-
Article type: Cover
Pages
Cover1-
Published: February 13, 2016
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS
FREE ACCESS
-
Article type: Index
Pages
Toc1-
Published: February 13, 2016
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS
FREE ACCESS
-
Takamitsu GOTO, Koki KOSAKA, Keisuke NONAKA, Sei NAITO, Naoki HASHIMOT ...
Article type: Article
Session ID: ME2016-1
Published: February 13, 2016
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS
FREE ACCESS
Recently, projectors are requested to project images in various situations including non-white surfaces. Many radiometric compensation methods with a ProCam system have been researched in order to project ideal images onto those surfaces regardless of their colors and patterns. However, ProCam system has to set its own camera in front of viewer's eyes, and that limitation is not suitable for practical use. Therefore, in this research, we propose a method that viewers can observe compensated images with no camera in front of them by using images from the other placed camera.
View full abstract
-
Mitsuhiko KITANO, Naoki HASHIMOTO
Article type: Article
Session ID: ME2016-2
Published: February 13, 2016
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS
FREE ACCESS
In this research, we propose high-speed and high-precision radiometric compensation by taking account of the pixel correspondence in ProCam. Radiometric compensation is a technique for projecting an image without being influenced by the reflectance on an arbitrary surface. However, the accurate compensation need several ten times of a feedback process with ProCam because of pixel corresponding errors. Therefore, the purpose of this research is to achieve radiometric compensation without using the feedback by measuring the effect of the error.
View full abstract
-
Yuki SAKAMOTO, Tomoaki NAKAMURA, Masahide KANEKO
Article type: Article
Session ID: ME2016-3
Published: February 13, 2016
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS
FREE ACCESS
Accurate detection of facial feature points from photograph is crucial to analyze facial features and to synthesize impressive caricatures by computer. A lot of studies on automatic detection of facial feature points have done until now. However, accurate detection methods which can be used in above-mentioned applications have not been developed yet. This paper proposes a method to extract the nares and wings of nose accurately based on the characteristics of luminance distribution. The effectiveness of the proposed method is shown by applying it to various facial images captured under different conditions.
View full abstract
-
Miyuki FUNADA, Tomoaki NAKAMURA, Takayuki NAGAI, Masahide KANEKO
Article type: Article
Session ID: ME2016-4
Published: February 13, 2016
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS
FREE ACCESS
It is considered that concepts are not only formed based on perceptual information in a bottom-up fashion but also affected by language and culture in a top-down fashion. Different concepts depending on each culture can be formed by such interaction between concepts and language. In this paper, we target a concept formation of colors based on multimodal LDA, which we have proposed, and analyze the interaction between concepts and language. Furthermore, we verify a validity of the proposed method by comparison with a human concept formation.
View full abstract
-
Ami Nagano, Yasuyuki Saito, Eita Nakamura, Shigeki Sagayama
Article type: Article
Session ID: ME2016-5
Published: February 13, 2016
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS
FREE ACCESS
Automatic music accompaniment system synchronizes with human player's performance. Eurydice is one of such systems that model the generation process of the performance with a probabilistic model, which can therefore follow flexibly the performance in presence of tempo changes, mistakes, skips and repeats. As the tempo control is made by following the player's input, it is a problem while the player's part has a long rest in the score. In this study, we discuss tempo control of the accompaniment by key tapping. To determine a key for the tap operation, wrong key rate was experimentally evaluated across various distances between keys. Then, as a preprocessing, we inserted necessary notes for the tap operation into the Standard MIDI file (SMF) for Eurydice, so that Eurydice controls tempo and proceeds forward when the player tapped the inserted notes. The inserted notes cause no sounds, because they do not exist in the original musical score. We implemented these new features and confirmed a successful operation.
View full abstract
-
Youjirou DAIMON, Yasuyuki SAITO
Article type: Article
Session ID: ME2016-6
Published: February 13, 2016
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS
FREE ACCESS
In a recycling company, it is necessary to detect boundary between water and oil in drum carried in process of recycling. Currently the employees detect boundary from touch of a rod inserted in drum. However, this method depends heavily to individual senses and also rainwater will mix in drum in case of rain. Therefore, the purpose of this study is to detect boundary between water and oil in sealed drum using ultrasonic probe that can send and receive ultrasonic.
