本記事に「抄録」はありません。

抄録全体を表示

In this paper, the authors describe multi-dimensional sensing that can be adopted to realize security in everyday life. In this study, the applications of sensing based on the use of surveillance cameras for equipment mounted on cars, road transport, fields such as farming, and health maintenance have been investigated. Recently developed systems and methods involving multi-dimensional sensing technologies are introduced, and current issues and trends are described.

抄録全体を表示

Multi-dimensional sensing has been used for various purposes in the field of production systems. The members of the IEEJ MDS committee investigated the trends in sensing technologies and their applications. In this paper, the result of investigations of auto-guided vehicles, cell manufacturing robots, safety, maintenance, worker monitoring, and sensor networks are discussed.

抄録全体を表示

Multidimensional sensing (MDS) technologies have numerous applications in the field of digital media, including the development of audio and visual equipment for human-computer interaction (HCI) and manufacture of data storage devices; furthermore, MDS finds applications in the fields of medicine and marketing, i.e., in e-marketing and the development of diagnosis equipment.

抄録全体を表示

This paper reports a new method for visual tracking of humans using active RFID technology. Previous studies were based on the assumption that the radio intensity from an RFID tag will be linearly proportional to the distance between the tag and the antenna or will remain unchanged; however, in reality, the intensity fluctuates significantly and changes drastically with a small change in the environment. The proposed method helps to overcome this problem by using only accurate binary information that reveals whether the target person is close to the antenna. Several experimental results have shown that the information from the RFID tag was useful for reliable tracking of humans.

抄録全体を表示

Kyushu Electric Power Co., Inc. collects different types of sensor data and weather information to maintain the safety of hydroelectric power plants while the plants are in operation. We have to identify malfunction signs from among the collected sensor data. In this paper, we describe a method for identifying the conditions that could cause a malfunction; our method consists of two identification stages. In the first stage, we identify malfunction signs, which are different from normal-condition data, and in the second stage, we monitor aging degradation. Our proposed method is based on the use of a one-class support vector machine and a normal support vector machine. The experimental results obtained in this study show that our proposed method can be employed to identify malfunction signs, which are different from normal-condition data, and to monitor aging degradation.

抄録全体を表示

In this paper, we propose a system that can be used to simplify touch-panel operation for people with visual disability. In the system, a user specifies the target button on the panel by a verbal input. The system detects the button and user's fingertip by analyzing images obtained through a stereo camera. The system helps in guiding a finger to the panel by indicating the direction of motion of the fingertip relative to the panel through sounds generated in a headphone. To devise an efficient navigation method, ways to indicate the direction of motion of the finger, different types of navigation sounds, and ways to indicate the distance between the finger and the panel were considered. The experiment showed the effectiveness of the proposed system and the average navigation time for the most efficient method was 9.7s.

抄録全体を表示

In this study, we propose a method for detecting road markings recorded in an image captured by an in-vehicle camera by using a position-dependent classifier. Road markings are symbols painted on the road surface that help in preventing traffic accidents and in ensuring traffic smooth. Therefore, driver support systems for detecting road markings, such as a system that provides warning in the case when traffic signs are overlooked, and supporting the stopping of a vehicle are required. It is difficult to detect road markings because their appearance changes with the actual traffic conditions, e. g. the shape and resolution change. The variation in these appearances depend on the positional relation between the vehicle and the road markings, and on the vehicle posture. Although these variations are quite large in an entire image, they are relatively small in a local area of the image. Therefore, we try to improve the detection performance by taking into account the local variations in these appearances. We propose a method in which a position-dependent classifier is used to detect road markings recorded in images captured by an in-vehicle camera. Further, to train the classifier efficiently, we propose a generative learning method that takes into consideration the positional relation between the vehicle and road markings, and also the vehicle posture. Experimental results showed that the detection performance when the proposed method was used was better than when a method involving a single classifier was used.

