模式识别与人工智能
Home      About Journal      Editorial Board      Instructions      Ethics Statement      Contact Us                   中文
Pattern Recognition and Artificial Intelligence
22 Judgement and Disposal of Academic Misconduct Article
22 Copyright Transfer Agreement
22 Proof of Confidentiality
22 Requirements for Electronic Version
More....
22 Chinese Association of Automation
22 National ResearchCenter for Intelligent Computing System
22 Institute of Intelligent Machines,Chinese Academy of Sciences
More....
 
 
2013 Vol.26 Issue.6, Published 2013-06-30
Orignal Article   
   
Orignal Article
513 Forward Search Planning Based on Implicit Decomposition with Landmarks
WEI Wei,OUYANG Dan Tong

A planning algorithm that implicitly decomposes the forward search procedure with landmark information is proposed. The original planning task is decomposed into several smaller sub tasks according to the estimation of the landmark counting heuristic. Whenever the search visits a state with lower heuristic value,a sub task is completed.The procedure is iterated until the estimation of the landmark counting heuristic reduces to zero. The plan fragments of all the sub tasks are connected into the final solution. The experimental results show the superiority of the proposed algorithm. The implicit task decomposition is introduced by the landmark counting heuristic. Therefore,compared with the existing decomposition methods that treat landmarks as mandatory subgoals,the proposed algorithm guides the forward search procedure faster,cuts down the search space dramatically and makes considerable improvement in both planning efficiency and planning quality.
2013 Vol. 26 (6): 513-520 [Abstract] ( 540 ) [HTML 1KB] [ PDF 416KB] ( 666 )
521 A Probabilistically Quantitative Reasoning System Based on n-Valued ukasiewicz Propositional Logic
ZHOU Hong-Jun
The probabilistic truth degree functions of propositions in n-valued ukasiewicz propositional logic are abstracted as a modality,and their three basic identities are abstracted as axioms of the modality. Then,a modal logic system for reasoning about probabilistic truth degrees of propositions is proposed. The syntax and semantics are constructed. The completeness theorem with respect to probabilistic truth degree functions is finally obtained. A logic foundation is set for probabilistically quantitative logic.
2013 Vol. 26 (6): 521-528 [Abstract] ( 306 ) [HTML 0KB] [ PDF 393KB] ( 512 )
529 Gait Recognition Based on Spatio-Temporal HOG Feature of Plantar Pressure Distribution
XIA Yi,MA Zu-Chang,YAO Zhi-Ming,SUN Yi-Ning
A gait recognition method based on spatio-temporal histogram of oriented gradient (HOG) is proposed. Spatio-temporal HOG embodies the feature fusion of spatial and temporal plantar pressure information. Firstly,several key points,such as the maximum and the minimum points on the pressure-time curve,are picked out. Then,plantar pressure distribution images corresponding to the moment of key points are used to construct spatio-temporal HOG feature vector. Finally,support vector machine classification is applied to implement gait classification. Gait samples are collected from 30 persons at different walking speeds. When the walking speeds of the samples in the training and testing sets are not differentiated,the recognition rate is 93.5%. The experimental results demonstrate that spatio-temporal HOG feature accurately describes the dynamic plantar pressure distribution during walking,and it also has good speed-adaptable properties.
2013 Vol. 26 (6): 529-536 [Abstract] ( 691 ) [HTML 0KB] [ PDF 621KB] ( 1211 )
537 An Improved Particle Filter SLAM Algorithm
WANG Xiao-Hua,YANG Xing-Fang
The estimation accuracy of the conventional particle filter algorithm is low because the historical information is not fully utilized. Combining the high estimation accuracy of exactly sparse delayed-state filter(ESDF) and the high efficiency of exactly sparse extended information filter(ESEIF),an improved particle filter SLAM algorithm is proposed. In this algorithm,the information matrix of ESDF,maintaining the historical relationship of robot pose and characteristics,improves the accuracy of the estimate,and ESEIF overcomes the defects of robot rotational state and characteristics density.Results of both emulational and factual experiments show that the proposed algorithm is valid and feasible.
2013 Vol. 26 (6): 537-542 [Abstract] ( 612 ) [HTML 0KB] [ PDF 892KB] ( 711 )
543 Visibility Measurement with Image Understanding
XU Xi,YIN Xu-Cheng,LI Yan,HAO Hong-Wei,CAO Xiao-Zhong
High hardware cost,complex operation and narrow application are main problems of existing measuring methods of atmospheric visibility. In this paper,machine learning is introduced into the study of visibility measurement and a method of daytime visibility measurement is proposed based on image understanding. Firstly,image features and vectors based on pixel contrast are designed and extracted grounded on the segmentation of the regions of interest in measured scene images. Then,the relational model between image features and visibility is constructed by training support vector regression. Finally,visibility of images to be measured is computed according to the model. Experimental results show that the proposed method has both high visibility measuring precision and good flexibility. Moreover, it reduces the limitations of existing visibility measurement methods.
