電気学会論文誌C(電子・情報・システム部門誌)
Online ISSN : 1348-8155
Print ISSN : 0385-4221
ISSN-L : 0385-4221
126 巻, 8 号
選択された号の論文の24件中1~24を表示しています
論文
<センシング>
<通信・ネットワーク>
<音声画像処理・認識>
  • Xiaowei Zhang, Jianming Lu, Nuo Zhang, Takashi Yahagi
    2006 年 126 巻 8 号 p. 942-949
    発行日: 2006年
    公開日: 2006/11/01
    ジャーナル フリー
    In this paper we propose a separating system for convolutive nonlinear mixtures, since it is rare to find linear instantaneous mixture in practical problems. The observed mixtures are first transformed into frequency domain and then separated on each frequency bin. After solved permutation problem, estimates are obtained. We show results of simulations with acoustic signals both of instantaneous and convolutive mixtures.
  • 福山 忠男, 荻巣 高博, 金 珍祐, 岡崎 耕三
    2006 年 126 巻 8 号 p. 950-956
    発行日: 2006年
    公開日: 2006/11/01
    ジャーナル フリー
    Although an IGO record is a valuable work, it takes a lot of human laborious work. In this paper, we propose an automatic record system by image processing. First, we control a camera shutter by judging the time of brightness change which is caused by the action of the player's hand, capture the n-th image in order. The contrast among white stone, black stone and board depend on the brightness, which are affected by shadows , etc. To cope with the situation, we use a subtraction image between the board and n-th image. The stones are not always put on the centers of the intersection points and are often drifted when they are touched with another stones or the hand. Light reflects on the surface. The brightness of the image often changes. Therefore, we estimate the intersection points of highlighted and distorted board. The stones recognition are based on template matching using RMS errors and correlations. The results are classified automatically by the application of K-means clustering algorithm. We show the proposed algorithm's effectiveness by experiments.
  • 大島 直也, 齊藤 剛史, 小西 亮介
    2006 年 126 巻 8 号 p. 957-962
    発行日: 2006年
    公開日: 2006/11/01
    ジャーナル フリー
    This paper proposed a method in which location of a person is estimation with one active camera in real-time. The approach taken in this paper is as follows. First, we detect the person region with proposed method which is the region detection method based on short time distribution of the density to avoid the influence of the camera movement. Then we obtain his height and the gravity point. The rotate angle and the distance between the person and the camera are calculated by the information of the height and the gravity point. Experiments were applied with several examples. We obtained that the average error distance is less than 12[cm]. The average error distance is small enough in comparison with shoulder width of the persons. Our method can estimate the location in real-time.
  • Xue Yuan, Jianming Lu, Takashi Yahagi
    2006 年 126 巻 8 号 p. 963-971
    発行日: 2006年
    公開日: 2006/11/01
    ジャーナル フリー
    In this paper, we propose a method of 3D face recognition using parallel pyramid neural networks (NNs). We first compensate for the poses of 3D original facial images using feature points and geometrical measurement. Then, the shape and texture images are extracted from compensated 3D images, respectively. The dimensionalities of the shape and texture images are reduced using PCA, processed by LDA for enhanced generalization. The corresponding reduced shape and texture features are combined to form the integrated features for face recognition. In the second step, a method is proposed for face recognition based on parallel pyramid NNs. Compared with conventional NN, the performance of the proposed parallel pyramid NNs is improved by utilizing lower layer effectively. Experimental results for 60 persons with different poses, facial expression and illumination conditions demonstrate the efficiency of our algorithm. In particular, the proposed method achieves 99.17% recognition accuracy using only 120 features.
  • 鳥居 恵多, 岡本 教佳
    2006 年 126 巻 8 号 p. 972-979
    発行日: 2006年
    公開日: 2006/11/01
    ジャーナル フリー
    In personal authentication systems (PAS) using fingerprints, there is a decline in authentication accuracy because of the dirt on the contact surface with the present input equipment. We have already examined the possibility of noncontact fingerprint authentication to address these problems. In this paper, we propose an improvement for the extraction of a fingerprint and a compensation method for a rotation error for the noncontact fingerprint authentication. In the preprocessing, the compensation of a rotation error from two dimensions to three dimensions. In the proposed rotation compensation technique, the front view and side view of a fingertip are used. Moreover, we examined the cause of decreasing the quality of the principal line in our technique. The extraction error is caused by the change in a red component of the skin caused in blood circulation. Then we propose the technique for correcting an input image by using the approximation image obtained by the wavelet transform. As a result of the experiment, the quality of the output image has improved greatly that the extraction errors of the principal lines are decreased.
  • 芝 浩二郎, 古屋 保, 西 省吾, 森 邦彦
    2006 年 126 巻 8 号 p. 980-989
    発行日: 2006年
    公開日: 2006/11/01
    ジャーナル フリー
    We developed a prototype of the automatic go-record system using some image processing techniques. This system is composed of three units. Their units are a 1.10GHz laptop PC, a TV camera, and a pair of switches which tell the system to set a go-stone on the go-board. In this paper, we describe our implemented image processing techniques which are the method to generate a go-coordinates, the method to detect a location of a go-stone, and so on. As a result of the experiment on the prototype system, we confirmed that the correct go-record was generated in a stable indoor lighting condition.