抄録全体を表示

In this paper, we propose a part-based human detection method in which the faces and upper parts of human bodies are first detected separately and then merged. In a crowded scene, it is very difficult to merge the detected faces and upper body parts accurately. We solve the problem by using the depth information obtained from the images. We use a TOF (Time of Flight) camera to capture amplitude and depth images. The detector detects the faces and upper body parts from the amplitude images and merges the parts by using a modified Mean Shift clustering algorithm with depth information. Experimental results show that the proposed method shows good performance in the case of crowded scenes. The proposed method is also better than the human detection method in which depth images are used.

抄録全体を表示

A camera self-calibration method based on the results of human detection is proposed. This method extracts the positions and heights of people in the target scene from the results of human detection and human-area segmentation, and it estimates camera parameters such as the location of the camera in the world coordinates and a vanishing line in the image coordinates. Calibrating a camera generally requires intensive effort, but the proposed method can perform self-calibration using parameters that are automatically extracted from the target image. As a result, our method can estimate the three-dimensional position of an object even when a camera that has not been previously calibrated is used. Experimental results show that the accuracy of the estimated camera parameters can be improved by using the results of human-area segmentation.

抄録全体を表示

A wearable walking navigation system without any special infrastructures has been developed to guide visually impaired. It is important to estimate a position correctly so that safe navigation can be realized. In our system, different sensor data are fused to estimate a pedestrian's position. An image processing system and a laser range finder were used to estimate the positions indoors. In this paper, we introduce the concept of “similarity” between map information and sensor data. This similarity is used to estimate the positions. Experimental results show that highly accurate position estimation can be achieved by sensor fusion. The positions in a linear passage were estimated using image processing data, and when the passage turns, the positions were estimated using LRF data.

抄録全体を表示

Color perception varies from individual to individual. It is difficult for a person with defective cones in the retina to recognize the difference between specific colors. We propose a method for presenting color information by using a projector camera system. The system projects border lines or color names on real object surfaces for specific color combinations. Registration of the projected images and real objects is executed by using projected markers. The effectiveness of the proposed method is verified through experiments.

抄録全体を表示

This paper describes a method for model-based 3D pose estimation. The object model includes a triangular surface mesh, model points, and model geometrical features. Model points and model geometrical features are generated using contour generators, which are estimated by the occluding contours of the images of the triangular surface mesh projected along multiple viewing directions, and are stored depending on the viewing direction. This modeling allows easy parallel processing. Multiple hypotheses for an approximate model pose are generated by comparing the model geometrical features and geometrical features extracted from pose data. These hypotheses are limited by the viewing directions used to generate model geometrical features. Each hypothesis is verified and improved by using model points and 3D boundaries, which are reconstructed by segment-based stereovision. In addition, each hypothesis is improved by using the triangular surface mesh and 3D boundaries. Experimental results obtained for objects with various shapes show the effectiveness of the proposed method.

抄録全体を表示

In this paper, we refer to images obtained by mapping dots observed during defect inspections as secondary images. We propose a method of extracting information for specifying the causes of distinctive structural patterns that appear in the secondary images. We suggest the use of a light process for conducting real-time inspection and the feature extraction process simultaneously. The patterns that appear in the secondary images are lines and random dots. In this study, we focus on the recognition of vertical or horizontal line patterns. Vertical or horizontal line patterns are characteristic of secondary images and appear frequently in these images. In parametric space, vertical and horizontal lines are constrained in a specific area. Since our method involves the use of constraints, it performs fast characteristic extractions. Our method is robust to the variation of the average dot densities of the secondary image and helps to realize low-cost recognition.

抄録全体を表示

This study aims to establish an error model for a stereo measurement system by considering camera vibration. In stereo measurements, we can approximate the measurement error distribution in images as a uniform distribution. Therefore, the parameters of the measurement error distribution are the mean value µ and the standard deviation σ, while the amplitude A and the frequency F are regarded as the parameters of the camera vibration. In order to verify the relationships between the parameters of the measurement error distribution and those of the camera vibration, we performed an experiment using a vibration-testing system. The results of the experiment showed that the vibration did not affect µ. Moreover, we found a positive correlation between σ and A as well as a negative correlation betweenσ and F. Using these relationships, we estimated the parameters of the measurement error from the parameters of the camera vibration.

