institute of Pattern Recognition and Artificial Intelligence Huazhong University of Science and Technology Wuhan China

ISBN: (纸本)9781424450367;9781424450374

In this paper, a self-adaptive unscented Kalman filtering for underwater gravity aided navigation is constructed. It is more accurate and far easier to implement than an extended Kalman filter. Then the novel navigation algorithm based on the self-adaptive unscented Kalman filter is explored. With this method submerged position fixes for autonomous underwater vehicle can be obtained from comparing gravity fields' measurements with gravity maps. Specifically, simulation results show that navigation errors can be reduced more effectively and efficiently by the presented algorithm.

关键词： Kalman filters Gravity Error correction Equations information filtering information filters Underwater vehicles Inertial navigation Remotely operated vehicles Laboratories

来源：评论

学校读者我要写书评

暂无评论

An 800mW fully-integrated 130nm CMOS DC-DC step-down multi-phase converter, with on-chip spiral inductors and capacitors

An 800mW fully-integrated 130nm CMOS DC-DC step-down multi-p...

引用

IEEE Energy Conversion Congress and Exposition, ECCE

作者： Michael Wens Michiel Steyaert State Key Lab. of Multi-spectral Information Processing Technology Huazhong Univ. of Sci. & Tech Wuhan

ISBN: (纸本)9781424428922

A fully-integrated DC-DC step-down four-phase converter in a 130 nm 1.2 V CMOS technology is realized. Integrated metal-track inductors and integrated MOS and MIM capacitors are used for the output filter. No external components are required. The converter is designed to generate an output voltage of 1.2 V out of a 2.6 V supply voltage. Stacked transistors are used to cope with the high voltage. The maximum power conversion efficiency is 58 %, for a voltage conversion ratio of 0.46. This yields a 21 % efficiency enhancement factor over a linear regulator. The maximum output power of 800 mW is the highest ever reported for a fully-integrated DC-DC converter, by the authors knowledge. A novel control system, based on a semi-constant on/off-time keeps the output voltage constant at varying load and line conditions. Moreover it ensures stability over the entire load range from 0 mW to 800 mW.

关键词： DC-DC power converters Spirals Inductors Voltage MIM capacitors CMOS technology MOS capacitors Filters Power conversion Regulators

来源：评论

学校读者我要写书评

暂无评论

Pneumoconiosis's Gross Tissue Imaging Classification based on Morphological Feature Description

Pneumoconiosis's Gross Tissue Imaging Classification based o...

引用

2012 Fifth International Symposium on Computational Intelligence and Design [第五届计算智能与设计国际会议 ISCID 2012]

作者： Delie Ming Linying Yu Liping Xiao State Key Laboratory for Multi-spectral Information Processing Technology Huazhong University of Sci Beijing Aerospace Automatic Control Institute Laboratory of Aerospace Intelligence Control Technolog

Referring to pneumoconiosis's gross tissue imaging classification, it is fatal to know the importance of distinguishing lung nodule from macule and how many nodules and macule the pneumoconiosis's gross tissue image really get. In achieving the distinguishing and counting results, the morphology processing method and some auxiliary digital image processing methods are adopted. After processing, the statistic work begins. The statistic results of abundant candidate parameters give us the chance to pick the parameter performing extremely well. In that way, the classification of pneumoconiosis's gross tissue images is quite feasible.

关键词： pneumoconiosis morphology Otsu segmentation lung nodule and macule sobel center extraction

来源：评论

学校读者我要写书评

暂无评论

Memory-Efficient Architecture for Fast Two-Dimensional Discrete Wavelet Transform

Memory-Efficient Architecture for Fast Two-Dimensional Discr...

引用

International Conference on Computational Intelligence and Software Engineering (CiSE)

作者： Xin Tian Jiaolong Wei Jinwen Tian Department of Electronics and Information Engineering Huazhong University of Science and Technology Wuhan China State Key Laboratory of Multi-spectral Information Processing Technology Huazhong University of Science and Technology Wuhan China

Memory-efficient architecture for fast two- dimensional Discrete Wavelet Transform(DWT) with high speed and small size of on-chip memory is proposed. It is composed of one row-wise one-dimensional(1-D) DWT module and two identical column-wise 1-D DWT modules. The input data samples are directly processed by the row-wise 1-D DWT module. This architecture reduces the size of on-chip memory and output latency. In further, the time multiplexing technology makes the column-wise 1-D DWT module process different column data samples generated from the row-wise 1-D DWT module. The proposed architecture is finally compared with the state of art architectures in terms of computing time, output latency and size of on-chip memory. The comparison results demonstrate that the proposed architecture is a high-speed architecture with less consumption of on-chip memory and shorter output latency.

关键词： Computer architecture Discrete wavelet transforms Hardware Very large scale integration System-on-a-chip

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：