IEICE Transactions on Information and Systems
Online ISSN : 1745-1361
Print ISSN : 0916-8532
Volume E94.D , Issue 2
Showing 1-30 articles out of 30 articles from the selected issue
Special Section on Foundations of Computer Science — Mathematical Foundations and Applications of Algorithms and Computer Science —
  • Shuichi MIYAZAKI
    2011 Volume E94.D Issue 2 Pages 181
    Published: February 01, 2011
    Released: February 01, 2011
    JOURNALS FREE ACCESS
    Download PDF (52K)
  • Hee-Kap AHN, Yoshio OKAMOTO
    Type: PAPER
    2011 Volume E94.D Issue 2 Pages 182-189
    Published: February 01, 2011
    Released: February 01, 2011
    JOURNALS FREE ACCESS
    We study problems in computational geometry from the viewpoint of adaptive algorithms. Adaptive algorithms have been extensively studied for the sorting problem, and in this paper we generalize the framework to geometric problems. To this end, we think of geometric problems as permutation (or rearrangement) problems of arrays, and define the “presortedness” as a distance from the input array to the desired output array. We call an algorithm adaptiveif it runs faster when a given input array is closer to the desired output, and furthermore it does not make use of any information of the presortedness. As a case study, we look into the planar convex hull problem for which we discover two natural formulations as permutation problems. An interesting phenomenon that we prove is that for one formulation the problem can be solved adaptively, but for the other formulation no adaptive algorithm can be better than an optimal output-sensitive algorithm for the planar convex hull problem. To further pursue the possibility of adaptive computational geometry, we also consider constructing a kd-tree.
    Download PDF (263K)
  • Takehiro ITO, Naoki SAKAMOTO, Xiao ZHOU, Takao NISHIZEKI
    Type: PAPER
    2011 Volume E94.D Issue 2 Pages 190-195
    Published: February 01, 2011
    Released: February 01, 2011
    JOURNALS FREE ACCESS
    Let C be a set of colors, and let ω(c) be an integer cost assigned to a color c in C. An edge-coloring of a graph G is to color all the edges of G so that any two adjacent edges are colored with different colors in C. The cost ω(ƒ) of an edge-coloring ƒ of G is the sum of costs ω(ƒ(e)) of colors ƒ(e) assigned to all edges e in G. An edge-coloring ƒ of G is optimal if ω(ƒ) is minimum among all edge-colorings of G. In this paper, we show that the problem of finding an optimal edge-coloring of a tree T can be simply reduced in polynomial time to the minimum weight perfect matching problem for a new bipartite graph constructed from T. The reduction immediately yields an efficient simple algorithm to find an optimal edge-coloring of T in time O(n1.5Δlog (nNω)), where n is the number of vertices in T, Δ is the maximum degree of T, and Nω is the maximum absolute cost |ω(c)| of colors c in C. We then show that our result can be extended for multitrees.
    Download PDF (539K)
  • Yutaka IWAIKAWA, Naoyuki KAMIYAMA, Tomomi MATSUI
    Type: PAPER
    2011 Volume E94.D Issue 2 Pages 196-199
    Published: February 01, 2011
    Released: February 01, 2011
    JOURNALS FREE ACCESS
    The firefighter problem is used to model the spread of fire, infectious diseases, and computer viruses. This paper deals with firefighter problem on rooted trees. It is known that the firefighter problem is NP-hard even for rooted trees of maximum degree 3. We propose techniques to improve a given approximation algorithm. First, we introduce an implicit enumeration technique. By applying the technique to existing (1-1/e)-approximation algorithm, we obtain $(1- {k-1 \\over (k-1)e + 1})$-approximation algorithm when a root has k children. In case of ternary trees, k=3 and thus the approximation ratio satisfies $(1- {k-1 \\over (k-1)e + 1})$ ≥ 0.6892, which improves the existing result 1-1/e ≥ 0.6321. Second technique is based on backward induction and improves an approximation algorithm for firefighter problem on ternary trees. If we apply the technique to existing (1-1/e)-approximation algorithm, we obtain 0.6976-approximation algorithm. Lastly, we combine the above two techniques and obtain 0.7144-approximation algorithm for firefighter problem on ternary trees.
