Journal of Signal Processing
Online ISSN : 1880-1013
Print ISSN : 1342-6230
ISSN-L : 1342-6230
Volume 18 , Issue 4
Special Issue on Papers Awarded the Student Paper Award at NCSP'14 (Editor-in-Chief: Keikichi Hirose, Editor: Nozomu Hamada, Guest Editor: Hironori Takimoto, Honorary Editor-in-Chief: Takashi Yahagi)
Showing 1-22 articles out of 22 articles from the selected issue
  • Yutaka Izumi, Hiroyuki Asahara, Takuji Kousaka
    2014 Volume 18 Issue 4 Pages 153-156
    Published: July 30, 2014
    Released: July 30, 2014
    JOURNALS FREE ACCESS
    In this paper, we experimentally study border-collision bifurcation, which occurs in the fast-scale dynamics in an interrupted electric circuit. First, we show a circuit model, and explain the behavior of the waveform. Next, we define a discrete map by sampling the data of the waveform. Finally, we discuss bifurcation phenomena in the circuit. In particular, we focus on the waveform behavior in the clock pulse interval, which we call the fast-scale dynamics, and study bordercollision bifurcation experimentally.
    Download PDF (2228K)
  • Takayuki Fujii, Hiroyuki Asahara, Takuji Kousaka
    2014 Volume 18 Issue 4 Pages 157-160
    Published: July 30, 2014
    Released: July 30, 2014
    JOURNALS FREE ACCESS
    We propose a simple stability analysis method for a forced self-excited system with stick-slip vibration. First, we show the physical model. Next, we sample time date in switching events and define the Poincaré map. Then, by using the Jacobian matrix and the characteristic equation, we discuss the stability analysis algorithm. Finally, we numerically verify the effectiveness of our proposed method.
    Download PDF (2056K)
  • Taku Otani, Shota Ueguchi, Ryohei Hori, Masaya Yoshikawa, Takeshi Fuji ...
    2014 Volume 18 Issue 4 Pages 161-164
    Published: July 30, 2014
    Released: July 30, 2014
    JOURNALS FREE ACCESS
    We have been studying the via-programmable structured ASIC architecture "VPEX3 (Via-Programmable logic using EXclusive-or array 3)", which can realize arbitrary logic by customizing only via layers. Our previous research found that VPEX3 cannot achieve the same maximum operation speed as ASIC and other mask-programmable devices. In this paper, we propose and evaluate a new architecture "VPEX3S". In this architecture, the output drivability of LEs is strengthened, and the via resistance between LEs is reduced in order to decrease the critical path delay. From an evaluation using logic synthesis, maximum operating speed of VPEX3S was found to be improved by about 34% compared to VPEX3. On the other hand, the circuit area was estimated to be 50%-60% larger than that of VPEX3 from the logic synthesis result. This is because the LE size increased by a factor of 1.85 because of the increase in transistor gate width and the number of vias. However, after placing and routing, the implementation area was estimated to be comparable or smaller. This is because the number of routing resources determines the area. As a result, the proposed architecture VPEX3S can realize the same speed performance as ASICs, with 3.3-4 times larger implementation area.
    Download PDF (964K)
  • Masafumi Mori, Toshiyuki Itou, Masayuki Ikebe, Tetsuya Asai, Tadahiro ...
    2014 Volume 18 Issue 4 Pages 165-168
    Published: July 30, 2014
    Released: July 30, 2014
    JOURNALS FREE ACCESS
    In this study, we propose an architecture for estimating motion vectors by searching for one neighbor pixel in high-speed images and a machine learning algorithm that uses the estimated motion vectors. In high-speed imaging, the motion of pixels between frames is considerably small. Our architecture estimates motion vectors by assuming that the pixels move less than one pixel between frames. We verified that our method could classify images into two classes, i.e., dangerous (something is approaching) or safe (others), by employing a simple perceptron after extracting the features of the estimated motion vectors using a method based on Poggio's HMAX (Hierarchical Model and X) model. We used the target images captured by an in-vehicle camera for learning and verified that another set of images could be classified using our method. We confirmed that the proposed architecture can estimate motion vectors using a small number of operations and perform classification based on machine learning.
