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....
 
 
2012 Vol.25 Issue.3, Published 2012-06-25
Orignal Article   
   
Orignal Article
361 A Granular Space Reduction Approach to Pessimistic Multi-Granulation Rough Sets
SANG Yan-Li, QIAN Yu-Hua
Multi-granulation rough set method (MGRS) is one of new directions in rough set theory. It is a data modeling method in the context of multiple granular spaces. Firstly, a concept of distribution reduction is introduced to pessimistic multi-granulation rough model, and a granular space selection under multiple granular spaces is investigated. Then, the important measure of a granular space in this model is defined, and an algorithm is designed to obtain a granular space reduction in the pessimistic multi-granulation rough model. Finally, an example is employed to verify the validity of the proposed algorithm. The obtained results are much closer to the practical decision.
2012 Vol. 25 (3): 361-366 [Abstract] ( 858 ) [HTML 1KB] [ PDF 324KB] ( 788 )
367 Update Summarization Based on Heat Conduction Model
DU Pan, GUO Jia-Feng, ZHANG Jin, CHENG Xue-Qi, ZHANG Xu
Besides the problems of topic relevance and information diversity tackled by traditional topic-focused multi-document summarization, the update summarization is required to address the problem of information novelty as well. In this paper, HeatSum, an extractive approach based on heat conduction for update summarization, is proposed. The process can naturally make use of the relationships among the given topic, the old sentences, the new sentences, and the sentences selected and to be selected to find proper sentences for update summarization. Therefore, HeatSum is able to simultaneously address the challenging problems above for update summarization in a unified way. The experiments on benchmark of TAC2009 are performed and the ROUGE evaluation results show that the HeatSum achieves fine performance compared to the best existing performing systems in TAC tasks and it significantly outperforms other baseline methods.
2012 Vol. 25 (3): 367-374 [Abstract] ( 712 ) [HTML 1KB] [ PDF 440KB] ( 858 )
375 Comparison of Application between Evolutionary Algorithm and Algorithm Based on Homotopy
DING Jun-Xiang, GE Yun-Jian, LI Shan-Hong, XU Fei, SHUANG Feng
To access the tactile information of the large-scale flexible tactile array sensor, the applications of evolutionary algorithm and algorithm based on homotopy theory for sensor decoupling are compared. The decoupling results show that the performance of evolutionary algorithm is much better than other artificial intelligence algorithms for the sensor array of a certain size, but it fails in the real time capability and the accuracy for the sensor array with a large scale size. The algorithm based on the homotopy theory changes the traditional static decoupling into the dynamic decoupling by introducing a time parameter into the decoupling processing of sensor. It can not merely be used for information decoupling of a large scale flexible tactile array sensor, moreover, it is applied to the real-time and accurate decoupling for the information acquisition problem of high-dimensional and multi parameter sensor.
2012 Vol. 25 (3): 375-381 [Abstract] ( 365 ) [HTML 1KB] [ PDF 563KB] ( 767 )
382 Discovering News Topics from Microblogs Based on Hidden Topics Analysis and Text Clustering
LU Rong,XIANG Liang, LIU Ming-Rong, YANG Qing
A method of news topics extraction from large-scale short posts of microblogging-service is proposed. Through the hidden topic analysis, the similarity measurement of short texts is solved well. In every time window, the short posts which are most likely to talk about news events are selected according to the characteristics of the news. Then, a two-level K-means-hierarchical hybrid clustering method is used to cluster all the selected data into different news topics. The experimental results show the proposed method works well on large-scale microblog dataset.
2012 Vol. 25 (3): 382-387 [Abstract] ( 858 ) [HTML 1KB] [ PDF 478KB] ( 1226 )
388 Rule Acquisition Algorithm Based on Maximal Granule
ZHANG Qing-Hua, WANG Guo-Yin, LIU Xian-Quan
Granular computing (GrC) is a method for simulating human thinking and solving complicated problems. It is a powerful tool for solving complicated problems, mining massive data sets, and dealing with fuzzy information. In this paper, the shortcoming of the traditional rule extraction methods is presented, and then the granularity principle of rule extraction is analyzed based on granular computing method. The essence of attribute reduction is to choose a maximum approximation partition space of decision-making knowledge space, and the rules acquired from maximum approximation partition space may not be the simplest. Therefore, a rule extraction algorithm based on granular computing is proposed. In the proposed algorithm, the rules based on maximal granule can be acquired from information system in a hierarchical knowledge space in top-down manner, and the results of the simulation experiments illustrate that the generalization ability of rough set method is improved.