    Download PDF (186K)
  • Bingbing ZHUANG, Hiroshi NAGAMOCHI
    Type: PAPER
    2011 Volume E94.D Issue 2 Pages 200-210
    Published: February 01, 2011
    Released: February 01, 2011
    JOURNALS FREE ACCESS
    In a rooted triangulated planar graph, an outer vertex and two outer edges incident to it are designated as its root, respectively. Two plane embeddings of rooted triangulated planar graphs are defined to be equivalent if they admit an isomorphism such that the designated roots correspond to each other. Given a positive integer n, we give an O(n)-space and O(1)-time delay algorithm that generates all biconnected rooted triangulated planar graphs with at most n vertices without delivering two reflectively symmetric copies.
    Download PDF (888K)
  • Bingbing ZHUANG, Hiroshi NAGAMOCHI
    Type: PAPER
    2011 Volume E94.D Issue 2 Pages 211-219
    Published: February 01, 2011
    Released: February 01, 2011
    JOURNALS FREE ACCESS
    In a rooted graph, a vertex is designated as its root. An outerplanar graph is represented by a plane embedding such that all vertices appear along its outer boundary. Two different plane embeddings of a rooted outerplanar graphs are called symmetric copies. Given integers n ≥ 3 and g ≥ 3, we give an O(n)-space and O(1)-time delay algorithm that generates all biconnected rooted outerplanar graphs with exactly n vertices such that the size of each inner face is at most g without delivering two symmetric copies of the same graph.
    Download PDF (724K)
  • Akihiro MATSUURA
    Type: PAPER
    2011 Volume E94.D Issue 2 Pages 220-225
    Published: February 01, 2011
    Released: February 01, 2011
    JOURNALS FREE ACCESS
    In this paper, we analyze recurrence relations generalized from the Tower of Hanoi problem of the form T(n,α,β)=min1≤tnT(n-t,α,β)+βS(t,3)}, where S(t,3)=2t-1 is the optimal total number of moves for the 3-peg Tower of Hanoi problem. It is shown that when α and β are natural numbers, the sequence of differences of T(n,α,β)'s, i.e., {T(n,α,β)-T(n-1,α,β)}, consists of numbers of the form β2iαj (i,j≥0) lined in the increasing order.
    Download PDF (147K)
  • Ro-Yu WU, Jou-Ming CHANG, Yue-Li WANG
    Type: PAPER
    2011 Volume E94.D Issue 2 Pages 226-232
    Published: February 01, 2011
    Released: February 01, 2011
    JOURNALS FREE ACCESS
    In this paper, we introduce a concise representation, called right-distance sequences (or RD-sequences for short), to describe all t-ary trees with n internal nodes. A result reveals that there exists a close relationship between the representation and the well-formed sequences suggested by Zaks [Lexicographic generation of ordered trees, Theoretical Computer Science 10 (1980) 63-82]. Using a coding tree and its concomitant tables, a systematical way can help us to investigate the structural representation of t-ary trees. Consequently, we develop efficient algorithms for determining the rank of a given t-ary tree in lexicographic order (i.e., a ranking algorithm), and for converting a positive integer to its corresponding RD-sequence (i.e., an unranking algorithm). Both the ranking and unranking algorithms can be run in O(tn) time and without computing all the entries of the coefficient table.
    Download PDF (357K)
  • Akio FUJIYOSHI, Masakazu SUZUKI
    Type: PAPER
    2011 Volume E94.D Issue 2 Pages 233-239
    Published: February 01, 2011
    Released: February 01, 2011
    JOURNALS FREE ACCESS
    In this paper, we study the minimum spanning tree problem with label selection, that is, the problem of finding a minimum spanning tree of a vertex-labeled graph where the weight of each edge may vary depending on the selection of labels of vertices at both ends. The problem is especially important as the application to mathematical OCR. It is shown that the problem is NP-hard. However, for the application to mathematical OCR, it is sufficient to deal with only graphs with small tree-width. In this paper, a linear-time algorithm for series-parallel graphs is presented. Since the minimum spanning tree problem with label selection is closely related to the generalized minimum spanning tree problem, their relation is discussed.
    Download PDF (461K)
  • Yu-Kumg CHEN, Chen-An FANG, Fan-Chieh CHENG
    Type: LETTER
    2011 Volume E94.D Issue 2 Pages 240-242
    Published: February 01, 2011
    Released: February 01, 2011
    JOURNALS FREE ACCESS
    The Towers of Hanoi problem is a classical problem in puzzles, games, mathematics, data structures, and algorithms. In this letter, a least memory used algorithm is proposed by combining the source array and target array for comparing the sizes of disk and labeling the disks in the towers of Hanoi problem. As a result, the proposed algorithm reduces the space needed from 2n+2 to n+5, where n represents the disks number.