    Download PDF (287K)
  • Yuki Shoji, Yoko Uwate, Yoshifumi Nishio
    2014 Volume 18 Issue 4 Pages 169-172
    Published: July 30, 2014
    Released: July 30, 2014
    JOURNALS FREE ACCESS
    The ring inhibitory neural network is a typical neural system used to generate oscillation with a simple structure. Neurons in the ring inhibitory neural network are connected in one direction. In this study, we propose polygonal structure models of the ring inhibitory neural network that are composed of multiple ring inhibitory neural networks. We confirm that the polygonal structure models based on triquetrous and pentagonal models composed of three and five neurons respectively can oscillate. In our simulations, the polygonal structure models show various synchronization and oscillation phenomena based on their network topology.
    Download PDF (1055K)
  • Kei Chiba, Tsuyoshi Otake, Hisashi Aomori, Nobuaki Takahashi, Mamoru ...
    2014 Volume 18 Issue 4 Pages 173-176
    Published: July 30, 2014
    Released: July 30, 2014
    JOURNALS FREE ACCESS
    Cycle spinning cellular neural networks (CS-CNNs) are artificial neural networks that work effectively to solve large-scale problems. In our previous work, a CS-CNN is applied to enhance the resolution of images with an arbitrary magnification parameter. In this paper, a novel adaptive architecture using a CS-CNN is developed to prevent the unnecessary smoothing of image detail. While a discrete-time cellular neural network (DT-CNN) transforms all pixel values into coefficients to predict the original pixel values using the A-template, the adaptive cycle spinning method with "minmod" functions is applied to estimate the optimal coefficients from individual outputs of the DT-CNN as above. The minmod functions are defined on the basis of the interpolation error theorem. Experimental results indicate that the proposed method produces better results than the conventional image resolution enhancement methods.
    Download PDF (462K)
  • Kai Morimoto, Masahiro Saito, Satoshi Inose, Atsushi Kannari, Tomoya ...
    2014 Volume 18 Issue 4 Pages 177-180
    Published: July 30, 2014
    Released: July 30, 2014
    JOURNALS FREE ACCESS
    The present study improves the nonlinear portfolio model by using principal component analysis. To enhance the portfolio effect of spreading risks efficiently, we aim for lower correlations among each asset movement. For this reason, we apply the principal components of assets to the nonlinear portfolio model, which uses nonlinear prediction to estimate future movements. However, because we are not sure whether these principal components have nonlinearity, we perform Fourier-shuffled surrogate tests on the principal components. Finally, we confirm the efficiency of our nonlinear principal-component portfolio model through some investment simulations with real financial data.
    Download PDF (273K)
  • Kyohei Sumikawa, Hiromasa Habuchi
    2014 Volume 18 Issue 4 Pages 181-184
    Published: July 30, 2014
    Released: July 30, 2014
    JOURNALS FREE ACCESS
    In this paper, the influence of scintillation and background noise on the optical low-density generator matrix code with binary pulse position modulation (LDGM-BPPM) with the unequal transmission power allocation (UTPA) scheme is investigated. In particular, the bit error rate (BER) performance of LDGM-BPPM with the UTPA scheme is evaluated in the presence of avalanche photodiode (APD) noise. Consequently, the following points are clarified. (1) The BER performance increases steadily in proportion to the background noise. (2) The BER of the UTPA scheme deteriorates gradually with increasing variance of scintillation. (3) The optimum value of the power ratio (optimum R) varies with the amount of background noise and scintillation. (4) The quasioptimum R is 3.1. Thus, the UTPA scheme is effective for optical LDGM-BPPM.
    Download PDF (299K)
  • Shinsuke Ohtake, Akiro Shimada, Yuichi Miyaji, Hideyuki Uehara, Takash ...