View full abstract
-
Tomoki KUSANO, Yasuyuki SAITO
Article type: Article
Session ID: ME2016-7
Published: February 13, 2016
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS
FREE ACCESS
This study personalizes Head-Related Transfer Function (HRTF). HRTF is major physical property to feel sound sterically. HRTF has individual difference for every person by head and auricle form. Using other person's HRTF causes false localization of sound and lowering realistic. So we focused individual difference, simulated to effect on HRTF by changing head and auricle form and estimated HRTF from subject's head and auricle form.
View full abstract
-
Riku Sato, Eita Nakamura, Yasuyuki Saito, Suguru Agata, Yuu Igarashi, ...
Article type: Article
Session ID: ME2016-8
Published: February 13, 2016
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS
FREE ACCESS
The current mainstream of music score employs staff notation. Not limited within piano, staff notation is used for many types of instruments as it can contain a large amount of information. It is, however, often hard to read staff notation for novice players who are not familiar with it because of its complexity. Vertical line notation is a new type of notation, which is simple, intuitively designed and thus easy to understand. In this study, we examined a method of creating music scores on vertical line notation including piano fingering automatically from standard MIDI files (SMF), furthermore, a method of regulating upper and lower limits of displaying scale.
View full abstract
-
Honoka FUJII, Yasuyuki SAITO
Article type: Article
Session ID: ME2016-9
Published: February 13, 2016
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS
FREE ACCESS
In this study, the development of software which can be degree of difficulty judgment and can be complication and simplification the SMF of the music data. First, we analyze each event by reading the existing SMF, and determining the degree of difficulty of the music based on the information of the amount of the note, the range to be used, and the rhythm. Then, to add the arrangement function that makes it difficult to insert the music pattern to any location, and also makes it easy more by deleting or simplifies any phrase in the music data.
View full abstract
-
Hidemitsu Watanabe, Misae Yagi, Mie Sato
Article type: Article
Session ID: ME2016-10
Published: February 13, 2016
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS
FREE ACCESS
In recent years, demands for 3D modeling systems have been increasing because 3D printers are widely sold generally. In this study, to create a 3D model easier, we propose operation methods in a 3D modeling system using an infrared sensor. We examine methods of pulling and splitting a virtual object. Using the infrared sensor to detect bare hands and a stick, this system can deform the virtual object with user's hands and the stick when the user pulls the virtual object and splits the virtual object by passing the stick through the virtual object.
View full abstract
-
Yoshiharu URA, Sho KATO, Mie SATO
Article type: Article
Session ID: ME2016-11
Published: February 13, 2016
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS
FREE ACCESS
Augmented reality (AR) technology becomes familiar and AR experience with a see-through head mounted display (HMD) becomes popular. AR can append information to the real world. Using a see-through HMD, this study aims to find a particular character string from a lot of information in sight. We set the character string that we want to look for as an AR marker. Then, by feature point matching, we search for the character string in a view from a camera attached to the see-through HMD. As a result, an area of the character string is emphasized in red color on the see-through HMD.
View full abstract
-
Daiki Ebihara, Sota Suzuki, Mie Sato
Article type: Article
Session ID: ME2016-12
Published: February 13, 2016
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS
FREE ACCESS
Interaction between a user and a virtual object is one of studies on augmented reality (AR). In a previous study, an AR system with which the user could grasp a spherical virtual object with his/her bare hand was developed. However, if the user grasps a non-spherical virtual object, he/she can feel strange against the operability and visuality. Therefore, in this study, we provide an AR system that enables the user to grasp, move, and rotate a virtual object with various shapes.