    Download PDF (785K)
Regular Section
  • David G. ROSADO, Eduardo FERNÁNDEZ-MEDINA, Javier LÓPEZ
    Type: PAPER
    Subject area: Fundamentals of Information Systems
    2011 Volume E94.D Issue 2 Pages 243-254
    Published: February 01, 2011
    Released: February 01, 2011
    JOURNALS FREE ACCESS
    The systematic processes exactly define the development cycle and help the development team follow the same development strategies and techniques, thus allowing a continuous improvement in the quality of the developed products. Likewise, it is important that the development process used integrates security aspects from the first stages at the same level as other functional and non-functional requirements. Grid systems allow us to build very complex information systems with different and remarkable features (interoperability between multiple security domains, cross-domain authentication and authorization, dynamic, heterogeneous and limited mobile devices, etc). With the development of wireless technology and mobile devices, the Grid becomes the perfect candidate for letting mobile users make complex works that add new computational capacity to the Grid. A methodology of development for secure mobile Grid systems is being defined. One of the activities of this methodology is the requirements analysis which is based in reusable use cases. In this paper, we will present a UML-extension for security use cases and Grid use case which capture the behaviour of this kind of systems. A detailed description of all these new use cases defined in the UML extension is necessary, describing the stereotypes, tagged values, constraints and graphical notation. We show an example of how to apply and use this extension for building the diagram of use cases and incorporating common security aspects for this kind of systems. Also, we will see how the diagrams built can be reused in the construction of others diagrams saving time and effort in this task.
    Download PDF (4606K)
  • Ali AKRAMIZADEH, Ahmad AFSHAR, Mohammad Bagher MENHAJ, Samira JAFARI
    Type: PAPER
    Subject area: Fundamentals of Information Systems
    2011 Volume E94.D Issue 2 Pages 255-263
    Published: February 01, 2011
    Released: February 01, 2011
    JOURNALS FREE ACCESS
    Model-based reinforcement learning uses the gathered information, during each experience, more efficiently than model-free reinforcement learning. This is especially interesting in multiagent systems, since a large number of experiences are necessary to achieve a good performance. In this paper, model-based reinforcement learning is developed for a group of self-interested agents with sequential action selection based on traditional prioritized sweeping. Every single situation of decision making in this learning process, called extensive Markov game, is modeled as n-person general-sum extensive form game with perfect information. A modified version of backward induction is proposed for action selection, which adjusts the tradeoff between selecting subgame perfect equilibrium points, as the optimal joint actions, and learning new joint actions. The algorithm is proved to be convergent and discussed based on the new results on the convergence of the traditional prioritized sweeping.
    Download PDF (287K)
  • Keivan NAVI, Mohammad ESMAEILDOUST, Amir SABBAGH MOLAHOSSEINI
    Type: PAPER
    Subject area: Computer System
    2011 Volume E94.D Issue 2 Pages 264-273
    Published: February 01, 2011
    Released: February 01, 2011
    JOURNALS FREE ACCESS
    This paper presents a general architecture for designing efficient reverse converters based on the moduli set {2α, 22β+1-1, 2β-1}, where β ≤ α ≤ 2β, by using a parallel implementation of mixed-radix conversion (MRC) algorithm. The moduli set {2α, 22β+1-1, 2β-1} is free from modulo (2k+1)-type which can result in an efficient arithmetic unit for residue number system (RNS). The values of α and β can be selected to provide the required dynamic range (DR) and also to adjust the desired equilibrium between moduli bit-width. The simple multiplicative inverses of the proposed moduli set and also using novel techniques to simplify conversion equations lead to a low-complexity and high-performance general reverse converter architecture that can be used to support different DRs. Moreover, due to the current importance of the 5n-bit DR moduli sets, we also introduced the moduli set {22n, 22n+1-1, 2n-1} which is a special case of the general set {2α, 22β+1-1, 2β-1}, where α=2n and β=n. The converter for this special set is derived from the presented general architecture with higher speed than the fastest state-of-the-art reverse converter which has been designed for the 5n-bit DR moduli set {22n, 22n+1-1, 2n-1}. Furthermore, theoretical and FPGA implementation results show that the proposed reverse converter for moduli set {22n, 22n+1-1, 2n-1} results in considerable improvement in conversion delay with less hardware requirements compared to other works with similar DR.
