检索结果-内蒙古大学图书馆

Finger spelling recognition using depth information and support vector machine

MULTIMEDIA TOOLS AND APPLICATIONS 2018年第21期77卷 29043-29057页

作者： Hu, Yong Jinling Inst Technol Sch Software Engn Nanjing Jiangsu Peoples R China

In Sign Language fingerspelling systems, letters of the alphabet are presented by a diverse form or/and movement of the fingers. In this study, the presented work focus on developing a real-time translation framework of static fingerspelling alphabets. At first an adaptive k-means based cluster method for depth segmentation is proposed, where a flexible cluster number n is used instead of the pre-defined definitive one. Based on the segmentation step, a recognition framework using intensity and depth information is proposed and compared with some distinctive works. Discriminative features extracted from Histogram of Oriented Gradients (HOG), local binary patterns (LBP), and Zernike moments are used due to their simplicity and good performance. The experiments are executed on a public fingerspelling dataset, which consisted of 120,000 images representing 24 alphabet letters over five different users. The results show that the presented framework is efficient, easy implementation, and performs better than the compared approaches.

关键词： Finger spelling recognition Depth image local binary patterns Histograms of oriented gradients Zernike moments k-means Support vector machine

来源：评论

学校读者我要写书评

暂无评论

Facial Expressions Recognition Based on Cognition and Mapped binary patterns

引用

IEEE ACCESS 2018年 6卷 18795-18803页

作者： Qi, Chao Li, Min Wang, Qiushi Zhang, Huiquan Xing, Jinling Gao, Zhifan Zhang, Huailing Shaanxi Normal Univ Sch Comp Sci Engn Lab Teaching Informat Technol Shaanxi Prov Minist EducKey Lab Modern Teaching Technol Xian 710062 Shaanxi Peoples R China AVIC Beijing Aeronaut Mfg Technol Res Inst Beijing 100024 Peoples R China Ecole Polytech Dept Genie Informat & Genie Logiciel Montreal PQ H3T 1J4 Canada Chinese Acad Sci Shenzhen Inst Adv Technol Shenzhen 518055 Peoples R China Guangzhou Med Univ Informat Engn Sch Guangzhou 510182 Guangdong Peoples R China

In this paper, a new expression recognition approach is presented based on cognition and mapped binary patterns. At first, the approach is based on the LBP operator to extract the facial contours. Secondly, the establishment of pseudo-3-D model is used to segment the face area into six facial expression sub-regions. In this context, the sub-regions and the global facial expression images use the mapped LBP method for feature extraction, and then use two classifications which are the support vector machine and softmax with two kinds of emotion classification models the basic emotion model and the circumplex emotion model. At last, we perform a comparative experiment on the expansion of the Cohn - Kanade (CK +) facial expression data set and the test data sets collected from ten volunteers. The experimental results show that the method can effectively remove the confounding factors in the image. And the result of using the circumplex emotion model is obviously better than the traditional emotional model. By referring to relevant studies of human cognition, we verified that eyes and mouth express more emotion.

关键词： Facial expressions recognition local binary patterns human cognition emotion model machine learning

来源：评论

学校读者我要写书评

暂无评论

Minimization of Parallax Artifacts in Video Stitching for Moving Foregrounds

引用

IEEE ACCESS 2018年 6卷 57763-57777页

作者： Kakli, Muhammad Umer Cho, Yongju Seo, Jeongil Korea Univ Sci & Technol Daejeon 34113 South Korea Elect & Telecommun Res Inst Tera Media Res Grp Daejeon 34129 South Korea

