检索结果-内蒙古大学图书馆

Representation of freshwater aquaculture fish behavior in low dissolved oxygen condition based on 3D computer vision

引用

MODERN PHYSICS LETTERS B 2018年第34-36期32卷

作者： Bao, Y. J. Ji, C. Y. Zhang, B. Gu, J. L. Changzhou Inst Technol Sch Elect & Photoelect Engn Changzhou 213032 Peoples R China Nanjing Agr Univ Coll Engn Nanjing 210031 Jiangsu Peoples R China

Dissolved oxygen (DO) plays an important role in industrialized freshwater aquaculture. Such deficiencies such as the high cost of water-quality monitoring system and the failure to accurately monitor or describe aquaculture water-quality existed in freshwater aquaculture water-quality monitoring system. Here, a kind of representation method applied to characterize industrialized aquaculture fish behavior in different degrees of DO deficiency is based on three-dimensional (3D) computer vision. 3D coordinate values of aquaculture fishes in water acquired from 3D computer vision Device by processing aquaculture fish image are applied to represent such parameters as the average activity and height of aquaculture fish in water. This method for representing different behaviors of industrialized freshwater aquaculture fish under the condition of anoxia is realized by using these parameters and combing with the experience of aquaculture. The results show that the representation of industrialized freshwater aquaculture fish based on 3D computer vision System can be applied to describe industrialized aquaculture fish behavior and effectively compensate for the shortfall spatial location of aquaculture fish unable to acquire from 2D monitoring system, which is helpful for the accurate and reasonable control of DO in aquaculture.

关键词： Fish behavior dissolved oxygen three-dimensional vision system

来源：评论

学校读者我要写书评

暂无评论

People Counting for Indoor Monitoring 3

People Counting for Indoor Monitoring

引用

3rd international conference ON COMPUTING, ENGINEERING, AND DESIGN (ICCED)

作者： Wahyuni, Elvira Sukma Alinra, Rizqi Renafasih Setiawan, Hendra Univ Islam Indonesia Dept Elect Engn Yogyakarta Indonesia

ISBN: (纸本)9781538634950

This study aims to develope an automatic system of people counting for indoor monitoring. People detection was performed using Histogram of Oriented Gradien (HOG) based feature descriptor and Support Vector Machine (SVM) classification. In this study several experiment were done to obtain best detection accuraccy i.e., Camera distance, light intensity and object distance. The result showed inaccurate detection is affected by the distance between camera and object, less of light intensity and objects coincide.

关键词： People Counting HOG SVM image processing computer vision

来源：评论

学校读者我要写书评

暂无评论

Time Efficient Design and FPGA Implementation of Distributed Canny Edge Detector Algorithm

Time Efficient Design and FPGA Implementation of Distributed...

引用

IEEE international conference on Recent Trends in Electronics, Information & Communication Technology (RTEICT)

作者： V Raghavendra Lakshmi Shrinivasan Affiliated to VTU Bangalore India

ISBN: (数字)9781538624401

ISBN: (纸本)9781538624418

Edge detection is one of the basic operations carried out in image processing and object identification. As it plays an important role in image processing which needs to be optimized, accurate and less latency architecture. This paper focuses on Edge detection using Adaptive threshold technique. The efficient Canny edge detector has been used in real time application. In this paper, the Gaussian filter is applied as preprocessing block to remove the high frequency edges. The Canny coefficients are approximated to suite the hardware requirements with less LUT's and further the adaptive threshold technique is applied to obtain the finer details of edges. The proposed method of edge computation using adaptive threshold reduces memory elements significantly, because it occupies less area, reduction in delay and increased efficiency without effecting detection performance. The proposed edge detection architecture is implemented using Xilinx system generator tool on Spartan6 ATLYS (XC6SLX45) board. It is observed that the proposed method is better in terms of time performance as compared to existing architectures.

关键词： image edge detection Kernel computer architecture Detectors Thresholding (Imaging) Field programmable gate arrays

来源：评论

学校读者我要写书评

暂无评论

An Adaptive Approach For image Denoising Using Pixel Classification And Gaussian Conditional Random Field Technique

An Adaptive Approach For Image Denoising Using Pixel Classif...