    Download PDF (522K)
  • Ning DENG, Weixing JI, Jiaxin LI, Qi ZUO, Feng SHI
    Type: PAPER
    Subject area: Computer System
    2011 Volume E94.D Issue 2 Pages 274-285
    Published: February 01, 2011
    Released: February 01, 2011
    JOURNALS FREE ACCESS
    Many state-of-the-art embedded systems adopt scratch-pad memory (SPM) as the main on-chip memory due to its advantages in terms of energy consumption and on-chip area. The cache is automatically managed by the hardware, while SPM is generally manipulated by the software. Traditional compiler-based SPM allocation methods commonly use static analysis and profiling knowledge to identify the frequently used data during runtime. The data transfer is determined at the compiling stage. However, these methods are fragile when the access pattern is unpredictable at compile time. Also, as embedded devices diversify, we expect a novel SPM management that can support embedded application portability over platforms. This paper proposes a novel runtime SPM management method based on the core working set (CWS) theory. A counting-based CWS identification algorithm is adopted to heuristically determine those data blocks in the program's working set with high reference frequency, and then these promising blocks are allocated to SPM. The novelty of this SPM management method lies in its dependence on the program's dynamic access pattern as the main cue to conduct SPM allocation at runtime, thus offloading SPM management from the compiler. Furthermore, the proposed method needs the assistance of MMU to complete address redirection after data transfers. We evaluate the new approach by comparing it with the cache system and a classical profiling-driven method, and the results indicate that the CWS-based SPM management method can achieve a considerable energy reduction compared with the two reference systems without notable degradation on performance.
    Download PDF (1437K)
  • Kazuhiro YOSHIMURA, Takuya IWAKAMI, Takashi NAKADA, Jun YAO, Hajime SH ...
    Type: PAPER
    Subject area: Computer System
    2011 Volume E94.D Issue 2 Pages 286-297
    Published: February 01, 2011
    Released: February 01, 2011
    JOURNALS FREE ACCESS
    Recently, we have proposed using a Linear Array Pipeline Processor (LAPP) to improve energy efficiency for various workloads such as image processing and to maintain programmability by working on VLIW codes. In this paper, we proposed an instruction mapping scheme for LAPP to fully exploit the array execution of functional units (FUs) and bypass networks by a mapper to fit the VLIW codes onto the FUs. The mapping can be finished within multi-cycles during a data prefetch before the array execution of FUs. According to an HDL based implementation, the hardware required for mapping scheme is 84% of the cost introduced by a baseline method. In addition, the proposed mapper can further help to shrink the size of array stage, as our results show that their combination becomes 88% of the baseline model in area.
    Download PDF (3724K)
  • Lei WANG, Yueming CAI, Weiwei YANG
    Type: PAPER
    Subject area: Information Network
    2011 Volume E94.D Issue 2 Pages 298-307
    Published: February 01, 2011
    Released: February 01, 2011
    JOURNALS FREE ACCESS
    In this paper, we analyze the impact of channel estimation errors for both decode-and-forward (DF) and amplify-and-forward (AF) cooperative communication systems over Nakagami-m fading channels. Firstly, we derive the exact one-integral and the approximate expressions of the symbol error rate (SER) for DF and AF relay systems with different modulations. We also present expressions showing the limitations of SER under channel estimation errors. Secondly, in order to quantify the impact of channel estimation errors, the average signal-to-noise-ratio (SNR) gap ratio is investigated for the two types of cooperative communication systems. Numerical results confirm that our theoretical analysis for SER is very efficient and accurate. Comparison of the average SNR gap ratio shows that DF model is less susceptible to channel estimation errors than AF model.
    Download PDF (452K)
  • Dean LUO, Yu QIAO, Nobuaki MINEMATSU, Keikichi HIROSE
    Type: PAPER
    Subject area: Educational Technology
    2011 Volume E94.D Issue 2 Pages 308-316
    Published: February 01, 2011
    Released: February 01, 2011
    JOURNALS FREE ACCESS
    This study focuses on speaker adaptation techniques for Computer-Assisted Language Learning (CALL). We first investigate the effects and problems of Maximum Likelihood Linear Regression (MLLR) speaker adaptation when used in pronunciation evaluation. Automatic scoring and error detection experiments are conducted on two publicly available databases of Japanese learners' English pronunciation. As we expected, over-adaptation causes misjudgment of pronunciation accuracy. Following the analysis, we propose a novel method, Regularized Maximum Likelihood Regression (Regularized-MLLR) adaptation, to solve the problem of the adverse effects of MLLR adaptation. This method uses a group of teachers' data to regularize learners' transformation matrices so that erroneous pronunciations will not be erroneously transformed as correct ones. We implement this idea in two ways: one is using the average of the teachers' transformation matrices as a constraint to MLLR, and the other is using linear combinations of the teachers' matrices to represent learners' transformations. Experimental results show that the proposed methods can better utilize MLLR adaptation and avoid over-adaptation.