View full abstract
-
Tsubasa KUROSAKA, Tomoaki NAKAMURA, Masahide KANEKO
Article type: Article
Session ID: ME2016-13
Published: February 13, 2016
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS
FREE ACCESS
In this paper, we proposed a novel path planning and speed control method for autonomous mobile robot under dynamic environment. The robot not only needs to avoid stationary objects such as walls and obstacles, but also needs to consider mobile objects such as human beings and pets. In order to avoid the collisions with mobile objects, the robot should be able to predict the motion of the objects, plan its moving path, and control its speed to move towards the destination while avoiding the mobile objects. Previous researches tried to solve this problem by using a spatiotemporal RRT method, which can spatiotemporally predict the motion of mobile objects and plan the path of the robot. However, dangerous paths that go through the places nearby the objects may be generated, as it does not consider the moving ambiguity of the mobile objects. Thus we proposed a method to plan paths which leave away from the dynamic obstacles by changing the local sampling probability. Simulation results show the usefulness of the proposed method.
View full abstract
-
Takafumi WATANUKI, Tomoaki Nakamura, Masahide KANEKO
Article type: Article
Session ID: ME2016-14
Published: February 13, 2016
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS
FREE ACCESS
Autonomous mobile robot that accompanies with people and offers services to him should move autonomously by adjusting the relative position with companion properly depending on surrounding conditions. As one of such robots, this paper considers the mobile type telepresence robot. This robot accompanies with companion and moves under dynamic environments while keeping the conversation between the companion and remote partner displayed on the screen of robot. To realize such function this paper employs an artificial potential method that is optimized to above accompanying behavior. The effectiveness of the proposed method is shown through simulation and experiments using a real robot.
View full abstract
-
Kohei HAYASHI, Tomoaki NAKAMURA, Masahide KANEKO
Article type: Article
Session ID: ME2016-15
Published: February 13, 2016
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS
FREE ACCESS
In this paper we aim to apply the drawing methods of caricature artist to the synthesis of facial caricatures by computer. Caricature artist modifies the drawing styles and the way of emphasis depending on facial features of individual person. The characteristics in drawing methods of caricature artist are treated numerically by applying principal component analysis to shapes and arrangement of facial parts in caricatures drawn by artist. The general framework of the face recognition by humans and techniques of caricature artist are also considered to reflect the facial features of an individual more effectively to caricatures. Facial caricatures drawn by the proposed method are compared to those by the previous method to show its usefulness.
View full abstract
-
Tetsutoshi NOMURA, Asako SOGA
Article type: Article
Session ID: ME2016-16
Published: February 13, 2016
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS
FREE ACCESS
In this study, we simply created three-dimensional animations of the human body at low cost to support dance creation. We developed a choreography synthesis system by inputting body-part motions using a Kinect. During the playback of the existing motion data, the motions of both arms inputted by Kinect were synthesized to the motion data in real time. Users combine steps from 40 hip-hop dance steps that were prepared in four counts in time series and input the arm motions to the created sequence. The inputted arm motions are blended to the corresponding parts of the hip-hop steps.
View full abstract
-
Keisuke SUMIDA, Asako SOGA
Article type: Article
Session ID: ME2016-17
Published: February 13, 2016
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS
FREE ACCESS
A Buddhist ceremonial procession called Nerikuyo is difficult to accurately record and transmit because visible things change by time and place. We developed a system using 3DCG and interactive techniques to simulate the human animation of Nerikuyo for its digital archiving and simulation. Users can edit the number of people walking, the distances among them, their walking patterns in the text file, and simulate the lines of people, the human movements, and the elapsed time. The created Nerikuyo 3DCG can be previewed from different viewpoints using HMD and Leap Motion.
View full abstract
-
Anna Kobayashi, Michimi Inoue, Mie Sato, Miyoshi Ayama
Article type: Article
Session ID: ME2016-18
Published: February 13, 2016
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS
FREE ACCESS
We examined relation of the bit depth and the gamma characteristic in image display to image impressions by subjective assessment. Our assessment was conducted using a 10bits display in a non-darkroom. As the result of this assessment, we confirmed that the bit depth and the gamma characteristic have relation to image impressions. In addition, it is suggested that appropriate gamma characteristics can increase effects of the 10 bits display in the non-darkroom. However, it is also suggested that the10 bits display decreases image impressions in some conditions.
View full abstract
-
Yuta YOSHINO, Tomoharu ISHIKAWA, Yoshiko YANAGIDA, Kazuya SASAKI, Miyo ...