Parallax handling in the presence of moving foregrounds is a challenging problem in video stitching. The artifacts such as blurring or ghosting become prominent when objects move across multiple views. In this paper, we propose a novel scheme to minimize the parallax artifacts caused by moving objects in a fixed video stitching framework. We first stitch one frame from input videos and generate a Lookup Table (LUT) for every input video. These LUTs provides the mapping information of input videos to the panorama domain. For every frame of the input videos, we use Structural Similarity (SSIM) Index to identify the LUT points in the overlapping region where parallax exists due to moving object. Therefore, our method does not require to match and track the features across frames to detect moving object. For every detected LUT point having parallax, a descriptor based on Histogram of Oriented Gradients (HOG) and local binary patterns (LBP) is generated and matched such that the parallax is minimized in the overlapping region. This may lead to the undesirable geometric artifacts in the non-overlapping region. To cater this issue, we smoothly propagate the correction in the non-overlapping region closer to the moving object. The proposed scheme is tested on different videos, where the overlapping region is very narrow, ranging from similar to 2.3% to similar to 9%. The experimental results demonstrated that it can significantly reduce the parallax artifacts which occur due to moving foregrounds.

关键词： Histogram of oriented gradients local binary patterns object motion parallax video stitching

来源：评论

学校读者我要写书评

暂无评论

Parallel Heat Kernel Volume Based local binary Pattern on Multi-Orientation Planes for Face Representation

引用

INTERNATIONAL JOURNAL OF PARALLEL PROGRAMMING 2018年第5期46卷 943-962页

作者： Lu, Wei Yang, Xiaomin Gou, Xu Jian, Lihua Wu, Wei Jeon, Gwanggil Beijing Jiaotong Univ Sch Software Engn Beijing Peoples R China Sichuan Univ Coll Elect & Informat Engn Chengdu Sichuan Peoples R China Incheon Natl Univ Coll Informat & Technol Incheon South Korea

Appropriate representation is one of the keys to successful face recognition technologies. Actual facial appearance sometimes differs dramatically because of variations in pose, illumination, expression, and occlusion. However, existing face representation methods remain insufficiently powerful and robust. Hence, we propose a new feature extraction approach for face representation based on heat kernel volume and local binary patterns. Multi-scale heat kernel faces are created in our proposed framework. We then reformulate these multi-scale heat kernel faces as three-dimensional volume. Furthermore, we generate multi-orientation planes from the heat kernel volume, which reflects orientation co-occurrence statistics among different heat kernel faces. Finally, we apply local binary pattern (LBP) analysis on the multi-orientation planes of the heat kernel volume to capture the microstructure and macrostructure of face appearance. Hence, we obtain the heat kernel volume based local binary pattern on multi-orientation planes (HKV-LBP-MOP) descriptor. The proposed method is successfully be paralleled. We applied the method to face recognition and obtain the performance of 99.28 and 87.82% on ORL and Yale databases respectively. Experimental results on the show that the proposed algorithm significantly outperforms other well-known approaches in terms of recognition rate.

关键词： Face presentation Face recognition Parallel multi-scale heat kernel face local binary patterns

来源：评论

学校读者我要写书评

暂无评论

Wear Debris Classification of Steel Production Equipment using Feature Fusion and Case-based Reasoning

引用

ISIJ INTERNATIONAL 2018年第7期58卷 1293-1299页

作者： Wang, Hongbing Huang, Rong Gao, Liyuan Wang, Weishen Xu, Anjun Yuan, Fei Univ Sci & Technol Beijing Sch Comp & Commun Engn Beijing 100083 Peoples R China Univ Sci & Technol Beijing Sch Met & Ecol Engn Beijing 100083 Peoples R China

Wear debris classification is of great significance for identifying machine wear states. In this paper, a method of wear debris classification using feature fusion and CBR is proposed. The method integrates local feature LBP, global feature FD and Tamura coarseness, and then the fused features are applied in CBR system with different weights and different similarity, which is adaptable, extendable, modular and fast. The results show that the subdivision of wear debris images into size 32*32 when calculating LBP is helpful for improving the classification, the combination of local features and global features can get better results. The comparative experimental results of different classification methods show that the CBR system has the shortest time-consuming while maintaining high classification accuracy.