2013 Vol. 26 (6): 543-551 [Abstract] ( 782 ) [HTML 0KB] [ PDF 1132KB] ( 1294 )
552 Maximum Similarity Matching Emotion Model Based on Mapping between State Space and Probability Space
WANG Hao,ZHANG Quan-Yi,FANG Bao-Fu,FANG Shuai
Robot emotion modeling is a hot issue in emotion robot research. Based on the emotion psychology knowledge,a dynamic emotion transfer model of the emotion robot is presented with different personalities under different external stimulation. The influences of personality and external stimulation are discussed. The emotion model based on state space is used to describe the emotion states of robot. The emotiontransfer process is simulated by hidden Markov model (HMM) process. However,the HMM process can only work out the current probability of the emotion state. To get the concrete emotion state,the maximum similarity matching emotion transfer model based on mapping between state space and probability space is proposed. Firstly,the current emotion probability is calculated by HMM process. Then,the current concrete emotion state is obtained by maximum similarity matching. Different personalities and stimulation can be built by adjusting the parameters of the model. The proposed model simulates the transformation process effectively. The experimental results show that the emotion transfer process simulated by the proposed model corresponds with the general rules of human emotion transformation.
2013 Vol. 26 (6): 552-560 [Abstract] ( 679 ) [HTML 0KB] [ PDF 559KB] ( 759 )
561 Loop Closure Detection AlgorithmBased on Monocular Vision Using Visual Dictionary
LIANG Zhi-Wei,CHEN Yan-Yan,ZHU Song-Hao,GAO Xiang,XU Guo-Zheng
Aiming at the problem of loop closure detection in monocular simultaneous localization and mapping for mobile robots,a detection algorithm based on visual dictionary (VD) is presented. Firstly, feature extraction is performed for each required image using SURF methods. Subsequently,a fuzzy K-means algorithm is employed to cluster these visual feature vectors into visual words based on VD which is constructed online. To precisely represent the similarities between each visual word and corresponding local visual features ,Gaussian mixture model is proposed to learn the probability model of every visual word in bags of visual words. Consequently,every image can be denoted as a probabilistic vector of VD,and thus the similarities between any two images can be computed based on vector inner product. To guarantee the continuity of the closed-loop detection,a Bayesian filter method is applied to fuse historical closed-loop detection information and the obtained similarities to calculate the posterior probability distribution of closed-loop hypothesis. Furthermore,two memory management mechanisms,shallow memory and deep memory,are introduced to improve the process speed of the proposed algorithm. The experimental results demonstrate the validity of the proposed approach.
2013 Vol. 26 (6): 561-570 [Abstract] ( 724 ) [HTML 0KB] [ PDF 1607KB] ( 1783 )
571 Natural Landmark Detection of Mobile Robots Based on Bayesian Surprise of Salient Scenes
QIAN Kun,MA Xu-Dong,DAI Xian-Zhong,FANG Fang,YANG Hong
Nature landmark detection of mobile robot in unknown and unstructured environment is a basis of hierarchical environmental mapping. A natural landmark detection method is proposed based on Bayesian Surprise of salient scenes. Visual attention map of scene images is computed to guide the SURF feature sampling within the scope of salient regions. The improved spatial bag-of-words model (sBoW) is employed to construct the pattern vectors of scene appearance. Multivariate Polya model based on the spatial bag-of-words paradigm is proposed for representing the place,and the detection of landmarks corresponding to salient scenes is achieved by computing the surprise of sensor measurements. The experimental results validate the low miss alarm rate and false alarm rate of the nature landmark detection method in large-scale and complex environment,as well as the effectiveness of generating topological nodes with the combination of hierarchical SLAM method.
2013 Vol. 26 (6): 571-576 [Abstract] ( 348 ) [HTML 0KB] [ PDF 916KB] ( 870 )
577 Precise Facial Feature Localization under Non-Restraint Environment with Limited Training Images
CHEN Ying,ZHANG Long-Yuan
After analyzing the limitation of current methods,a precise localization strategy with limited training data is proposed in a probability framework. Texture and geometry information of facial elements are extracted as model features after comparison analysis with other traditional descriptors. Gaussian mixture model is used for the probability modeling,which describes the distribution of each model features extracted from different facial conditions well. Then,a series of fusion strategies are designed for the facial features localization,which considers the probability distribution of each facial feature,the distribution characters of their surrounding elements and their geometry constraints. The experimental results show that the proposed method can realize precise localization for the facial features with limited training sample images which belong to a single subject,and it outperforms other methods in localization accuracy.