抄録全体を表示

We propose a method for high-speed image matching, a small number of pixels representing a statistical subset of a template image are used in this method. Generally, when a small number of template pixels are taken into consideration, high-speed matching can be achieved. However, it is often difficult to eliminate the trade-off between speed and reliability in image matching. In order to achieve reliable matching, unique and significant pixels with specific locations and intensities should be selected. For this purpose, we analyze a co-occurrence histogram of multiple local pixels because it provides beneficial information about the probability of simultaneous occurrence. Using the proposed method, we preferentially select pixels with low co-occurrence probability as appropriate template pixels. We also propose a method for determining the approximate probability of co-occurrence of pixels by using some two-dimensional co-occurrence histograms; considerable memory space can be saved when these two-dimensional histograms are used. On the basis of the results of experiments conducted using more than 480 test images, it has been proved that approximately 0.4∼1% of template pixels selected by the proposed method are required to achieve a reasonable object-detection success rate.

抄録全体を表示

In this paper, we propose an effective measure to compute “rareness” of pixel blocks by comparing with other blocks and also introduce a region segmentation method to extract significant blocks such as defects in a visual inspection system. Our basic idea is based on the simple and realistic assumption that the target blocks that are to be detected must have statistical characteristics different from the other blocks. In order to calculate the measure, we utilize histograms of three features: luminance, gradient, and gradient direction. Some blocks that have extraordinary characteristics can automatically be detected by estimating the relative distances between all blocks. Results of experiments involving the use of actual images have proved that our method works well for extracting a rare defect region without a priori knowledge about the defects.

抄録全体を表示

In material engineering, it is widely recognized that deformation and fracture (D/F) characteristics are important because the safety of developed materials can be determined on the basis of D/F characteristics. The D/F characteristic is defined as the load required to break the material and the strain caused by applying a particular load. To observe the effect of grain boundary slip at the micron level, we have proposed a method of obtaining displacement vectors of internal structures from submicron 3D CT images. In this paper, we introduce an improved method for accurately acquiring D/F characteristics. The results of a simulation and a real test confirm the effectiveness of the improved method.

抄録全体を表示

In this paper, we propose a method for producing an informative wide-range composite image of a target object from a video taken by a camera moving parallel to the ground plane. Generally, when one generates an image that presents the scene in a video, it is difficult to select the information to be shown in the generated image from the large amount of information contained in the video. Our method involves the use of trajectories of feature points for selecting informative regions and generates a composite image on the basis of the trajectories. First, we create a 3D volume by piling up frame images on a time line. Second, we set a curved surface in that volume. Finally, a wide-range image is formed by considering the pixels along the curved surface. Our method selects a curved surface that passes through a large number of feature points of the target object. Therefore, we obtain an informative wide-range composite image of the target object from a video. By performing experiments with some actual videos, we verified that our method can generate an informative wide-range image of the target object effectively.

抄録全体を表示

This paper presents a system based on the control of PTZ cameras for automated real-time tracking of individual figure skaters moving on an ice rink. In the video images of figure skating, irregular trajectories, various postures, rapid movements, and various costume colors are included. Therefore, it is difficult to determine some features useful for image tracking. On the other hand, an ice rink has a limited area and uniform high intensity, and skating is always performed on ice. In the proposed system, an ice rink region is first extracted from a video image by the region growing method, and then, a skater region is extracted using the rink shape information. In the camera control process, each camera is automatically panned and/or tilted so that the skater region is as close to the center of the image as possible; further, the camera is zoomed to maintain the skater image at an appropriate scale. The results of experiments performed for 10 training scenes show that the skater extraction rate is approximately 98%. Thus, it was concluded that tracking with camera control was successful for almost all the cases considered in the study.