    Download PDF (572K)
  • Masako OMACHI, Shinichiro OMACHI
    Type: PAPER
    Subject area: Pattern Recognition
    2011 Volume E94.D Issue 2 Pages 317-324
    Published: February 01, 2011
    Released: February 01, 2011
    JOURNALS FREE ACCESS
    Precise estimation of data distribution with a small number of sample patterns is an important and challenging problem in the field of statistical pattern recognition. In this paper, we propose a novel method for estimating multimodal data distribution based on the Gaussian mixture model. In the proposed method, multiple random vectors are generated after classifying the elements of the feature vector into subsets so that there is no correlation between any pair of subsets. The Gaussian mixture model for each subset is then constructed independently. As a result, the constructed model is represented as the product of the Gaussian mixture models of marginal distributions. To make the classification of the elements effective, a graph cut technique is used for rearranging the elements of the feature vectors to gather elements with a high correlation into the same subset. The proposed method is applied to a character recognition problem that requires high-dimensional feature vectors. Experiments with a public handwritten digit database show that the proposed method improves the accuracy of classification. In addition, the effect of classifying the elements of the feature vectors is shown by visualizing the distribution.
    Download PDF (939K)
  • Ayaka YAMAMOTO, Yoshio IWAI, Hiroshi ISHIGURO
    Type: PAPER
    Subject area: Image Recognition, Computer Vision
    2011 Volume E94.D Issue 2 Pages 325-335
    Published: February 01, 2011
    Released: February 01, 2011
    JOURNALS FREE ACCESS
    Background subtraction is widely used in detecting moving objects; however, changing illumination conditions, color similarity, and real-time performance remain important problems. In this paper, we introduce a sequential method for adaptively estimating background components using Kalman filters, and a novel method for detecting objects using margined sign correlation (MSC). By applying MSC to our adaptive background model, the proposed system can perform object detection robustly and accurately. The proposed method is suitable for implementation on a graphics processing unit (GPU) and as such, the system realizes real-time performance efficiently. Experimental results demonstrate the performance of the proposed system.
    Download PDF (2661K)
  • Vinh Ninh DAO, Masanori SUGIMOTO
    Type: PAPER
    Subject area: Image Recognition, Computer Vision
    2011 Volume E94.D Issue 2 Pages 336-348
    Published: February 01, 2011
    Released: February 01, 2011
    JOURNALS FREE ACCESS
    This paper describes a technique for reconstructing dynamic scene geometry using a handheld video projector-camera system and a single checkerboard image as a structured light pattern. The proposed technique automatically recognizes a dense checkerboard pattern under dynamic conditions. The pattern-recognition process is adaptive to different light conditions and an object's color, thereby avoiding the need to set threshold values manually for different objects when the scanning device is moving. We also propose a technique to find corresponding positions for the checkerboard pattern, when displayed by a projector, without needing any position-encoding techniques. The correspondence matching process is based on epipolar geometry, enabling the checkerboard pattern to be matched even if parts of it are occluded. By using a dense checkerboard pattern, we can construct a handheld projector-camera system that can acquire the geometry of objects in real time, and we have verified the feasibility of the proposed techniques.
    Download PDF (1505K)
  • Somying THAINIMIT, Chirayuth SREECHOLPECH, Vuttipong AREEKUL, Chee-Hun ...