2013 Vol. 26 (6): 577-583 [Abstract] ( 599 ) [HTML 0KB] [ PDF 1333KB] ( 597 )
584 Weighting Binary Transformation Algorithm for Non Co-occurrence Data
JI Bo,YE Yang-Dong
The assumption that all data features are equally important in the categorical data-sequential information bottleneck(CD-sIB) lowers the transformation quality. A weighting binary transformation method is proposed to reveal the feature of non co-occurrence data by highlighting the representative features and depressing the redundancy features. Meanwhile,two weighting rules,the applicability of stochastically distributed data and the non supervision of weighting schemes,are introduced. Then,the weighted categorical data-sequential information bottleneck(WCD-sIB) algorithm is presented based on the weighting granularity concept. The experimental results show that the weighting binary transformation method generates good co-occurrence data representation,and the WCD-sIB algorithm is superior to the other algorithms.
2013 Vol. 26 (6): 584-591 [Abstract] ( 506 ) [HTML 0KB] [ PDF 465KB] ( 596 )
592 A Swarm Pattern Global Search Algorithm
QU Liang-Dong,HE Deng-Xu,WU Jin-Zhao
Pattern search algorithm often falls into local optimization and its efficiency is low. Inspired by swarm intelligence algorithm,a global optimization algorithm,swarm pattern global search algorithm (SPGSA),is proposed. Swarm intelligence is introduced to SPGSA in the evolution process. Thus,SPGSA includes pattern search operator,pattern moving operator,pattern learning operator and pattern dispersion operator.It has a strong ability of global and local search as well as better features of fast convergence and good stability. Comparisons of the simulation results by using standard benchmark functions prove the effectiveness.
2013 Vol. 26 (6): 592-597 [Abstract] ( 548 ) [HTML 0KB] [ PDF 410KB] ( 687 )
598 An Uncorrelated Space Algorithm Based on Fisher Minimum Criterion and Its Application to Face Recognition
YANG Jun,LIU Yan-Li,FENG Chao-Sheng,FENG Lin
Uncorrelated space algorithm is a fast method for extracting uncorrelated discriminant vectors based on the generalized fisher criterion,but it requires the total-scatter matrix to be reversible. To solve this problem,an improved uncorrelated features extraction method based on the generalized Fisher minimum criterion and uncorrelated space algorithm is proposed. The main idea of the proposed method is to solve the discriminant vectors of generalized fisher minimum criterion in the non-null subspace of the between-class scatter matrix. The rationality of the idea is discussed. A strategy including two steps is proposed to get the non-null subspace efficiently from high dimensional data. Firstly,the original data are mapped to a low dimensional subspace by PCA algorithm. Then,the non-null subspace of the between-class scatter matrix can be solved efficiently in the subspace,and the rationality of the process is discussed. The experimental results on standard face database show that the proposed method is efficient with higher accuracies compared with Fisherface algorithm and the uncorrelated space algorithm.
2013 Vol. 26 (6): 598-603 [Abstract] ( 516 ) [HTML 0KB] [ PDF 354KB] ( 623 )
604 Night-Sky Cloud Image Segmentation Algorithm Based on Prior Threshold Surface
HUANG Qian,WANG Yu-Lin,WANG Shao-Long,TANG Da-Jun
Affected by the atmospheric pollution,the moonlight and the zodiac light,the night-sky cloud images vary greatly. The conventional threshold methods which only utilize pixel gray value as well as the neighborhood information are difficult to segment the image accurately because of their uneven backgrounds. In this paper,two prior features of cloud images are observed from the statistical analysis,and a prior threshold surface based cloud segmentation algorithm is presented. After reliable background regions are extracted according to the prior features,an adaptive threshold surface can be obtained by polynomial fitting on the background regions,and the values of the threshold surface are between the clouds and backgrounds. Thus the cloud can be segmented from the background. The experimental results show that the proposed algorithm is more feasible and effective compared with other existing algorithms. Moreover,it produces fine results on the cloud images of light influence.
2013 Vol. 26 (6): 604-608 [Abstract] ( 520 ) [HTML 0KB] [ PDF 1414KB] ( 629 )
模式识别与人工智能
 

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