抄録全体を表示

This paper proposes a method for estimating the 3D shape, color, and specular reflections of an object on the basis of images recorded with an image scanner equipped with multiple light sources. The proposed method is based on the photometric stereo method and it involves several steps. First, we carry out height estimation using four scanned images. Using linear light sources, we can determine the height between the scanning plane and the object surface, regardless of the amount of specular reflection. Next, we estimate the surface normal vectors and object color using the scanned images recorded by light sources located along a circular path. Then, we estimate the specular components using the same scanned images. We perform experiments for object reconstruction using synthetic images and for identifying the optimal arrangement of the light sources for the estimation. Finally, we confirm that the estimated color and specular components have some errors, but the object shape can be estimated accurately.

抄録全体を表示

In this study, we focused on the basic taste stimulation for the analysis of real facial expressions. We considered that the expressions caused by taste stimulation were unaffected by individuality or emotion, that is, such expressions were involuntary. We analyzed the movement of facial muscles by taste stimulation and compared real expressions with artificial expressions. From the result, we identified an obvious difference between real and artificial expressions. Thus, our method would be a new approach for facial expression recognition.

抄録全体を表示

In this paper, we propose a mouth extraction method based on an integrated combination of color difference features, texture information characterized by the CFI (Co-occurrence Frequency Image) feature of the pixels, and the topological features of the extracted mouth region. The negative known that the ill effects caused by the presence of a mustache can be successfully suppressed by using the proposed method.
Mouth extraction is performed by using the CIE L*a*b* color space and its behavior was investigated. The system brings L* of CIE L*a*b* to the same value in all the pixels. The presence of a mustache and/or wrinkles around the mouth affects the extraction. Mustache color and skin color are closer to the gray scale than mouth color. If the system brings L* to the same value in all the pixels, mouth extraction would be unaffected by the presence of a shadow or mustache. Moreover, the weight of a^* of CIE L*a*b* could provide a strong affect for emphasizing the difference between mouth and the surrounding skin region.
In the proposed method, the input image is binarized by referring to the CFI of the converted image. When the boundary between the mouth and the skin is not clear, CFI, the frequency of occurrence of a pixel pair of the image, is referred to detect such a boundary, because the frequency at the boundary is always low. We present the experimental results obtained and show that proposed method can be successfully used for mouth extraction.

抄録全体を表示

In this paper, we propose a method for estimating the JPEG blockiness in a given image. The most advanced method in this research area is the method that is based on the analysis of gradient information obtained for three types of image regions; the region inside of the block, the region at the boundary of the block, and the region over entire image. The method can be used to detect the blockiness and to calculate the image quality measure accurately. However, the correlation between the calculated measure and the subjective estimation by humans is not very high. Therefore, we propose a method for estimating the blockiness in the image by examining the relationship between the frequency structures of the pixels inside and on the boundary of the JPEG block. For analyzing the frequency, we used the co-occurrence histogram because it provides some useful information on the frequency structure. Experimental results have proved that the coefficient obtained in correlation of our method is greater than that obtained in the conventional method and a subjective evaluation by approximately 0.04.

抄録全体を表示

Movies can be used to analyze a player's performance and improve his/her skills. In the case of baseball, the pitching is recorded by using a high-speed camera, and the recorded images are used to improve the pitching skills of the players. In this paper, we present a method for estimating of the rotational velocity of a baseball on the basis of movies recorded by high-speed cameras. Unlike in the previous methods, we consider the original seam pattern of the ball seen in the input movie and identify the corresponding image from a database of images by adopting the parametric eigenspace method. These database images are CG Images. The ball's posture can be determined on the basis of the rotational parameters. In the proposed method, the symmetric property of the ball is also taken into consideration, and the time continuity is used to determine the ball's posture. In the experiments, we use the proposed method to estimate the rotational velocity of a baseball on the basis of real movies and movies consisting of CG images of the baseball. The results of both the experiments prove that our method can be used to estimate the ball's rotation accurately.