    Type: PAPER
    Subject area: Image Recognition, Computer Vision
    2011 Volume E94.D Issue 2 Pages 349-356
    Published: February 01, 2011
    Released: February 01, 2011
    JOURNALS FREE ACCESS
    Iris recognition is an important biometric method for personal identification. The accuracy of an iris recognition system highly depends on the success of an iris segmentation step. In this paper, a robust and accurate iris segmentation algorithm for closed-up NIR eye images is developed. The proposed method addressed problems of different characteristics of iris databases using local image properties. A precise pupil boundary is located with an adaptive thresholding combined with a gradient-based refinement approach. A new criteria, called a local signal-to-noise ratio (LSNR) of an edge map of an eye image is proposed for localization of the iris's outer boundary. The boundary is modeled with a weighted circular integral of LSNR optimization technique. The proposed method is experimented with multiple iris databases. The obtained results demonstrated that the proposed iris segmentation method is robust and desirable. The proposed method accurately segments iris region, excluding eyelids, eyelashes and light reflections against multiple iris databases without parameter tunings. The proposed iris segmentation method reduced false negative rate of the iris recognition system by half, compared to results obtained using Masek's method.
    Download PDF (768K)
  • Kwang-Wook LEE, Seung-Won JUNG, Seung-Kyun KIM, Sung-Jea KO
    Type: PAPER
    Subject area: Computer Graphics
    2011 Volume E94.D Issue 2 Pages 357-362
    Published: February 01, 2011
    Released: February 01, 2011
    JOURNALS FREE ACCESS
    The panorama image obtained by image stitching can have visible artifacts due to the limitation of alignment accuracy and defects of the optical systems. Moreover, conventional image stitching algorithms cannot be directly applied to a real-time video stitching due to its complexity and waving artifacts. In this paper, we propose a real-time content-aware stitching algorithm which not only finds a seam by using path searching based on the greedy method, but also adaptively updates the seam by detecting objects moving toward the seam. Experimental results show that the proposed algorithm can successfully produce stitched video sequences without the waving and ghost artifacts commonly found in conventional stitching algorithms.
    Download PDF (708K)
  • Sihyoung LEE, Sunil CHO, Yong Man RO
    Type: PAPER
    Subject area: Biological Engineering
    2011 Volume E94.D Issue 2 Pages 363-370
    Published: February 01, 2011
    Released: February 01, 2011
    JOURNALS FREE ACCESS
    The active shape model (ASM) has been widely adopted by automated bone segmentation approaches for radiographic images. In radiographic images of the distal radius, multiple edges are often observed in the near vicinity of the bone, typically caused by the presence of thin soft tissue. The presence of multiple edges decreases the segmentation accuracy when segmenting the distal radius using ASM. In this paper, we propose an enhanced distal radius segmentation method that makes use of a modified version of ASM, reducing the number of segmentation errors. To mitigate segmentation errors, the proposed method emphasizes the presence of the bone edge and downplays the presence of a soft tissue edge by making use of Dual energy X-ray absorptiometry (DXA). To verify the effectiveness of the proposed segmentation method, experiments were performed with 30 distal radius patient images. For the images used, compared to ASM-based segmentation, the proposed method improves the segmentation accuracy with 47.4% (from 0.974mm to 0.512mm).
    Download PDF (966K)
  • Hiroyuki GOTO
    Type: LETTER
    Subject area: Fundamentals of Information Systems
    2011 Volume E94.D Issue 2 Pages 371-374
    Published: February 01, 2011
    Released: February 01, 2011
    JOURNALS FREE ACCESS
    This research aims to accelerate the computation module in max-plus algebra using CUDA technology on graphics processing units (GPUs) designed for high-performance computing. Our target is the Kleene star of a weighted adjacency matrix for directed acyclic graphs (DAGs). Using a inexpensive GPU card for our experiments, we obtained more than a 16-fold speedup compared with an Athlon 64 X2.
    Download PDF (400K)
  • Hangkyu KIM, Chang-Sup PARK, Yoon Joon LEE
    Type: LETTER
    Subject area: Artificial Intelligence, Data Mining
    2011 Volume E94.D Issue 2 Pages 375-378
    Published: February 01, 2011
    Released: February 01, 2011
    JOURNALS FREE ACCESS
    Semantic search can be divided into three steps. Keyword matching, the first step, significantly impacts the search results, since the following steps are based on it. In this paper, we propose a keyword matching methodology that aggregates relevance scores of the related text to define the score of an object. Validity of the approach is shown by experiments performed with three public data sets and the detailed analysis of the results.