2012 Vol. 25 (3): 388-396 [Abstract] ( 806 ) [HTML 1KB] [ PDF 453KB] ( 723 )
397 Image Segmentation Using Cloud Model and Data Field
WU Tao,QIN Kun

An image thresholding method is proposed to select the optimal threshold for image segmentation. The data field is introduced to map the original image from grayscale space to the corresponding potential space by the proposed method, and the relative and the absolute data fields are produced with two different mass functions. Then, by considering the features of the data fields and combining the global and the local statistical characteristics of the image, a potential threshold for data field can be yielded. Next, the pixels are divided into possible backgrounds and objects, which are represented with background and object cloud models generated by backward cloud generator. According to the membership degrees of pixels over two cloud models, the final result of image thresholding is obtained by maximum determination method. It is indicated by the experiments that the proposed method yields accurate and robust result, and it is reasonable and effective.
2012 Vol. 25 (3): 397-405 [Abstract] ( 705 ) [HTML 1KB] [ PDF 598KB] ( 1044 )
406 Neighbor Class Linear Discriminate Analysis
WANG Yan-Wei, DING Xiao-Qing, LIU Chang-Song
A method of neighbor class linear discriminant analysis (NCLDA) is proposed. Linear discriminant analysis (LDA) is a special case of this method. LDA finds the optimal projections by maximum between-class scatter while by minimum within-class scatter. The between-class scatter is an average over divergences among all classes. In NCLDA,between-class scatter is defined as average divergences between one class and its k nearest neighbor classes. By selecting proper numbers of neighbor class, NCLDA alleviates overlaps among classes caused by LDA. The experimental results show that the proposed NCLDA is robust and outperforms LDA.
2012 Vol. 25 (3): 406-410 [Abstract] ( 694 ) [HTML 1KB] [ PDF 499KB] ( 521 )
411 Fast and Precise Two-Dimensional Renyi Entropy Image Thresholding
ZHANG Xin-Ming, XUE Zhan-Ao, ZHENG Yan-Bin
In view of the inaccurate segmentation and the high computational complexity of the traditional two-dimensional (2-D) Renyi entropy (RE) thresholding method, a fast and precise 2-D RE image thresholding method is presented. Firstly, the 2-D histogram is divided into inner, edge and noise areas by four oblique lines in parallel with the main diagonal line, and the noise points of the noise areas in the 2-D histogram are eliminated to obtain better segmentation performance. Then, the values of inner and edge areas in the 2-D RE formula are calculated precisely to get a more accurate threshold. Finally, a recursive algorithm of the precise 2-D RE image thresholding method is proposed, and an approach based on the recursive algorithm is inferred with the computational features and two formulas of 2-D RE to reduce the computational complexity. The experimental results show that the proposed method achieves more accurate segmentation results and more robust anti-noise capability compared with other contrast methods, and its running time is much less, almost the same as that of the current RE recursive algorithm based on 2-D histogram oblique segmentation.
2012 Vol. 25 (3): 411-418 [Abstract] ( 806 ) [HTML 1KB] [ PDF 951KB] ( 675 )
419 Space Feature Based Spectral Clustering for Noisy Image Segmentation
LIU Han-Qiang, ZHAO Feng
To overcome the problem that the traditional spectral clustering is easily influenced by image noise while applied to noisy image segmentation, a space feature based spectral clustering algorithm for noise image segmentation is proposed. In this method, gray value, local spatial information and non-local spatial information of each pixel are utilized to construct a 3-dimensional feature dataset. Then, the space compactness function is introduced to compute the similarity between each feature point and its K nearest neighbors. Finally, the final image segmentation result is obtained by spectral clustering algorithm. Some noisy artificial images, nature images and synthetic aperture radar images are utilized and normalized. Cut, FCM_s and Nystrom method are compared with the proposed method in the experiments. The experimental results show that the proposed method is robustness and obtains the satisfying segmentation result.
