Recently, various kinds of 3D imaging devices such as laser range finder or time-of-flight sensor become popular, so a lot of practical algorithms for recognizing 3D objects using “point cloud data” have been proposed. In this paper, we will introduce typical object recognition approaches and survey various kinds of 3D features proposed by many researchers. As for some important techniques, their principle and characteristics are explained in detail. Also we will mention about the LRF (Local Reference Frame) which is very important factor to realize stable feature description and accurate pose estimation in practical use.
In this paper, we aim to classify two classes in children by using single-channel electroencephalogram (EEG). EEG has been used to define neural patterns and to adjust the wide applicability to a larger population of healthy and diseased users. Specialized EEG devices have recently developed as for compact and portable measurement system using them in the real environment. If there is a multiplex state estimation system with EEG through a specialized EEG device, it would be a powerful tool for neuroscience studies and clinical applications. We firstly focused on the state of concentration; therefore, two kinds of single-channel EEG signals (during meditation and concentration) from 10 children were measured. Recordings were processed to remove artifacts, and then extracted their periodic or non-periodic features by three methods (Fourier transform, wavelet transform, and empirical mode decomposition). Elastic net logistic regression constructed predictive models to classify two classes of the optimized extracted features. A model showed 0.988 area under the receiver operating characteristic curve when wavelet transform was selected as feature extraction method. Our next is to construct a multiplex state estimation system. Finally, we will make portable applications using a specialized EEG device that include the multiplex model and encourage children to develop the child's sense.
Ultrasonography, which is one of the quality inspection methods, is able to inspect inner material in a non-destructive. However, with the conventional method, we can discriminate existence of inner defects of the measurement object, but it is difficult to detect the position of inner defects as three-dimensional information. Thus in this study, we focused ultrasonic holography, which is a method to detect of the sound source position. The purpose of this study is building a system to obtain three-dimensional information inside the material instantly by analyzing ultrasonic holography with the two-dimensional sound distribution which appears on the measurement object surface by penetrating through objects of ultrasound. As problems of ultrasonic holography, there are a long measurement time and appearance of Aliasing Image. Therefore, in our study, we suggested a method to obtain sound distribution by Speckle Interferometry, as method to solve two problems at once.
3D shape measurement systems that use contactless methods are required for quality inspection of metal molds and electronic parts in industrial fields. A fringe projection method with phase-shifting method offers the advantages of high precision and high speed. Recently, since the size of electronic parts has become smaller, the pitch of a grating pattern projected onto a specimen should also become smaller. In this paper, a small pitch fringe projection method using Talbot effect with an SLD (Super luminescent diode) and a method to incline the appearing area of the fringe pattern are proposed. A linear fiber array with four cores is prototyped. A phase-shifting method using the linear fiber array is also proposed. The effectiveness is confirmed with the experimental results of shape measurement using a 3D shape measurement setup built with proposed method.
Output feedback control of performing feedback control directly without using an observer by the output of the system detected by the sensor is proposed. As for the output feedback control proposed until now, output-observation noise is not included although system noise is included in the controlled object. We consider an output feedback control problem which evaluates only the state to a system including system noise and output observation noise. Our goal is an output feedback law that minimizes the state covariance matrix. A necessary and sufficient condition is found for a static output feedback law to be optimal among all the dynamic output feedback laws. The condition is stated in terms of an algebraic Riccati equation, which is same form as that appearing in the Kalman filter.
In this paper, we propose a novel keypoint detection and feature description method called “SHORT” (Shell Histograms and Occupancy from Radial Transform) for fast 3D object recognition. Conventional keypoint detection and feature description methods such as the SHOT method have been necessary to calculate many normal vectors or other statistical values from the point cloud data in local regions, so its computational costs are expensive. By contrast, the SHORT method consists of a fast keypoint detector that does not calculate statistics and a fast feature descriptor that uses a small number of points in the restricted local regions. The keypoint detector uses the occupancy measure which can be estimated by only counting the number of points in multiple spherical shell regions. Also the feature descriptor uses a small number of points included in distinctive shell regions of multiple scales. Experimental results in 3D object recognition using real dataset show that the processing speed of the proposed method is approximately nine times faster than that of comparative methods.
