Home      About Journal      Editorial Board      Instructions      Ethics Statement      Contact Us                   中文

Judgement and Disposal of Academic Misconduct Article Copyright Transfer Agreement Proof of Confidentiality Requirements for Electronic Version More.... Chinese Association of Automation National ResearchCenter for Intelligent Computing System Institute of Intelligent Machines,Chinese Academy of Sciences More....

2016 Vol.29 Issue.5, Published 2016-05-30 Papers and Reports    Researches and Applications        Papers and Reports 385 3D Visualization Method for Tongue Movements in Pronunciation LI Rui, YU Jun, LUO Changwei, WANG Zengfu Problem of 3D visualization of tongue movements in pronunciation is studied. Firstly, a precise 3D tongue model according to magnetic resonance imaging scan data is built. Based on the 3D tongue model, the electromagnetic articulometer(EMA) data collected from three points on tongue dorsum surface are used as the driven data. The mass spring technique is used to realize realistic tongue movements in pronunciation. To evaluate the effect of modeling and synthesis methods for tongue movements, the computer graphics techniques are employed to simulate the detailed effect of the tongue movements. Finally, the simulation results are compared with X-ray video of the motion characteristics of articulators for Mandarin Chinese recorded by a pronunciation specialist. The experimental result shows the proposed method achieves precise and realistic results of 3D tongue movements and it has a wide application prospect. 2016 Vol. 29 (5): 385-392 [Abstract] ( 667 ) [HTML 1KB] [ PDF 1434KB] ( 815 ) 393 Image Object Segmentation Algorithm Combining Color and Depth Information ZHENG Qingqing, WU Jin, WEI Longsheng, LIU Jin The object contour is difficult to be extracted by the existing methods only using appearance information in the image with shadow, low-contrast edges and ambiguous areas. The depth discontinuities provide useful information for object boundaries identification. An image object segmentation algorithm is proposed by combining color and depth information. Firstly, the image is over-segmented into small homogeneous regions by mean-shift algorithm, and then color and depth information are combined to describe the characteristics of regions adequately. Next, seed regions of target and background are automatically selected according to depth information. Finally, an object contour is extracted by maximal similarity based region merging (MSRM). Experiment results on Middlebury and NYU-V2 databases show that the proposed algorithm is simple and effective compared with state-of-the-art algorithms. Besides, it improves the segmentation accuracy and enhances the visual effect of the segmentation image. 2016 Vol. 29 (5): 393-399 [Abstract] ( 565 ) [HTML 1KB] [ PDF 4457KB] ( 1511 ) 400 Graph-Based Algorithm for Pattern Matching with Bounded Length Gaps and One-off Constraint HU Xuegang, WANG Haiping, GUO Dan, LI Peipei The problem of pattern matching with bounded length gaps and one-off constraint (PMGO) is discussed. The structure of individual occurrences is changed by the bounded gaps, and the relation between occurrences is restricted by the one-off constraint. Thus, a large-scale sparse space of all candidate occurrences is generated. Based on the framework of the constraint satisfaction, the PMGO problem is transformed into path search in a directed acyclic graph (DAG) structure. Meanwhile, the equivalence of transformation is proved. Then, a graph-based pruning and matching (GPM) algorithm is presented. In GPM algorithm, a constraint relationship between vertexes is built under the one-off constraint, and then the path search is combined with a pruning procedure in an alternating and iterative manner. The loss rate of occurrences is used to measure existing heuristic algorithms and the completeness of the proposed GPM algorithm. The experimental results demonstrate that the GPM algorithm provides a complementary method for heuristic algorithms and it efficiently reduces the loss rate of occurrences. 