2012 Vol. 25 (3): 419-425 [Abstract] ( 904 ) [HTML 1KB] [ PDF 1164KB] ( 869 )
426 Model for Transforming Vague Sets into Fuzzy Sets Based on Convex Simplex
WEI Bo, PENG Jun-Huan, YANG Hong-Lei
Making use of a corresponding relation between Vague sets and the three triangles within a same plane of convex simplex, a simplex geometric expression is presented for Vague sets and transforming them into Fuzzy sets. It effectively solves the problem of geometric interpretation in methods or models for transforming Vague sets into Fuzzy sets. A simplex transformation model(S-TM) is proposed for transforming Vague sets into Fuzzy sets, and the corresponding transformation rules are also presented. Compared with the existing transformation methods or models, the proposed S-TM is more effective and visualized in geometric expression, and more explicit and determinate in geometric interpretation. Comparative results show that transforming Vague sets into Fuzzy sets is fuzzy and gradual.
2012 Vol. 25 (3): 426-434 [Abstract] ( 661 ) [HTML 1KB] [ PDF 420KB] ( 579 )
435 Correlation Analysis on Multidimensional Data Streams Based on Base-Windows
QIAN Jiang-Bo, WANG Zhi-Jie, CHEN Hua-Hui, DONG Yi-Hong, XIE Zhi-Jun, WANG Yong-Li
Multidimensional data stream analysis is seldom studied, even the minor contribution is mainly from the analytical works on a single sliding window model. An on-line correlation analysis algorithm called Base_win_CCA algorithm is presented, which significantly reduces space and time complexity by performing simultaneous correlation analysis on multidimensional data streams. Technically, the algorithm achieves the correlation of multiple windows in a flexible and accurate way by dynamically maintaining statistics data. Theoretical analysis and experimental results indicate that the proposed algorithm is remarkable in performance when the window is larger with sufficient data streams and users.
2012 Vol. 25 (3): 435-444 [Abstract] ( 508 ) [HTML 1KB] [ PDF 723KB] ( 772 )
445 Chinese Event Type Recognition Based on Conditional Random Fields
HU Bo-Lei, HE Rui-Fang, SUN Hong, WANG Wen-Jun
The result of event argument recognition cannot guide event type recognition in the traditional multi-step event extraction methods. Nevertheless the performance of event extraction system largely depends on event type recognition. In order to address the backward dependency of event type recognition on event argument recognition, event extraction is considered as a sequence labeling. In this paper, an improved conditional random field joint labeling model is proposed. The event type and event argument are labelled simultaneously in the graph model. The solution of the unbalanced data problem is discussed through embedding trigger word. The experiments on ACE 2005 Chinese corpus show that the performance of event type recognition is improved by the proposed method and F-score achieves 63.53%.
2012 Vol. 25 (3): 445-449 [Abstract] ( 520 ) [HTML 1KB] [ PDF 319KB] ( 933 )
450 Image Retrieval Method Based on Combined Euler Vector and Edges Direction Histogram
YANG Hong-Ju, JIAN Xiao-Yan, CAO Fu-Yuan, QIAN Yi-Li
How to retrieve and query images effectively is an important research topic in the image database. An image retrieval method is proposed which integrates combined Euler vector and edges orientation histogram(EOH). Firstly, the combined Euler vector feature is extracted from the edge images for image retrieval (EEXO algorithm). Then, to make a distinction among the images which have different shapes and similar Euler features, an image retrieval algorithm named EEXOEOH is presented which combines the EEXO algorithm and the EOH algorithm. The experimental results demonstrate that EEXOEOH has higher retrieval efficiency than the other four algorithms.
2012 Vol. 25 (3): 450-455 [Abstract] ( 599 ) [HTML 1KB] [ PDF 1191KB] ( 763 )
456 Breakthroughs in Artificial Intelligence and Innovation in Methodology
ZHONG Yi-Xin

The following breakthroughs have been made in the field of artificial intelligence (AI) research over the last decade: 1) The common kernel mechanism of intelligence formation, the information-knowledge-intelligence conversion, was discovered. Thus, the mechanism simulation of intelligence was established. 2)The knowledge ecological structure, the empirical knowledge-regular knowledge-commonsense knowledge growth, which is all supported by innate knowledge, was discovered. 3) The combination of 1) and 2) has led to another discovery that the existing and independent AI approaches, the structural simulation approach to cortex of the brain, the functional simulation approach to the logical thinking, and the behavior simulation approach to the intelligent beings are three special harmonious cases of the mechanism simulation of intelligence under respective knowledge. Therefore, the unified simulation approach and AI theory are achieved, which opens up prospects for AI research. It is believed that the radical source for the breakthroughs in AI is the innovation of scientific methodology.