    2014 Volume 18 Issue 4 Pages 185-188
    Published: July 30, 2014
    Released: July 30, 2014
    JOURNALS FREE ACCESS
    The two-dimensional multihop wireless power transfer (2DM-WPT) system can supply electrical power to a wide range of devices. In this system, location information is needed for high power transfer efficiency. We propose a location estimation method based on a fingerprinting technique for the 2DM-WPT system. The proposed method utilizes the reflection coefficient as a location fingerprint. We conduct an electromagnetic simulation to validate the proposed method. When the same receiver is used in both the learning phase and the estimation phase, the correct estimation rate is 96.05%. When two receivers with different shapes and sizes are used in the learning phase and the estimation phase, the correct estimation rate is 59.78%. However, the location is estimated correctly over the couplers where power can be transferred with high efficiency. These results show the effectiveness of the proposed method when the receiver shapes and sizes are the same in the learning phase and the estimation phase. In contrast, when the receiver shapes and sizes are different in each phase, some locations are estimated incorrectly.
    Download PDF (1110K)
  • Takahiro Nagai, Takashi Yasuno, Akinobu Kuwahara
    2014 Volume 18 Issue 4 Pages 189-192
    Published: July 30, 2014
    Released: July 30, 2014
    JOURNALS FREE ACCESS
    This paper proposes a semi-autonomous remote control system for a mobile robot using the degree of collision danger. To successfully realize obstacle avoidance depending on the current condition, the degree of collision danger between the mobile robot and an obstacle is defined by using simplified fuzzy reasoning and is used to decide the priority of manipulation by an operator or autonomous navigation. Several experimental results using our developed legged-type mobile robot demonstrate the validity of the proposed semiautonomous remote control system.
    Download PDF (1871K)
  • Tomomi Nomura, Yasue Mitsukura
    2014 Volume 18 Issue 4 Pages 193-196
    Published: July 30, 2014
    Released: July 30, 2014
    JOURNALS FREE ACCESS
    In this study, we propose an evaluation approach for web form usability. With the popularization of online activities, a web form is used as a means of intermediation between a company and users. It is a crucial problem to provide a user-friendly web form. Therefore, our aim is to investigate users' feelings and behavior during the completion of a web form in different ways: a questionnaire, an electroencephalogram (EEG), and eye tracking. Moreover, we quantify and visualize the evaluation results in order to easily understand a large amount of information at a glance. As a result, we were able to extract valuable information for web form usability from several angles. Thus, we have developed an overall evaluation approach in order to clearly comprehend web form usability.
    Download PDF (527K)
  • Masaru Ando, Daisuke Morikawa, Masashi Unoki
    2014 Volume 18 Issue 4 Pages 197-200
    Published: July 30, 2014
    Released: July 30, 2014
    JOURNALS FREE ACCESS
    Human beings can localize a target sound by using binaural cues. On the other hand, we can also localize a target sound by using monaural cues. The monaural modulation spectrum (MMS) can be regarded as an important cue of monaural sound localization. A method of estimating direction of arrival (DOA) using a machine learning scheme for classification of MMS patterns has been proposed. However, this method cannot account for the monaural DOA mechanism by using MMS patterns. To further investigate how the MMS plays an important role in monaural sound localization, we aimed to find cues in the human ability formonaural sound localization and propose a method of estimating DOA by using these cues. We investigated how the MMS of observed signals vary with the azimuth. As a result, shapes of the MMS were drawn as arcs with azimuth variations. We then proposed a method of estimating monaural DOA using these results. Simulations were carried out to verify the effectiveness of the proposed method. We found that the proposed method could estimate DOA using the MMS, except with front-back confusion discrimination.
    Download PDF (289K)
  • Akikazu Miyazaki, Shota Morita, Masashi Unoki
    2014 Volume 18 Issue 4 Pages 201-204
    Published: July 30, 2014
    Released: July 30, 2014
    JOURNALS FREE ACCESS
    The authors previously proposed a method for blindly estimating the speech transmission index (STI) based on the concept of the modulation transfer function (MTF). This method, however, over- or under-estimates STIs from observed signals in real environments due to the effect of background noise. In this paper, the proposed method was developed from the previous method to resolve the problem by simultaneously estimating the effects of inverse MTFs in noisy and reverberant environments. Simulations were carried out to verify that the proposed method can correctly estimate STIs in noisy reverberant environments by using noisy reverberant AM signals. The results revealed that the proposed approach could be used to effectively estimate STIs from noisy reverberant signals in various room acoustics.