2016 Vol. 29 (5): 400-409 [Abstract] ( 399 ) [HTML 1KB] [ PDF 514KB] ( 486 ) 410 Modularization Based Large-Scale Ontology Mapping Approach SUN Yufei, MA Liangli, GUO Xiaoming, QIN Jiwei The mapping efficiency is the key to some dynamic mapping applications. A modularization based large-scale ontology mapping approach is proposed. Firstly, it uses a weighted semantic distance and information content based method is employed to calculate the similarity of ontology concepts. Then, by an improved efficient agglomerative hierarchical clustering algorithm, the concepts are clustered and the sub-ontologies are extracted. Finally, an elaborate information retrieval based method is designed to find related sub-ontologies from heterogeneous ontologies. The proposed approach reduces time complexity by pruning candidate search space effectively. The experimental results show that the proposed approach improves the mapping efficiency significantly with high-quality mapping results. 2016 Vol. 29 (5): 410-416 [Abstract] ( 496 ) [HTML 1KB] [ PDF 434KB] ( 470 ) 417 Distributed Parallel Reasoning Algorithm with Rete for RDF Data WANG Jingbin, ZHENG Cuichun Most of the current distributed parallel reasoning algorithms for resource description framework (RDF) data need multiple MapReduce tasks. However, the reasoning of instances of triple antecedents under resource description framework schema (RDFS) /ontology web language (OWL) rules can not be performed expeditiously by some of these algorithms during processing massive RDF data, and the overall efficiency in reasoning process is not satisfactory. To solve this problem, a distributed parallel reasoning algorithm with Rete for RDF data on MapReduce (DRRM) is proposed to perform reasoning on distributed systems. Firstly, lists of schema triples and models for rule markup with the ontology of RDF data are built,and then alpha stage and beta stage of Rete algorithm are implemented with MapReduce at the phase of RDFS/OWL reasoning. Finally, the dereplication of reasoning results is conducted and a whole reasoning procedure of all the RDFS/OWL rules is executed. Experimental results show that the results of parallel reasoning for large-scale data can be achieved efficiently and correctly by the proposed algorithm. 2016 Vol. 29 (5): 417-426 [Abstract] ( 447 ) [HTML 1KB] [ PDF 654KB] ( 605 ) 427 Medical Image Classification Algorithm Based on KAP Directed Graph Model WU Ping, PAN Haiwei, GAO Linlin, HAN Qilong, XIE Xiaoqin, FENG Xiaoning Brain CT images have good texture features and similar texture angular point positions between them. Thus, a classification algorithm based on K nearest neighbor texture angular points (KAP) directed graph model is put forward to classify medical images. Firstly, the T-Harris method is proposed to extract texture angular points. Then, the KAP directed graph model is presented by using texture angular points and combining the inherent characteristics of medical images. Finally, a medical image classification algorithm based on the KAP directed graph model is proposed. Experimental results show good results of the presented algorithm in time complexity and accuracy. 2016 Vol. 29 (5): 427-438 [Abstract] ( 551 ) [HTML 1KB] [ PDF 774KB] ( 617 ) Researches and Applications 439 Loose Sparse Representation Based Undersampled Face Recognition with Auxiliary Dictionaries MA Xiao, ZHUANG Wenjing, FENG Jufu In the undersampled face recognition problem with uncontrolled intra-class variations, the auxiliary dictionary can not work quite well. The training dictionary and the auxiliary dictionary in the sparse representation face recognition methods have different representation abilities for the query image. Thus, different demands on the sparsity constraints of these dictionaries at representation stage are discussed. In this paper, a loose sparse representation based classification with auxiliary dictionaries (LSRCAD) is proposed by using different constraints on two types of dictionary respectively. The experiments confirm the effectiveness and the robustness of LSRCAD. LSRCAD outperforms the original sparse representation face recognition methods with auxiliary dictionaries for undersampled face recognition problems. 