2012 Vol. 25 (3): 456-461 [Abstract] ( 605 ) [HTML 1KB] [ PDF 364KB] ( 1666 )
462 Disparities Assisted Video Inpainting
GAO Hui, XU Wei, ZHANG Mao-Jun, WANG Yun-Li, WANG Wei
A disparities assisted streetscape video inpainting method is proposed. Occlusions in the video are detected and the positional relationship between target region and reference region is built according to the disparities. A camera motion associated filling order as well as a progressive repairing strategy is applied in the proposed method. Because the disparity is a kind of pixel-wise positional relationship between images, the effectiveness of the proposed method is not related with the complexity of the scenes. Furthermore, the experimental results show that the proposed method has the capability to remove occlusions with lager area.
2012 Vol. 25 (3): 462-468 [Abstract] ( 304 ) [HTML 1KB] [ PDF 1287KB] ( 647 )
469 Long Query User Satisfaction Analysis Based on User Behaviors
ZHU Tong, LIU Yi-Qun, RU Li-Yun, MA Shao-Ping
Performance evaluation is one of the most important issues in web search. Long queries contain much information which describes users information demand correctly. Thus, a long query search user satisfaction detection framework is proposed. The concept of user satisfaction is defined. The relevant user behavior features in user logs are extracted which are combined with Decision Tree and SVM to identify satisfactory or unsatisfactory queries. The experimental results on large scale practical search engine data show the effectiveness of the proposed framework. Furthermore, the classification accuracies of satisfactory and unsatisfactory queries reach 86% and 70%, respectively.
2012 Vol. 25 (3): 469-474 [Abstract] ( 486 ) [HTML 1KB] [ PDF 471KB] ( 896 )
475 Vein Image Segmentation Algorithm Based on Function Optimization in Regions of Interest
JIA Xu, CUI Jian-Jiang, XUE Ding-Yu, PAN Feng
A near infrared dorsal hand vein image segmentation algorithm is proposed based on function optimization about entropy and gradient in local regions of interest. The noise is removed by applying compressed sensing theory in vein image firstly. Then, the Bandelet transform is used to extract the regions of interest including vein information, and the established function about entropy and gradient is constrained and optimized in these regions so that the vein information and the background can be separated. Finally, the segmentation results in all the regions of interest are fused, and the whole vein image segmentation process is accomplished. The experimental results show that the proposed algorithm makes the acquired segmentation image reserve better vein features than other segmentation algorithms. In addition, the proposed algorithm has good reference value in the segmentation of finger vein and palm vein images with texture features.
2012 Vol. 25 (3): 475-480 [Abstract] ( 462 ) [HTML 1KB] [ PDF 623KB] ( 746 )
481 PPI Network Clustering Based on Artificial Bee Colony and Breadth First Traverse Algorithm
TIAN Jian-Fang, LEI Xiu-Juan
The clustering of protein-protein interaction (PPI) network is one of the principal methods to reveal and research the protein function.The traditional clustering methods are inefficient for PPI network due to its special characters. Therefore, a clustering method is proposed based on the optimal search of artificial bee colony (ABC) algorithm and the breadth first traverse (BFT) clustering algorithm. To avoid noisy interference on experimental results, the distance-density algorithm is used to roughly determine the number of clustering in the preprocessing stage. Then, the initial clustering center is determined based on the comprehensive feature value of nodes in the network. The BFT algorithm is used in the clustering process and the improved ABC algorithm is employed to automatically search the optimal merging threshold. Finally, the performance of the proposed algorithm is estimated by precision and recall and some key parameters of the algorithm is analyzed. The experimental results show that the proposed algorithm improves the clustering effect of the PPI network efficiently.