    Download PDF (203K)
  • Katsuhiko Yamamoto, Zhi Zhu, Masashi Unoki, Naofumi Aoki
    2014 Volume 18 Issue 4 Pages 205-208
    Published: July 30, 2014
    Released: July 30, 2014
    JOURNALS FREE ACCESS
    Speech scrambling methods are widely used for copyright protection and encrypting digital speech signals in order to guarantee the confidentiality of the original signals. They are very important methods for preventing eavesdropping and unauthorized copying. However, it seems to be impossible to completely recover a scrambled speech signal into the original signal. Moreover, nobody can comprehend the partial speech content from speech signals scrambled with these methods. In this paper, we propose a semi-scramble method for speech signals based on phonemic restoration. By using a speech scrambling method based on the random-bit shift of quantization bits, speech signals are converted to scrambled signals in partial intervals. We evaluated the confidentiality and efficiency of the proposed method by using two objective measures, signal-to-error ratio (SER) and perceptual evaluation of speech quality (PESQ). As a result, we confirmed that the proposed method can play a role in copyright protection for an original signal and recover a semi-scrambled speech signal into the original one. Finally, we indicated that the acoustic characteristics of signal semi-scrambled with the proposed method enable the listener to understand the speech information.
    Download PDF (429K)
  • Kento Yoshida, Yasue Mitsukura
    2014 Volume 18 Issue 4 Pages 209-212
    Published: July 30, 2014
    Released: July 30, 2014
    JOURNALS FREE ACCESS
    In this study, we emphasized sound sources using nonnegative matrix factorization (NMF) with the likelihood of the directions obtained from the estimated localizations. NMF is one of the most commonly used techniques for the separation of music signals. The emphasis of sound sources by NMF has several problems: the uncertainty of the decomposition of the bases corresponding to the sources and the necessity to group the obtained bases according to the sources. These problems have to be solved simultaneously to improve the separation performance. In order to simultaneously solve these problems, we proposed a novel NMF incorporating the cost based on the likelihood of the localizations. According to the experimental results, we could improve the separation performance by using our proposed method.
    Download PDF (244K)
  • Toshihide Aketo, Hiroshi Saruwatari, Satoshi Nakamura
    2014 Volume 18 Issue 4 Pages 213-216
    Published: July 30, 2014
    Released: July 30, 2014
    JOURNALS FREE ACCESS
    In this paper, we propose a novel sound field reproduction system that combines an image sensor and a wavefront synthesis method. This system employs the spectral division method using a spatial localization filter and performs listener detection via image sensing, which enables us to reproduce the sound field at a specific listener's position with high accuracy. A simulation experiment and subjective assessments of directional perception and sound quality perception show the proposed method's superiority in terms of directional perception.
    Download PDF (1174K)
  • Tomo Miyauchi, Daichi Kitamura, Hiroshi Saruwatari, Satoshi Nakamura
    2014 Volume 18 Issue 4 Pages 217-220
    Published: July 30, 2014
    Released: July 30, 2014
    JOURNALS FREE ACCESS
    In this paper, we propose a new depth estimation method that utilizes the variation of the direction of arrival (DOA) of individual sound sources. In addition, we propose activationshared nonnegative matrix factorization (NMF) as a feature extraction method. The result of an experiment indicated the efficacy of the proposed method.
    Download PDF (519K)
  • Sho Sawada, Tomohiro Fujita, Kaihei Iwanaga, Takeshi Kumaki, Mamoru Na ...
    2014 Volume 18 Issue 4 Pages 221-224
    Published: July 30, 2014
    Released: July 30, 2014
    JOURNALS FREE ACCESS
    One of the variations of the active-contour method, namely, multiple-contour expansion, is realized on cellular automaton (CA)-based pixel-level snakes (CA-based PLS). As for the PLS algorithm, the active contour steps to next pixel one by one. The processing speed of PLS thus depends on image size. In the case of the multiple-contour-expansion method, the contours are successively deformed, their topology is transformed, and objects are extracted from images. The advantage of this method is that the processing speed does not depend on image size. It was experimentally demonstrated that multiple contour expansion is made possible by the CA-based PLS and that the processing speed of PLS does not depend on image size.