抄録全体を表示

Direct yaw-moment control (DYC) is an effective method for achieving stable vehicle motion. In the DYC of vehicles having in-wheel motors (IWMs) and active front and rear steering systems, some of the control inputs are generally redundant. This means that input variables can not be decided uniquely to control each longitudinal, lateral, and yawing motion. The equalization of workload for each wheel on the basis of longitudinal and lateral force distribution enhances the cornering performance of vehicles. Therefore, we have proposed a method for obtaining longitudinal- and latitudinal-force distributions on the basis of the least squares solution of the equations for longitudinal, lateral, and yawing motion. Further, we have proposed a lateral-force control method with tire lateral force sensors and active front/rear steering and a DYC method for reducing the tracking error in this controller. In this paper, we show that the equalization of the workload for each wheel and quick yaw-rate response are achieved by adopting these proposed methods. Simulations and experiments are carried out to confirm the effectiveness of the proposed methods.

抄録全体を表示

This paper presents a new constant V/f control method for induction motors and discusses the speed characteristics for steady-state and transient operations. In the proposed method, vector control is applied, and auto-boost voltage compensation for the voltage drop across the stator impedance and slip frequency compensation for improving speed control accuracy are included. The feasible operating region of the proposed method is discussed by using a vector locus. A linear model is proposed, and the stability of the V/f control system is analyzed by using the root locus to design a control parameter. The effectiveness of proposed V/f control method is demonstrated by simulation and experiment for motoring and regenerating operations.

抄録全体を表示

This paper presents operating characteristics of a diode-clamped linear amplifier (DCLA) in regenerative-operating conditions. The DCLA consists of series-connected MOSFETs, series-connected dc power supplies, and clamping diodes. The circuit configuration is similar to that in a diode-clamped multilevel inverter. The DCLA uses complementary MOSFETs. One of the series-connected MOSFETs is only operates in the active state, while the other MOSFETs are in that on- or off- state. Thus, the DCLA is much smaller in the power loss than a conventional class-B amplifier. In addition, the DCLA enables the regeneration of the electric power from the ac terminals to the dc power supplies, when connected to a regenerative load. The power loss under a regenerative condition is theoretically derived and verified by using the experimental setup of the 12-series DCLA. As a result, the experimental setup have demonstrated that the loss in the DCLA in a regenerative operation is only 8% of the regenerative power.

抄録全体を表示

This paper proposes a new Fourier-type phase estimation method (FPEM) for estimating the rotor phase in the sensorless drives of salient-pole permanent-magnet synchronous motors. The method involves the use of a stator current induced by high-frequency voltage injection. The proposed FPEM has the following characteristics: 1) it is versatile, i.e., it can be applied to almost all voltage injections, 2) it does not require band-pass filters that extract the high-frequency component from the stator current, 3) it could have the potential to enhance the positive correlation between the rotor phase and high-frequency current, 4) it does not require redesigning of the integrated PLL when the amplitude and/or frequency of the injected high-frequency voltage is changed. The usefulness of the proposed method is verified through extensive experiments.

抄録全体を表示

DC machine has been used in steel industries and paper industries, Owing to their advantages for adjustable speed drive. The design objective for DC machine is to increase their output power and decrease their size and weight, though DC machine have more than 100 years of application history. Commutation is one of the most important for phenomenon which determines machine reliability, when the design objective is required in design stage. In other words, The Commutation phenomenon gives the restriction to increase the output power and decrease their size and weight of the DC machine.

抄録全体を表示

Single-phase boost rectifier with a resonant snubber switch cell is studied. The resonant snubber switch cell consists of an only active power switching device and a LC resonant circuit, and can also provide the necessary control over the current. To turn on and off an active power device under the soft switching conditions, the rectifier is operated in the discontinuous conduction mode (DCM). The experimental results under the charge control confirm that the input current can nearly be wave-shaped sinusoidally with a near-unity power factor.

抄録全体を表示

本記事に「抄録」はありません。

抄録全体を表示

Our research focuses on electrical machine design, controls, drives and power generation systems with respect to permanent magnet synchronous machines and reluctance machines. Three faculty members, a support staff and 20 students are with the research group.

抄録全体を表示