关键词： wear debris local binary patterns fractal dimension tamura coarseness feature fusion case-based reasoning

来源：评论

学校读者我要写书评

暂无评论

Signature recognition using binary features and KNN

引用

INTERNATIONAL JOURNAL OF BIOMETRICS 2018年第1期10卷 1-15页

作者： Hezil, Hedjaz Djemili, Rafik Bourouba, Houcine Univ May 8th 1945 Dept Elect & Telecommun Fac Sci & Technol Guelma Algeria Univ 20 Aout 1955 Dept Elect Engn Fac Technol Skikda Algeria

This paper proposes the use of binary features in offline signature recognition systems. Indeed, offline signature recognition finds mainly its importance for the authentication of administrative and official documents in which a higher accuracy is needed. In the proposed approach, features are extracted by using two descriptors: binary statistical image features (BSIF) and local binary patterns (LBP). To assess the reliability of the method, experiments were carried out using two publicly available datasets, MCYT-75 and GPDS-100 databases. Using a k-nearest neighbour classifier, recognition performances reach values high as 97.3% and 96.1% for MCYT-75 and GPDS-100 databases respectively. In signature verification, the classification accuracy measured with equal error rate (EER) achieved 4.2% and 4.8% respectively on GPDS-100 and GPDS-160. In addition, the EER for the MCYT-75 database has attained 7.78%. All those accuracies outperformed various performance results reported in literature.

关键词： offline signature recognition feature extraction biometric local binary patterns LBP binary statistical image features BSIF KNN

来源：评论

学校读者我要写书评

暂无评论

local binary patterns for multi-view facial expression recognition

引用

COMPUTER VISION AND IMAGE UNDERSTANDING 2011年第4期115卷 541-558页

作者： Moore, S. Bowden, R. Univ Surrey Ctr Vis Speech & Signal Proc Guildford GU2 7JW Surrey England

Research into facial expression recognition has predominantly been applied to face images at frontal view only. Some attempts have been made to produce pose invariant facial expression classifiers. However, most of these attempts have only considered yaw variations of up to 45 degrees, where all of the face is visible. Little work has been carried out to investigate the intrinsic potential of different poses for facial expression recognition. This is largely due to the databases available, which typically capture frontal view face images only. Recent databases, BU3DFE and multi-pie, allows empirical investigation of facial expression recognition for different viewing angles. A sequential 2 stage approach is taken for pose classification and view dependent facial expression classification to investigate the effects of yaw variations from frontal to profile views. local binary patterns (LBPs) and variations of LBPs as texture descriptors are investigated. Such features allow investigation of the influence of orientation and multi-resolution analysis for multi-view facial expression recognition. The influence of pose on different facial expressions is investigated. Others factors are investigated including resolution and construction of global and local feature vectors. An appearance based approach is adopted by dividing images into sub-blocks coarsely aligned over the face. Feature vectors contain concatenated feature histograms built from each sub-block. Multi-class support vector machines are adopted to learn pose and pose dependent facial expression classifiers. (C) 2010 Elsevier Inc. All rights reserved.

关键词： Facial expression recognition Multi-view facial expression recognition Head pose estimation local binary patterns local gabor binary patterns

来源：评论

学校读者我要写书评

暂无评论

Robust video tracking algorithm: a multi-feature fusion approach

引用

IET COMPUTER VISION 2018年第5期12卷 640-650页

作者： Wang, Howard Nguang, Sing Kiong Wen, Jiwei Univ Auckland Dept Elect & Comp Engn Auckland New Zealand Jiangnan Univ Dept Automat Wuxi Peoples R China

This study proposes a novel robust video tracking algorithm consists of target detection, multi-feature fusion, and extended Camshift. Firstly, a novel target detection method that integrates Canny edge operator, three-frame difference, and improved Gaussian mixture model (IGMM)-based background modelling is provided to detect targets. The IGMM-based background modelling divides video frames into meshes to avoid pixel-wise processing. In addition, the output of the target detection is utilised to initialise the IGMM and to accelerate the convergence of iterations. Secondly, low-dimensional regional covariance matrices are introduced to describe video targets by fusing multiple features like pixel location, colour index, rotation and scale invariant features as well as uniform local binary patterns, and directional derivatives. Thirdly, an extended Camshift based on adaptive kernel bandwidth and robust H state estimation is proposed to predict the states of fast moving targets and to reduce the mean shift iterations. Finally, the effectiveness of the proposed tracking algorithm is demonstrated via experiments.