Article type: Article
Session ID: ME2016-19
Published: February 13, 2016
Released on J-STAGE: March 30, 2018
CONFERENCE PROCEEDINGS
FREE ACCESS
The purpose of this study, the rotation duration and rotation speed is to reveal whether affect how the transmitted ease texture of the fabric in the video with rotating. In this experiment, we were prepared videos of total sixteen patterns of the rotation duration four patterns and the rotational speed four patterns for each fabric. We were examined transmitted ease of texture from the error of video evaluation and real evaluation. As result, softness or hardness of the fabric that is easily transmitted was shown by determining the combination of the rotational speed rotary duration for each softness and hardness fabric.
View full abstract
-
Irin Chin, Miyosi Ayama
Article type: Article
Session ID: ME2016-20
Published: February 13, 2016
Released on J-STAGE: March 30, 2018
CONFERENCE PROCEEDINGS
FREE ACCESS
In this study, different types of animations of dragon's movement are used to do the experiment and evaluation, and the result of the experiment will be analyzed to find out the influence of different elements to the sense of 3D and reality. Firstly, Low poly dragon model is used to do the experiment, drawing the pictures and analyzing according to the data from the experiment. High poly dragon model is also used to make new animation , using experiment to compare with Low poly dragon's animation. Because of the errors of the data, the ANOVA (Analysis of Variance) had also been analyzed.
View full abstract
-
Yuki HAYAKAWA, Hotaka TAKIZAWA, Hiroyuki KUDO, Toshiyuki OKADA
Article type: Article
Session ID: ME2016-21
Published: February 13, 2016
Released on J-STAGE: March 30, 2018
CONFERENCE PROCEEDINGS
FREE ACCESS
In this paper, we propose a method of non-rigid registration of abdominal 3D CT images. First, gas regions are eliminated from reference and floating CT images based on a viscous-fluid transformation model. Then, 3D global transformation is performed by fitting bone structures in the images. Finally, we use a 3D non-rigid registration method based on the Thirion's demons algorithm and the Freiman's local-Affine estimation algorithm. This registration method is applied to actual images, and experimental results are shown.
View full abstract
-
Tatsuya YASUDA, Hotaka TAKIZAWA, Toshiaki OKUMURA, Hiroyuki KUDO, Tosh ...
Article type: Article
Session ID: ME2016-22
Published: February 13, 2016
Released on J-STAGE: March 30, 2018
CONFERENCE PROCEEDINGS
FREE ACCESS
This paper proposes an extraction method of vertebrae in abdominal X-ray CT images. Bone regions are extracted using binarization and morphological operators, and then spinal columns and ribs are obtained from the bone regions. The positions of intervertebral disks are estimated, and the spinal columns are divided into vertebra columns based on the estimated positions. The vertebral columns are identified on the basis of their relative positions and connectivity. This paper focuses on the extraction of the 12th thoracic vertebra (T12) observed in different areas of abdominal CT images. The proposed method was applied to 15 cases. The experimental results demonstrated that the proposed method was effective.
View full abstract
-
Masafumi SHIMOYAMA, Sayaka SHIOTA, Hitoshi KIYA
Article type: Article
Session ID: ME2016-23
Published: February 13, 2016
Released on J-STAGE: March 30, 2018
CONFERENCE PROCEEDINGS
FREE ACCESS
In this paper, we propose an image retrieval method using signs of discrete wavelet transform (DWT) coefficients for JPEG 2000 images. For JPEG 2000 images, a number of image estimation methods using the number of zero bit-planes, extracted from the header information have been studied so far, and they have been applied to the calculation of the similarity between images. However, the estimation methods limited to the absolute values of DWT coefficients. The purpose of this article is to robustly estimate image features, and to apply it to image retrieval for JPEG2000 images. The proposed method uses only signs of low-band DWT coefficients to achieve robust image retrieval against the difference of compression levels. A number of simulations are provided to confirm the effectiveness of the proposed method.