2012 Vol. 25 (3): 481-490 [Abstract] ( 717 ) [HTML 1KB] [ PDF 758KB] ( 649 )
491 Multi-Robots Formation System Architecture Based on Flexible Structure
CAI Yun-Fei, TANG Zhen-Min, ZHAO Chun-Xia, YANG Jing-Yu
A flexible control structure based on constrained control is proposed to solve the problem of multi-robots formation architecture designing in the assumed mission of automatic minefield clearing. It utilizes the advantages of redundancy and impact resistance to enhance the adaptation to environmental perturbations. It utilizes filtering localization algorithm based on prediction, feedback and correction to generate the prestress of formation structure and enhance the system carrying capacity compatible with the appropriate formation shape. The multi-role allocation and collaboration model is used to improve the system performance. The experimental results show that the flexible structure of the multi-robots system adapts to environmental changes well. The system has good flexibility and survivability capabilities. Its robustness is improved and the control method based on flexible structure has good practical value.
2012 Vol. 25 (3): 491-499 [Abstract] ( 723 ) [HTML 1KB] [ PDF 1138KB] ( 771 )
500 Identification of Query Intents via Combining Multiple Features
WU Da-Yong,ZHAO Shi-Qi,LIU Ting,ZHANG Yu
Identifying underlying user intents of search engine queries is a hotspot in the field of web information retrieval. An approach to identifying user intents of search engine queries is proposed based on features from various sources. Specifically, the query intent identification is regarded as a classification problem. The classification features are extracted from various sources including query texts, search engine feedbacks and query logs. The method is evaluated on the real web query data. The experimental results show that the exploited features are helpful to improve the identification performance. Furthermore, about 88.5% of the test queries can be correctly identified with the classification framework via combining all the features.
2012 Vol. 25 (3): 500-505 [Abstract] ( 775 ) [HTML 1KB] [ PDF 356KB] ( 1683 )
506 A Fast Compressed Video Retrieval Method Based on Fluctuation of Frame Data Counts
GAO Hao-Lin, LI Bi-Cheng, ZHANG Bai-Yu
To implement the fast retrieval for compressed video data, a retrieval method based on the fluctuation of frame data counts is proposed. The data counts for each frame in compress domain are calculated to acquire the data counts curves of equal-length query segment and target video. Then, the query segment is slided on the target video after the alignment of frame data I, and the length of sliding window is the same as the length of a group of picture. The differences of data counts fluctuation between the query clip and the target video are measured. Finally, the similarity result is given according to the designated threshold. The extraction of high dimension feature vector for each frame is omitted in the proposed method, and a video clip is represented by a single vector instead of an array of high dimension vectors. The experimental results show that the video retrieval is speeded based on the proposed method. Meanwhile, a high accuracy is achieved. Therefore, the method can be used for fast retrieval on compressed video database and online video segment matching to find the target video.
2012 Vol. 25 (3): 506-512 [Abstract] ( 455 ) [HTML 1KB] [ PDF 757KB] ( 696 )
513 Ball Particle Filter Algorithm for Visual Tracking
XIA Yu,WU Xiao-Jun
Particle degeneration is a key issue which influences the performance of a particle filter. To improve the quality of particle sampling and the accuracy of visual tracking, a ball particle filter algorithm for visual tracking is proposed. Ball sampling mode guarantees the valid particles in state-space. Compared to the conventional particle filter, the proposed method uses much fewer particles to ameliorate the diversity of distribution, and overcomes the degeneration problem effectively. By iterative motion of ball, particles are moved towards the regions with larger values of posterior density function. Ball particle filter without depending on state-mode can track the maneuver object with irregular movement. The simulation results show that the proposed method improves the efficiency of particles and achieves fine tracking precision.
2012 Vol. 25 (3): 513-520 [Abstract] ( 675 ) [HTML 1KB] [ PDF 1606KB] ( 631 )
521 A Mongolian-English Word Alignment Approach Based on Discriminative Model
ZHANG Guan-Hong, Odbal, GONG Zheng
Word alignment is an essential issue in the field of natural language processing.A discriminative word alignment method is proposed using the linear CRF model for Mongolian-English language pair. According to the differences between Mongolian and English languages, morphological, lexical and part-of-speech features can be incorporated into the CRF model, and a dual-layer CRF word alignment model is constructed. In the first layer, the chunks that are split from the sentence are aligned. Then in the second layer, the words of chunks are aligned using CRF word alignment model. The experimental results on Mongolian-English task demonstrate that the proposed method improves the performance of word alignment.