引用

3rd international conference on Computing, Communication, Control and Automation (ICCUBEA)

作者： Thakare, Bhushan S. Deshmukh, Hemant R. St Gadge Baba Amravati Univ Amravati India Sinhgad Acad Engn Dept Comp Engn Pune Maharashtra India Dr Rajendra Gode Inst Technol & Res Dept Comp Engn Amravati India

ISBN: (纸本)9781538640081

Recently, digital image processing has attracted researchers due to its significant performance for real-time application such as bio-medical system, security system and automated computer vision based systems. This process is widely accepted in banking systems for automated cheques processing systems. During this automated process, images are captured and processed through computer vision application for signature verification and other aspect computation. However, various systems are present for this application but precise automatic verification remains a challenging task for researchers. During processing, images are capture where unwanted signals get added into original signal and degrade the image quality resulting in degraded performance. To overcome this, image pre-processing plays an important role where edge detection, image resampling, image enhancement and image denoising are the key component. In this work, our main aim is to improve the image denoising performances by developing an image denoising application. To overcome this issue, we introduce a novel approach for image denoising by applying pixel classification using multinomial logistic regression (MLR) (for classification) and Gaussian Conditional Random Field is used for denoising to generate efficient performance for image denoising application. Proposed work comprise of two procedures such as: (i) parameter generation by considering multinomial logistic regression (MLR) based on input noisy image and (ii) designing an inference network whose layer perform the computations which are tangled in GCRF formulation. An extensive simulation study is performed on open source benchmark databased and compared with state of art filtering schemes. Experimental study shows that proposed approach gives better performance when compared with existing models.

关键词： image processing image denoising conditional random field cheque fraud detection hybrid modeling

来源：评论

学校读者我要写书评

暂无评论

Mammogram classification method based on GMM and GLCM-PSO-PNN

Mammogram classification method based on GMM and GLCM-PSO-PN...

引用

2018 3rd international conference on Automation,Mechanical Control and Computational Engineering(AMCCE 2018)

作者： Xiaojian Zhang Chengjian Wei Xili Wan School of computer science and technology Nanjing Tech University

Facing the condition that the inefficient training of traditional classifiers in the classification process of mammography, a classification method is proposed combining image processing and supervised learning. Firstly, the improved adaptive median filter enhances the image contrast. Then, according to the result of breast segmentation based on Gauss Mixture Model(GMM), this paper proposed a classification model based on Probabilistic Neural Network optimized(PNN) optimized by Gray Level Co-occurrence Matrix(GLCM) and Particle Swarm Optimization(PSO). The eigenvector extracted from the GLCM can be used as input to simplify the network structure. The smoothing factor optimized by PSO used to train the network can improve accuracy. The results in public mammographic patches database demonstrate that the model can classify the types of mammography effectively and perform better than the previous methods.

关键词： mammogram gauss mixture model probabilistic neural network gray level co-occurrence matrix particle swarm optimization

来源：评论

学校读者我要写书评

暂无评论

2D Face Recognition on CNN with Half-Average-Face

2D Face Recognition on CNN with Half-Average-Face

引用

2018 3rd international conference on Communication,image and Signal processing（CCISP 2018）

作者： Yangfeng Zheng School of Data and Computer Science SUN YAT-SEN UNIVERSITY

As the size of training dataset of face recognition models becomes larger and larger, we are interested in a method called Average-Half-Face（AHF）, which could halve the size of training samples. The AHF method divides a full face into two halves and then averages them together（reversing the columns of one of the halves）. We preprocess the dataset with the method of AHF, and train them on two different models, Eigenfaces and Convolutional Neural Network（CNN）. We compare the prediction results with those models trained on the original dataset. Previous researches showed that AHF is superior to Full-Face（FF）, while our experiment results further showed that in some cases AHF also boosts CNN. The application of AHF can bring both saving in storage and reduction on training cost time.

关键词：

来源：评论

学校读者我要写书评

暂无评论

BAG: A Binary Descriptor for RGB-D images Combining Appearance and Geometric Cues 1

引用

3rd international conference on Cognitive Systems and Information processing (ICCSIP)

作者： Xiao, Xiuzi He, Songhua Guo, Yulan Lu, Min Zhang, Jun Hunan Univ Coll Informat Sci & Engn Changsha 410082 Hunan Peoples R China Natl Univ Def Technol Coll Elect Sci & Engn Changsha 410073 Hunan Peoples R China Chinese Acad Sci Inst Comp Technol Beijing 10080 Peoples R China

ISBN: (数字)9789811052309

ISBN: (纸本)9789811052309;9789811052293

Feature matching forms the basis for numerous computer vision applications. With the rapid development of 3D sensors, the availability of RGB-D images has been increased stably. Compared to traditional 2D images, the additional depth images in RGB-D images can provide more geometric information. In this paper, we propose a new efficient binary descriptor (namely BAG) for RGB-D image representation by combining appearance and geometric cues. Experimental results show that the proposed BAG descriptor produces better feature matching performance with faster matching speed and less memory than the existing methods.