    Download PDF (513K)
  • Lihua GUO, Lianwen JIN
    Type: LETTER
    Subject area: Pattern Recognition
    2011 Volume E94.D Issue 2 Pages 379-383
    Published: February 01, 2011
    Released: February 01, 2011
    JOURNALS FREE ACCESS
    The Laplacian support vector machine (LSVM) is a semi-supervised framework that uses manifold regularization for learning from labeled and unlabeled data. However, the optimal kernel parameters of LSVM are difficult to obtain. In this paper, we propose a multi-kernel LSVM (MK-LSVM) method using multi-kernel learning formulations in combination with the LSVM. Our learning formulations assume that a set of base kernels are grouped, and employ l2 norm regularization for automatically seeking the optimal linear combination of base kernels. Experimental testing reveals that our method achieves better performance than the LSVM alone using synthetic data, the UCI Machine Learning Repository, and the Caltech database of Generic Object Classification.
    Download PDF (719K)
  • Chaoke PEI, Li GAO, Donghui WANG, Chaohuan HOU
    Type: LETTER
    Subject area: Image Processing and Video Processing
    2011 Volume E94.D Issue 2 Pages 384-387
    Published: February 01, 2011
    Released: February 01, 2011
    JOURNALS FREE ACCESS
    The H.264/AVC standard achieves significantly high coding efficiency if multiple block size Motion Estimation is adopted. However, the complexity of Motion Estimation and DCT is dramatically increased as a result. In previous work we propose an early mode decision algorithm to control the complexity, based on all-zero-blocks detection in 16×16 size. In this paper, we improve the algorithm. Firstly, we propose to detect all-zero blocks in 16×16, 8×8 and 4×4 sizes to simplify the course of mode decision. Secondly, we define the thresholds which are used to terminate motion estimation and mode decision in advance for these sizes. Last, we present the whole proposed algorithm. Experiments show that about 77% encoding time and 85% motion estimation time can be saved on average, which is better than state-of-the-art approaches.
    Download PDF (177K)
  • Jonghyun PARK, Wanhyun CHO, Gueesang LEE, Soonyoung PARK
    Type: LETTER
    Subject area: Image Recognition, Computer Vision
    2011 Volume E94.D Issue 2 Pages 388-391
    Published: February 01, 2011
    Released: February 01, 2011
    JOURNALS FREE ACCESS
    This paper proposes a novel image segmentation method based on Clausius entropy and adaptive Gaussian mixture model for detecting moving objects in a complex environment. The results suggest that the proposed method performs better than existing methods in extracting the foreground in various video sequences composed of multiple objects, lighting reflections, and background clutter.
    Download PDF (150K)
  • Kazuya UEKI, Masashi SUGIYAMA, Yasuyuki IHARA
    Type: LETTER
    Subject area: Image Recognition, Computer Vision
    2011 Volume E94.D Issue 2 Pages 392-395
    Published: February 01, 2011
    Released: February 01, 2011
    JOURNALS FREE ACCESS
    Over the recent years, a great deal of effort has been made to estimate age from face images. It has been reported that age can be accurately estimated under controlled environment such as frontal faces, no expression, and static lighting conditions. However, it is not straightforward to achieve the same accuracy level in a real-world environment due to considerable variations in camera settings, facial poses, and illumination conditions. In this paper, we apply a recently proposed machine learning technique called covariate shift adaptationto alleviating lighting condition change between laboratory and practical environment. Through real-world age estimation experiments, we demonstrate the usefulness of our proposed method.
    Download PDF (160K)
  • Lv GUO, Yin LI, Jie YANG, Li LU
    Type: LETTER
    Subject area: Biological Engineering
    2011 Volume E94.D Issue 2 Pages 396-399
    Published: February 01, 2011
    Released: February 01, 2011
    JOURNALS FREE ACCESS
    Surface integrity of 3D medical data is crucial for surgery simulation or virtual diagnoses. However, undesirable holes often exist due to external damage on bodies or accessibility limitation on scanners. To bridge the gap, hole-filling for medical imaging is a popular research topic in recent years[1]-[3]. Considering that a medical image, e.g. CT or MRI, has the natural form of a tensor, we recognize the problem of medical hole-filling as the extension of Principal Component Pursuit (PCP) problem from matrix case to tensor case. Since the new problem in the tensor case is much more difficult than the matrix case, an efficient algorithm for the extension is presented by relaxation technique. The most significant feature of our algorithm is that unlike traditional methods which follow a strictly local approach, our method fixes the hole by the global structure in the specific medical data. Another important difference from the previous algorithm [4] is that our algorithm is able to automatically separate the completed data from the hole in an implicit manner. Our experiments demonstrate that the proposed method can lead to satisfactory results.
    Download PDF (155K)
feedback
Top