To achieve more realistic haptic rendering, it requires the haptic modeling based on the haptic sensing and analyzing of the physical phenomenon which occur between objects in the real world and requires the control of haptic device at real-time. In the field of the robotics, bilateral control is used to transmit the haptic information between multipoint. In our previous work, we developed a friction sensing and rendering system based on the bilateral control with ONE 3DOF haptic device. However, it was difficult to measure and represent the haptic perception with human finger. Therefore, we propose an estimate method of the haptic perception such as the hardness and friction of the target object based on the bilateral control with ONE 3DOF haptic device and an artificial finger.
We propose a texture-based haptic rendering system for the touch panel based on the pressure force analysis with an artificial finger. To add the haptic rendering for the touch panel, it is necessary to measure the real object in the real world and render the physical stimuli to the finger via the touch panel. Generally, force sensor is used for the force sensing. However, it is difficult to recreate the sensing information directly with haptic interface. Moreover, previous studies have not considered the haptic analysis of the several real objects in the haptic perception and manipulation. Therefore, we propose an estimate method of the haptic perception such as the contact force and hardness of the target object in the pressure motion based on the pressure force analysis with artificial finger and propose a texture-based haptic rendering system for the touch panel. As a result of the experiment, we were able to confirm the effectiveness of the proposal method.
We have developed a new virtual reality system for supporting the exercise. One of objectives in the system development is the generation of force sense to the system user without complex haptics devices by introducing the Pseudo-Haptic. In this paper, we had examined the generation of the Pseudo-Haptic with the change of operability. In the experimental system, the character on a computer screen moves with motion of user's hands captured by Kinect sensor. The experimental results suggest that the Pseudo-Haptic generate on examinee's arm with speed-up or slowdown of the character moving.
An independent component analysis (ICA) has received much attention in the fields of pattern recognition, such as face, fingerprint, figure, character, or number recognition. FastICA is an efficient and popular algorithm for ICA, and an improved algorithm is available for character recognition. In order to increase the rate of recognition, the appropriate setting of parameter or base number is indispensable. We investigate the effects of a change in base number on recognition rate to clarify the index for determining the number. As a result, we confirmed that recognition error rate tends to decrease in case the number is 3, 4, or 5.
It is well known that points on a plane in 3D world are related to corresponding image points in a view of a moving camera by projective translation. Good image feature have robust projectivity under any camera movements. In the standard performance evaluation of image processing, real captured images of a scene is used ordinary. However it is not enough to evaluate in detail because the variation of camera angle and distance to target objects is limited and the capturing cost is expensive. During the early stage of the image processing development, the basic performance measurement should be the most important in an easy way. We propose a projectivity diagnosis method to measure the performance of local descriptor base template matching between a template image and reference images which are created by deforming the template image. This template matching consist of a feature image point extraction and a local descriptor matching. The proposed method evaluates the positional accuracy of the extracted feature points and the matching with local descriptor. Four metrics are introduced to evaluate the projectivity of template matching. In the experiment, our proposed diagnosis method exposed the projectivity of SIFT, SURF and ORB. SIFT showed the better robustness than the others.
In this paper, we propose a method of surveillance of the plant growth using image processing. This method is able to observe the condition of raising the plant in the greenhouse by detecting small insects. Experts can classify small insects, but many general people cannot classify small insects. Therefore we propose a method of classifying small insects by using image processing. The plate is prepared for extracting small insect. The image is obtained by an image scanner. In first process, each region of small insects is extracted from the image scanner image by using color information. In next process, some features are detected from the region of small insects. In last process, small insects are classified using above features. The experiments using the experimental system are performed to demonstrate the efficacy of this method and the experimental results are shown.
In this paper, we propose a method to create a 3-dimensional model of space using 3-d Rotation Invariant Phase Only Correlation (RIPOC). RIPOC is the applications of Phase Only Correlation (POC). The RIPOC can correlate two data even though there is rotation between two data. In this paper, we extend the RIPOC to 3-dimensional. It estimates the rotation angle between two 3-d information. We create a 3-d model using the estimated results. And it shows the ability of the proposed method.
Special Issue on “International Conference on Electronics and Software Science 2015”
In this paper, we examine a technique wherein the production of body sway in a desired direction was considered using a vibro-tactile stimulator. Creating body sway in a specific direction to assist in postural control is necessary. Therefore, we aim to introduce a device that creates spatio-temporal tactile stimulation pattern by altering the interval times between vibrations. We measure body sway caused by each stimulation using a high-speed camera, stabilometer, and acceleration sensor. We then analyze the effects of each stimulation on body sway using an EM algorithm to estimate parameters of a contaminated normal distribution model. Consequently, this paper shows specific body sway for each stimulation, which suggests the optimal time duration and interval of stimulation for a specific direction.