2016 Vol. 29 (5): 439-446 [Abstract] ( 433 ) [HTML 1KB] [ PDF 985KB] ( 381 ) 447 Parameters Optimization Model Based on Interval Concept Lattice LI Mingxia, LIU Baoxiang, WANG Liya, ZHANG Chunying To meet user′s requirements of the interval concept lattice structures and further mine the effective association rules, the interval concept lattice structure update algorithm is proposed based on the changes of parameters. According to the analysis of the update degree of the lattice structure, the interval parameter optimization model of concept lattice is built. The method of parameter approximation is employed to obtain a strategy of finding optimal parameters, and thus the problem of subjective and unpredictable parameters is solved. Finally, the instance is utilized to demonstrate the effectiveness of the proposed method and model. 2016 Vol. 29 (5): 447-454 [Abstract] ( 410 ) [HTML 1KB] [ PDF 384KB] ( 287 ) 455 Lag Correlation Mining Method Based on Improved Boolean Reduction and Layered Series for Big Data Stream REN Yonggong, QIAN Haizhen, LANG Hongyu To enhance the efficiency of lag correlation sequences mining for big data stream, a lag correlation mining method based on Boolean reduction and layered series is proposed in this paper. Firstly, by two sequence averages of the original data stream, the big data stream sequence is transformed by the improved Boolean to effectively decrease the computational cost of Boolean reduction. Secondly, through conversion and reduction of sequence elements, the number of the sequence element is reduced. And the proposed method overcomes the drawback of the traditional algorithm in computing lag correlations of all sequence elements. The experiments show the effective reduction in computational time and obvious improvement in computational accuracy of the proposed method. 2016 Vol. 29 (5): 455-463 [Abstract] ( 396 ) [HTML 1KB] [ PDF 717KB] ( 425 ) 464 Sparse Representation Method for Human Interaction CHEN Changhong, ZHANG Jie, LIU Feng In this paper, a sparse representation method for human interaction is proposed. The trajectory feature embodying global changes is fused with spatio-temporal feature emphasizing local movement. Firstly, the sparse representation of the trajectory feature is obtained by the bag of words model. Then, multi-level spatio-temporal features are produced by three layered spatial-temporal pyramid and processed by sparse coding. Multi-scale maxpooling algorithm is employed to obtain the local sparse feature. Finally, two kinds of sparse features are weighted and connected to obtain the sparse representation of human interaction. The dynamic latent conditional random field model is employed to verify the proposed sparse representation and the experimental results demonstrate the effectiveness. 2016 Vol. 29 (5): 464-471 [Abstract] ( 562 ) [HTML 1KB] [ PDF 883KB] ( 536 ) 472 Affinity Propagation Based Evolutionary Clustering Algorithm for Uncertain Data Stream XIA Cong, LU Yihong The existing algorithms can not solve the clustering problems for uncertain data stream from the perspective of temporal evolution. An evolutionary clustering algorithm based on affinity propagation for uncertain data stream (EAP-UStream) is presented. A concept of change rate of uncertain micro-cluster is put forward with the consideration of the influence of the varying factors caused by the procedure of online uncertain data stream forming the micro-clusters on offline clustering. The degree of similarity between the micro-clusters is measured in terms of uncertain data stream evolution. A concept of coupling degree of uncertain micro-clusters is proposed. Thus, the uncertain similarity matrix is constructed, and evolutionary clustering for uncertain data stream is realized with the idea of affinity propagation. The experimental results show the effectiveness of EAP-UStream. 2016 Vol. 29 (5): 472-480 [Abstract] ( 409 ) [HTML 1KB] [ PDF 483KB] ( 494 )

Supervised by China Association for Science and Technology Sponsored by Chinese Association of Automation NationalResearchCenter for Intelligent Computing System Institute of Intelligent Machines, Chinese Academy of Sciences Published by Science Press

Copyright © 2010 Editorial Office of Pattern Recognition and Artificial Intelligence Address: No.350 Shushanhu Road, Hefei, Anhui Province, P.R. China Tel: 0551-65591176 Fax:0551-65591176 Email: bjb@iim.ac.cn Supported by Beijing Magtech 　Email:support@magtech.com.cn