<ロボティクス>
<ソフトコンピューティング>
  • 増田 和明, 栗原 謙三
    2006 年 126 巻 8 号 p. 997-1008
    発行日: 2006年
    公開日: 2006/11/01
    ジャーナル フリー
    This paper proposes a global optimization method based on thoroughly searching for equilibrium points of gradient-based dynamical systems. Such a method is possible due to the linkage between equilibrium points of nonlinear systems and the outstanding properties of gradient dynamics. As the essence of this study, a general form of computational procedure for efficiently finding equilibrium points of nonlinear dynamical system based on the use of trajectories initiating from already known points into their eigen-directions is provided. Then, optimization is realized by incorporating the procedure to gradient-based models for obtaining various local optima as their stable equilibrium points. Its application to constrained global optimization is also discussed, and the effectiveness of our method is demonstrated through numerical simulations.
  • Shingo Mabu, Hiroyuki Hatakeyama, Moe Thu Thu, Kotaro Hirasawa, Jinglu ...
    2006 年 126 巻 8 号 p. 1009-1015
    発行日: 2006年
    公開日: 2006/11/01
    ジャーナル フリー
    A new graph-based evolutionary algorithm called “Genetic Network Programming, GNP" has been proposed. The solutions of GNP are represented as graph structures, which can improve the expression ability and performance. In addition, GNP with Reinforcement Learning (GNP-RL) has been proposed to search for solutions efficiently. GNP-RL can use current information (state and reward) and change its programs during task execution. Thus, it has an advantage over evolution-based algorithms in case much information can be obtained during task execution. The GNP we proposed in the previous research deals with discrete information, but in this paper, we extend the conventional GNP-RL which can deal with numerical information. The proposed method is applied to the controller of Khepera simulator and its performance is evaluated.
<ソフトウェア・情報処理>
  • 高橋 大樹, 西 佑介, 祇園 智弘, 南 真一, 福永 辰也, 緒形 次郎, 吉江 修
    2006 年 126 巻 8 号 p. 1016-1025
    発行日: 2006年
    公開日: 2006/11/01
    ジャーナル フリー
    Semantic Web is the technology which determines the relevance of data over the Web using meta-data and which enables advanced search of global information. It is now desired to develop and apply this technology to many situations of facility management. In facility management, vocabulary should be unified to share the database of facilities for generating optimal maintenance schedule and so on. Under such situations, ontology databases are usually used to describe composition or hierarchy of facility parts. However, these vocabularies used in databases are not unified even between factories of same company, and this situation causes communication hazard between them. Moreover, concept involved in the hierarchy cannot be corresponded each other. There are some methods to correspond concepts of different hierarchy. But these methods have some defects, because they only attend target hierarchy itself and the number of instances. We propose improved method for corresponding concepts between different concepts' hierarchies, which uses other hierarchies all over the world of Web and the distance of instances to identify their relations. Our method can work even if the sets of instances belonging to the concepts are not identical.
  • 中江 達哉, 森津 俊之, 薦田 憲久
    2006 年 126 巻 8 号 p. 1026-1032
    発行日: 2006年
    公開日: 2006/11/01
    ジャーナル フリー
    We propose a method for designing CDO (Collateralized Debt Obligation) that meets investor needs about attributes of CDO. It is demonstrated that adjusting attributes (that are credit capability and issue amount) of CDO to investors' preferences causes a capital loss risk that the agent takes. We formulate a CDO optimization problem by defining an objective function using the above risk and by setting constraints that arise from investor needs and a risk premium that is paid for the agent. Our prototype experiment, in which fictitious underlying obligations and investor needs are given, verifies that CDOs can be designed without opportunity loss and dead stock loss, and that the capital loss is not more than thousandth part of the amount of annual payment under guarantee for small and midium-sized enterprises by a general credit guarantee institution.
  • 熊谷 正俊, 伏木 匠, 君田 和也, 横田 孝義
    2006 年 126 巻 8 号 p. 1033-1039
    発行日: 2006年
    公開日: 2006/11/01
    ジャーナル フリー
    This paper discusses an extended method of “Feature Space Forecast Method" which we proposed before. When we forecast traffic information, we have to consider various factors such as days, seasons, holidays, and so on. Furthermore, for nation-wide forecast services, the number of road links handled by a forecast model reaches more than 0.1 million. Therefore, in order to provide accurate nation-wide services, a forecast method that can efficiently deal with a large amount of traffic data is required. The proposed method achieves an efficient forecast process with a small forecast model that is one-tenth as large as that of traditional methods, by performing forecasting calculation in the feature space shared by multiple road links.
  • 栗原 謙三, 永井 学, 西内 信之, 増田 和明
    2006 年 126 巻 8 号 p. 1040-1048
    発行日: 2006年
    公開日: 2006/11/01
    ジャーナル フリー
    This paper proposes a new method to plan the project durations that are modeled as stochastic network such as GERT network. Since stochastic networks have a variable time and a branching probability for each arrow, the total duration of the network is also modeled as a variable. In order to complete the project by the desired date with a certain confidence, the variable times or branching probabilities of the network should be designed appropriately. We will propose a planning method for branching probabilities to realize the desired project duration using GA and Monte Carlo simulation.
研究開発レター
<ソフトコンピューティング>
エラータ
feedback
Top