There have been a number of investigations into image recognition and the assessment of human physiological states using infrared thermography. Assessing a human's physiological state by infrared thermography typically exploits the skin temperature of the nasal region and forehead, whereas other parts of the face are less frequently used. The present study has developed a method of analyzing facial thermal images (FTIs) by independent component analysis (ICA), a type of blind signal processing (BSP). ICA is a well-known statistical analysis tool that estimates the original source signal from observed mixture signals. When applied to thermal images, ICA is predicted to extract blind signals such as those from other parts of the face. In this study, the authors use ICA to conduct BSP on a series of FTIs. The extracted independent components are shown to represent temperature fluctuations from the opening and closing of the eyes, respiration, truncal sites such as the cheeks and forehead, and possibility of sympathetic nervous system activity. The FTIs reconstructed after the removal of artifacts indicate the local features that the blind signal cannot extract from the original FTIs.
In design process, miscellaneous knowledge is required for achieving desirable goal. Collaboration is a crucial method that contributes designers to create prime solutions by sharing their knowledge within a team. During a collaboration, valuable novel knowledge that is not held by the members could emerge due to synergistic effect. However, the way to generate such new knowledge is implicit. In this paper, a new model for collaborative design mechanism is proposed to investigate process of generating such knowledge. Effect of a collaboration can be visualized by using the proposed model. To show the usefulness of the model, a calculation tool has been developed based on the proposed model.
Aiming to help persons with mental health problems identify their own mental state and control it, not by visiting specialists passively but by proactively confronting their symptoms, this paper proposes an indirect biofeedback system that externalizes and objectifies the physiological state of users to allow them to self-control their inner state but also to control their physiological state. In the proposed system that we have developed, based on a display representation of physiological information with colors and shapes, the users can grasp their inner state and control it by different methods of breathing that help control their autonomic nervous system. Also, this paper clarifies the usefulness of the proposed system, showing the experimental results in comparison with the conventional direct feedback waveform display system.
In order to save energy, electric power of battery need to be used up as much as possible. It is necessary to design a DC-DC converter with high boost ratio. We elucidated a boost ratio of tapped-inductor DC-DC converter with considering element-resistance. The boost ratio of tapped-inductor DC-DC converter was analyzed in continuous conduction mode, critical conduction mode and discontinuous conduction mode. In experiment, tapped-inductor DC-DC converter was able to change 0.478V, in that voltage battery uses up its power energy, into 12.0V for 100Ω load resistance. Experimental boost ratio was improved to 1.51 times of that of conventional converter. Theoretical boost ratio, that considered series resistance in inductance devices and semiconductor devices was improved to 1.40 times of that of conventional converter. Furthermore, power efficiency of tapped-inductor DC-DC converter was from 62% to 75%, in that of conventional converter was from 71% to 73%.
A large fraction of the Internet traffic is organized by the Content Delivery Networks (CDNs). CDNs redirect end-users to an appropriate server using Domain Name Servers (DNSs). Meanwhile, in future networks, it is anticipated that network routers will be equipped with more processing power and storage modules for providing most effective end-user services. From this viewpoint, a Service-oriented Router (SoR) is introduced to accelerate content-based services. This paper presents a novel method, DNS and SoR Collaborative Redirection (DSCR), to accelerate the end-user redirection. DSCR uses SoRs to collect and store network and server state information by placing SoRs at edges of the ISP network. The collected information is then used by the DSCR to leverage the DNS-based end-user redirection. We present the design and a prototype implementation of DSCR and, show how both SoR and DNS jointly can take advantage of existing Internet architecture to accelerate end-user redirection. We used a detailed simulation to evaluate the proposed method using various strategies. Simulations clearly demonstrate the effectiveness of the collaborative method, which yields 5-10% reduction in end-user latency when compared with DNS-based redirection.
We developed a system incorporating comb-electrodes that can simply be laid on bedding for electrical detection of excrement leakage by bedridden elderly patients, without the need to install special sensors in general-purpose disposable diapers or auxiliary pads. The system is designed to detect urination externally from patient undergarments and estimate the quantity of urination based on changes in paper-diaper or auxiliary-pad impedance when wet by the fluid. When disposable diapers or auxiliary pads get wet with urine, the impedance changes. Test subjects lying full-length on beds showed that the system could distinguish effectively between urination and non-urination through detection of the externally placed comb electrodes, and it was found that the change in impedance correlated with the urination quantity.