    Download PDF (1134K)
  • Kazumasa Horie, Atsuo Suemitsu, Masahiko Morita
    2014 Volume 18 Issue 4 Pages 225-228
    Published: July 30, 2014
    Released: July 30, 2014
    JOURNALS FREE ACCESS
    During the application of surface electromyograms (EMGs) in human-machine interfaces, direct estimates of multiple hand motion speeds are required to facilitate operations that fully reflect the user's intentions. However, no practical methods are available for this purpose because conventional function approximators cannot learn the complex relationships between the motion speeds and surface EMGs within a practical period of time. By contrast, it has been shown that a selective desensitization neural network (SDNN) can learn complex input-output relationships with low computational costs. In this study, we propose a method for the direct estimation of hand motion speeds from surface EMGs using a SDNN. We estimated the motion speed in practice to assess the efficacy of this proposed method. Our experimental results show that the proposed method can estimate the approximate speeds of six basic motions in real time.
    Download PDF (395K)
  • Sangeun Lee, Keiichi Horio
    2014 Volume 18 Issue 4 Pages 229-232
    Published: July 30, 2014
    Released: July 30, 2014
    JOURNALS FREE ACCESS
    In previous work, we proposed a human tracking algorithm based on the reliable appearance model (RAM). The RAM is a set of discriminative local image descriptors that is selected by a boosting algorithm to identify a target in the initial frame, and is employed as an observation model in a particle filter. As the appearance model of the target in human tracking constantly changes as time passes owing to changes in pose, it is necessary to adaptively update the RAM to improve the tracking accuracy. In this paper, if necessary, an insufficient local image descriptor for robust tracking is updated. In order to classify whether local image descriptors are suitable or not during tracking, a distance histogram corresponding to a local image descriptor is constructed. When the histogram indicates that the local image descriptor is lacking in tracking performance, then it is updated. The experimental results demonstrate that the adaptive appearance model successfully tracks sport players even when their pose often changes.
    Download PDF (1507K)
  • Naoya Tsujita, Hiroshi Igarashi
    2014 Volume 18 Issue 4 Pages 233-236
    Published: July 30, 2014
    Released: July 30, 2014
    JOURNALS FREE ACCESS
    Concern for others (CFO) is an important factor in accomplishing efficient cooperative work by multiple human beings. Quantified CFO for evaluating CFO, named Estimation of the CFO (ECFO), has previously been proposed. In this study, relationships between the cooperative work performance and the ECFO are analyzed using a newly developed virtual ping-pong game. In our previous study, a driving simulator with two operators was developed and employed to investigate cooperative behavior. Furthermore, the ECFO was defined as the difference of the inputs between individual operation and cooperative operation by two operators. However, the task was too complicated to achieve satisfactory prediction accuracy. Therefore, it was difficult to obtain the exact relationship between the ECFO and the performance. For this reason, in this study, a ping-pong game was developed and used for the quantitative evaluation of the ECFO. More accurate prediction of the input in individual operation was obtained using this game.
    Download PDF (513K)
  • Taiyo Mineo, Haruhisa Takahashi
    2014 Volume 18 Issue 4 Pages 237-240
    Published: July 30, 2014
    Released: July 30, 2014
    JOURNALS FREE ACCESS
    Here we focus on website video-tag category clustering using kernel spectral clustering (KSC) [1], which has not been reported so far. KSC is a nonlinear clustering method used for the spectral clustering of data in a kernel feature space. Unfortunately KSC is intrinsically affected by the selection of kernel parameters, and this hinders complete automatic learning. This study proposes a unified learning method which enables automatic KSC optimal kernel parameter tuning based on a graph Laplacian representation of the clustering. This method is well suited to applications that require speedy and fully automatic analysis, since it can automatically find the optimal arameter, and generate the optimally connected components of the graph Laplacian. We demonstrate the application of our method to the above problem and show its excellent performance through computer experiments.
    Download PDF (308K)
feedback
Top