2012 Vol. 25 (3): 521-526 [Abstract] ( 503 ) [HTML 1KB] [ PDF 397KB] ( 941 )
527 Optimizing Merging Results of Multiple Resource Retrievals by a Particle Swarm Algorithm
XIE Xing-Sheng, ZHANG Guo-Liang, LI Bin
To automatically merge the result from multiple independent research engines (IREs) is a key component of the metasearch engine development and it is problem in distributed information retrieval applications as well. After testing a variety of existing result merging algorithms for multiple IRE results, a discrete particle swarm algorithm (DPSA) is proposed for further optimizing a group of merging results produced by other result merging algorithms.The experimental results show that the DPSA generally outperforms all the other result merging algorithms. It usually has better adaptability in application for not having to take into account the usefulness weights of IRE results and the overlap rate among different IRE results of a query. Compared to other result merging algorithms, the recognition precision of DPSA increases nearly 20%, while the precision standard deviation for different queries decreases about 50%.
2012 Vol. 25 (3): 527-533 [Abstract] ( 470 ) [HTML 1KB] [ PDF 454KB] ( 690 )
534 Multi-Robot Active Simultaneous Localization and Mapping Based on Cooperative Correction Approach
TAO Tong, HUANG Ya-Lou, YUAN Jing, SUN Feng-Chi
Cooperation strategy is the key point in multi-robot active simultaneous localization and mapping (SLAM). A cooperation strategy based on correcting each other, called cooperative correction,is proposed. The approach is applied to multi-robot active SLAM by state machine. The cooperative correction approach improves the accuracy of localization and mapping by optimizing the observations of landmarks in two modes: weak cooperative correction and strong cooperative correction. The former is an indirect mode, which is used when all the robots are in low accuracy. The latter is a direct mode, under which the robot with high accuracy will actively correct other robot and related landmarks. Compared the proposed approach with other multi-robot active SLAM approachs, the simulation results validate the advantage of high accuracy, and compared with single robot active SLAM, the simulation results validate the advantage of low navigation cost. Finally, the experimental result in real environment with two Pioneer3-DX mobile robots validates that the cooperative correction approach effectively improves the efficiency and accuracy of multi-robot active SLAM.
2012 Vol. 25 (3): 534-543 [Abstract] ( 560 ) [HTML 1KB] [ PDF 744KB] ( 585 )
544 Improvement for Migration Operator in Biogeography-Based Optimization Algorithm
XU Zhi-Dan,MO Hong-Wei
In original biogeography-based optimization (BBO), the migration and mutation operators are applied to evolve the population. BBO is often used to solve single-objective optimization problems. When the original migration operator of BBO is applied to solve continuous multi-objective optimization problems, the diversity of the population is decreased sharply. In this paper, the migration operator of BBO is developed and the perturbation factor is introduced to increase the diversity of the population. Thus, a biogeography-based multi-objective evolutionary algorithm (BBMOEA) is proposed. Compared with the algorithm under the action of the original migration operator on benchmark test problems, the simulation results illustrate the effectiveness and efficiency of the developed migration operator. Meanwhile, compared with SPEA2 and NSGA-Ⅱ, the experimental results show that the solution set gained by algorithm BBMOEA has good convergence and even distribution.
2012 Vol. 25 (3): 544-549 [Abstract] ( 513 ) [HTML 1KB] [ PDF 466KB] ( 665 )
550 Medical Image Deconvolution in Besov Space Based on Sparse Decomposition
WEN Qiao-Nong ,XU Shuang ,WAN Sui-Ren
In the framework of sparse decomposition, an image deconvolution functional variation model in the Besov smooth space is proposed. The data item is constrained in the negative Hilbert-Sobolev space, while the regularization item is constrained by sparsity and smoothness. Both the sparsity and the smoothness are taken into account by regarding the L1 norm of the redundant dictionary as a sparsity measure and semi-norm as image smoothness measure in Besov space. The operator splitting method is adopted due to the difficulty in solving this mode directly. The original model is split into two parts: image deconvolution and sparse decomposition. Thus, it is solved using cross-iterative method. The resolution of the model pseudo-code is given in detail and the convergence of the algorithm is verified experimentally. The results show that deconvolution model is better than other models.
2012 Vol. 25 (3): 550-556 [Abstract] ( 620 ) [HTML 1KB] [ PDF 639KB] ( 757 )
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