关键词： Feature matching Binary descriptor Local binary pattern RGB-D image

来源：评论

学校读者我要写书评

暂无评论

Automation inspection dimensional and parts of recognition and industrial materials 3

Automation inspection dimensional and parts of recognition a...

引用

3rd international conference on Numerical and Symbolic Computation: Developments and Applications, SYMCOMP 2017

作者： De Melo, Charles Luiz S. Junior, Almir K. Bastos, Moises P. Manaus Institute of Technology 69088-130 Brazil University of the State of Amazonas Control Engineering and Automation 69020-000 Brazil

ISBN: (纸本)9789899941038

Most industrial visual inspection systems still dealing with a low variety of object classes. This is because even the most recognition techniques and representation scheme are not flexible enough to support major changes in the field of objects to recogni ze and scale. To avoid this type of restriction, our system incorporates digital processing techniques of image, a software able to do all sorts of for recognition and sizing and a fully compatible hardware with all the operational requirem ents of the sof tware. From the analysis of the captured images, our system is able to choose, in a learning phase, the representation scheme best suited for each of the pieces, which facilitates and speeds up the identification and subsequent location of the same in the recognition phase. The advantages of applying these techniques representation were observed in several experiments, both with simulated real images as real-time images. The objective of this article is to describe the implementation of a computer vision system, able to choose the best strategy to recognize measure and locate planar objects, aiming at industrial applications. © ECCOMAS, Portugal.

关键词： computer vision

来源：评论

学校读者我要写书评

暂无评论

License Plate Recognition Using Deep FCN 1

引用

3rd international conference on Cognitive Systems and Information processing (ICCSIP)

作者： Wu, Yue Li, Jianmin Shenzhen Coll Int Educ Shenzhen Peoples R China Tsinghua Univ Dept Comp Sci & Technol Tsinghua Natl Lab Informat Sci & Technol State Key Lab Intelligent Technol & Syst Beijing Peoples R China

ISBN: (数字)9789811052309

ISBN: (纸本)9789811052309;9789811052293

In our work, we concentrate on the problem of car license plate recognition after the plate has been extracted from an image. Traditional methods approach this problem as three separate steps: preprocessing, segmentation, and recognition. In this paper, we propose a unified approach that integrates these steps using a fully convolutional network. We train a 36-class FCN on a dataset of single characters and apply it to height-normalized license plates. The architecture of this model successfully reduces the loss in detail during end-to-end convolution. Finally, we extract the results from the output sequences of probabilities using a variant of the NMS algorithm. The experiments on public license plate datasets show that our approach outperforms the state-of-the-art methods.

关键词： computer vision Fully Convolutional Network License plate recognition End-to-end recognition Residual learning

来源：评论

学校读者我要写书评

暂无评论

“FOLO”: A vision-Based Human-Following Robot

“FOLO”: A Vision-Based Human-Following Robot

引用

2018 3rd international conference on Automation,Mechanical Control and Computational Engineering(AMCCE 2018)

作者： Evan Chen Department of Information Art ＆ Design Tsinghua University

The main challenge in following robot is solving the real-time user localization problem. In this paper we introduce our solution for it together with our whole robot system called "FOLO". Using extra-tool（such as Bluetooth） may cause inconvenience in interaction;3 D detector plus tracker ID approaches are at risks of computing consume and ROI flash;Approaches based on 2 D appearance have challenges on appearance change, re-detecting and complex background. We present a 2 D-appearance approach on "FOLO" which can work overcome above issues. Our approach utilizes consensus of corresponding method to track, and then, updates features by supervised learning into classifier cascade to against challenges of re-detecting and complex background. Moreover, we give pre-processing procedure on each frame to sharpen edges on image to enhance quality of tracking and use adaptive background to overcome challenge of complex background. This paper illustrates that our tracking approach can work against common challenges of tracking in our own designed experiments, has a rapid speed over 25 fps and can achieve state-of-the-art results. We use two-layer-PID method to control "FOLO" for a long-term task which allows "FOLO" succeeding to follow user in office environment.

关键词： Human-following Learning Detection Autonomous navigation

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：