Recently, a novel human-machine interface known as the eye-gaze input system has been reported. This system is operated solely through the user's eye movements. Therefore, it can be used by people suffering from severe physical disabilities. We propose an eye-gaze input system that uses a personal computer and home video camera. This system detects the users' eye-gaze through image analysis under natural light including fluorescent or LED light. Our proposed system also has a high-level accuracy and confidence; that is, users can easily move the mouse cursor to their gazing point. We confirmed a large difference in the duration of voluntary (conscious) and involuntary (unconscious) blinks through a precursor experiment. In addition, we confirmed that these durations vary significantly depending on the subject. By using the duration of eye blink, voluntary blink can be detected automatically. Through this method, we developed an eye-gaze input interface that uses information of voluntary blinks. That is, users can decide their input by performing voluntary blinks that represent mouse clicking.
Inpatients of circulatory system disease must manage their urine volume every day since they have low ability to control fluid balance inside their bodies. In most hospitals, measuring cups are used which lead to nosocomial infection. Our study propose a new method of urine volume estimation without using a cup. We propose multiple cylindrical model to estimate the amount of liquid volume from images taken by a monocular camera. This model is based on the idea of calculating the total volume of cylinder extracted from each image. First, images of liquid simulating male urination are binarized to derive features for the model. Each volume of cylinder is calculated by the initial velocity and diameter of liquid in each image. We conducted experiments to evaluate the model. As a result, we suggest that this model could be a new way of urine volume management for inpatients.
Environmental concern has recently led to saving of electric power for water transmission and distribution systems in water utilities. To cope with the problem, effective water supply operation and appropriate management of water pressure have been researched. Moreover, after a large earthquake in eastern Japan, the problem of electricity shortage has been evident. To address the problem, the water saving which results in the electricity saving has been encouraged and the driving of waterworks facilities has been shifted from a peak demand time zone to a low demand time zone in order to achieve power peak cut. We propose a transmission pump operation scheduling method that minimizes the maximum value of power consumption in a time zone of demand response request. The optimal operational plan of pumps is made by mathematical programming so that the power consumption is minimized under constraints that the each reservoir keeps water level to enable 12-hour consecutive water supply and the water level does not go under lower limit during and after a time period of demand response. Simulation tests showed that the proposed method reduced power consumption by more than 65% compared with the current pump operation.
Q-learning is learning the optimal policy by updating in action-state value function(Q-value) to maximize a expectation reward by a trial and error search. However, there is major issues slowness of learning speed. Therefore, we added technique agent memorize environmental information and useing with update of the Q-value in many states. By updating the Q-value in the number of conditions to give a lot of information to the agent, be able to reduce learning time. Further, by incorporating the stored environmental information into action selection method, and the action selection to avoid the failure behavior, such as learning to stagnation, improved the learning speed of learning the initial stage. In addition, we design a new action area value function, in order to search for much more statas from the learning initial. Finally, numerical examples which solved maze problem showed the usefulness of the proposed method.
In this paper, we propose an adaptive noise canceller (ANC) with speech suppressors to reduce noise in car environment, where the noise consists of directional and diffuse noise signals. Although the ANC is useful for suppressing the directional noise, its capability is degraded when a desired speech signal is observed at both microphones simultaneously. Such situation is called as crosstalk. To solve this problem, we introduce two speech suppressors into the ANC, and employ one noise suppressor as a post-filter to suppress diffuse noise. Simulation results show that the proposed method effectively suppresses both of directional noise and diffuse noise.
Evolvable Hardware (EHW) is reconfigurable hardware which can adopt to unknown new environments. EHW can be implemented combining learning networks such as Neural Networks (NNs) and programmable devices such as FPGA (Field Programmable Gate Array). As such research of EHW, Block-Based Neural Networks (BBNNs) have been proposed. BBNNs have simplified network structures and have been attracting attention with their ease of hardware implementation. In particular, Block-Based Pulsed Neural Networks (BBPNNs) which adopt a pulsed neuron model instead of an analogue neuron model in BBNNs have been proposed in order to solve the problem that BBNNs use many multiplier circuits and require large scale hardware resources for implementation. In addition, applying Back Propagation (BP) which is common learning algorithm of NNs to BBPNNs has been proposed. In this paper, we propose two approximation methods in order to reduce hardware resources which are necessary to apply BP to BBPNNs. In the proposed methods, we approximate input values and derivative values of activation function in BP. Results of computational experiments indicate the validity of the proposed methods.