关键词： image fusion feature extraction video signal processing edge detection Gaussian processes mixture models covariance matrices robust video tracking algorithm multifeature fusion approach target detection extended Camshift target detection method Canny edge operator three-frame difference improved Gaussian mixture model IGMM-based background modelling video frames low-dimensional regional covariance matrices video targets multiple feature fusion pixel location colour index scale invariant features local binary patterns directional derivatives adaptive kernel bandwidth robust H state estimation mean shift iteration reduction tracking algorithm

来源：评论

学校读者我要写书评

暂无评论

Evaluation of fractal dimension effectiveness for damage detection in retinal background

引用

JOURNAL OF COMPUTATIONAL AND APPLIED MATHEMATICS 2018年 337卷 341-353页

作者： Colomer, Adrian Naranjo, Valery Janvier, Thomas Mossi, Jose M. Univ Politecn Valencia Inst Invest & Innovac Bioingn I3B Camino Vera S-N E-46022 Valencia Spain UPV FISABIO Unidad Conjunta Grp Tecnol Informat Aplicadas Oftalmol Valencia Spain Univ Orleans Lab I3MTO Rue ChartresBP 6759 F-45067 Orleans 2 France Univ Politecn Valencia ITEAM Res Inst Camino Vera S-N E-46022 Valencia Spain

This work investigates the characterization of bright lesions in retinal fundus images using texture analysis techniques. Exudates and drusen are evidences of retinal damage in diabetic retinopathy (DR) and age-related macular degeneration (AMD) respectively. An automatic detection of pathological tissues could make possible an early detection of these diseases. In this work, fractal analysis is explored in order to discriminate between pathological and healthy retinal texture. After a deep preprocessing step, in which spatial and colour normalization are performed, the fractal dimension is extracted locally by computing the Hurst exponent (H) along different directions. The greyscale image is described by the increments of the fractional Brownian motion model and the H parameter is computed by linear regression in the frequency domain. The ability of fractal dimension to detect pathological tissues is demonstrated using a home-made system, based on fractal analysis and Support Vector Machine, able to achieve around a 70% and 83% of accuracy in E-OPHTHA and DIARETDB1 public databases respectively. In a second experiment, the fractal descriptor is combined with texture information, extracted by the local binary patterns, improving the bright lesion detection. Accuracy, sensitivity and specificity values higher than 89%, 80% and 90% respectively suggest that the method presented in this paper is a robust algorithm for describing retina texture and can be useful in the automatic detection of DR and AMD. (C) 2018 Elsevier B.V. All rights reserved.

关键词： Retinal fundus image Fractal analysis Diabetic retinopathy Age-related macular degeneration local binary patterns

来源：评论

学校读者我要写书评

暂无评论

Ideal neighbourhood mask for speech enhancement

引用

ELECTRONICS LETTERS 2018年第5期54卷 317-318页

作者： Arcos, C. D. Vellasco, M. Alcaim, A. Catholic Univ CETUC PUC Rio Ctr Telecommun Studies BR-22451900 Rio De Janeiro RJ Brazil

A novel approach for speech enhancement applications by applying spectral mask estimation is introduced. The new application uses the local binary patterns to estimate an ideal neighbourhood mask. This will indicate which time-frequency units of the noisy speech are dominated by the noise. The performance assessment of the proposed application in conjunction with the traditional mask techniques, i.e. ideal binary mask and ideal ratio mask, are carried out under various environments in terms of the objective speech quality measures, as well as word error rate performance in speech recognition systems using deep neural networks. Results indicated that the proposed mask yielded significantly better performance than the conventional techniques.

关键词： speech recognition speech enhancement speech intelligibility time-frequency analysis neural nets deep neural networks speech recognition systems word error rate performance objective speech quality measures performance assessment time-frequency units ideal neighbourhood mask estimation local binary patterns spectral mask estimation speech enhancement

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：