View full abstract
-
Yohei Aoki, Keisuke Maruyama, Naoki Hashimoto
Article type: Article
Session ID: ME2016-24
Published: February 13, 2016
Released on J-STAGE: March 30, 2018
CONFERENCE PROCEEDINGS
FREE ACCESS
Recently, projection mapping which can change the looks of an object by projecting images attracts attention. In order to perform the projection mapping, we should calibration of a projects image's position to match a projection object's shape with a projects image. This work is too hard for us to do High-Accuracy Projection Mapping. Well, if we introduce the depth sensor, we can implement it easily, because it can get projection object's position and attitude. But, the depth sensor's accuracy of measurement is not high enough. Therefore, in this research, we propose high accuracy projection mapping system with a low-cost depth sensor by high accuracy calibration and improvement the accuracy of measurement using a robust estimation.
View full abstract
-
Naoya NIWATA, Naoki HASHIMOTO
Article type: Article
Session ID: ME2016-25
Published: February 13, 2016
Released on J-STAGE: March 30, 2018
CONFERENCE PROCEEDINGS
FREE ACCESS
Recently, spatial augmented reality (SAR) has been used as a method of editing appearance of real objects by using a projector in entertainment fields. In addition to the general SAR which projects images toward stationary objects, dynamic SAR which projects images toward moving objects is especially paid attention. In order to realize the dynamic SAR, it is necessary to obtain the position of the projection target in real time. However the measurement using non-contact device, such as cameras, is difficult to achieve enough processing speed. Therefore, in this research, we propose a minimization method of data used for the object tracking by selecting effective key point from a depth image, and accomplish acceleration of the pose estimation.
View full abstract
-
Masaki KOUCHI, Osamu WATANABE
Article type: Article
Session ID: ME2016-26
Published: February 13, 2016
Released on J-STAGE: March 30, 2018
CONFERENCE PROCEEDINGS
FREE ACCESS
-
Kei SATOU, Masashi KAMEDA
Article type: Article
Session ID: ME2016-27
Published: February 13, 2016
Released on J-STAGE: March 30, 2018
CONFERENCE PROCEEDINGS
FREE ACCESS
JPEG images include a visual distortion which is called the mosquito noise around edge region. It is difficult to estimate the mosquito noise from a given encoded image, since the visual effect is dependent with the characteristics of each input image. This paper presents an estimation method of the mosquito noise based on the cepstrum analysis. It is seen in our experimental results that the period and the intensity are relationship with the peak of the cepstrum coefficients.
View full abstract
-
Takuya NISHIMOTO, Shota NAKAMURA, Yoshiki MIZUKAMI
Article type: Article
Session ID: ME2016-28
Published: February 13, 2016
Released on J-STAGE: March 30, 2018
CONFERENCE PROCEEDINGS
FREE ACCESS
Local features as represented by Scale-Invariant Feature Transform are generally used for Content Based Image Retrieval. However, most previous methods using such local features rely only on intensity value of images and then they do not take advantage of the color information which may be helpful for the application. In this study, we propose Key Line Segment (KLS) feature which regards color information on line segments connecting two local feature points as the feature descriptor. The effectiveness of KLS is discussed based on the simulation results.
View full abstract
-
Shingo NAKATANI, Yoshiki MIZUKAMI
Article type: Article
Session ID: ME2016-29
Published: February 13, 2016
Released on J-STAGE: March 30, 2018
CONFERENCE PROCEEDINGS
FREE ACCESS
In this paper, we propose a method to utilize the color texture of an object for estimating the 3D posture of the human body from monocular photography image. In the proposed method, first, we map the object texture to a 3D human body model. Next, we assign postures in the motion capture database on the human body model and produce posture images by rendering the model from eight horizontally-surrounding directions. Finally, we compare the photography image with these posture images, and estimate the posture of the human body in the photography image. We compare the posture estimation results obtained by using RGB value, HOG feature and color histogram, and discuss the effectiveness of the proposed method.
View full abstract
-
Naoto SUYAMA, Shun TOKUDA, Yoshiki MIZUKAMI
Article type: Article
Session ID: ME2016-30
Published: February 13, 2016
Released on J-STAGE: March 30, 2018
CONFERENCE PROCEEDINGS
FREE ACCESS
Recently many researchers study 3D reconstruction using multiple images. Most objects have planner surfaces especially in urban and internal scenes so that detecting the planner surfaces in these scenes seems to be helpful for estimating the depth map accurately. In this study, we detect planner surfaces in an image by investigating the color distribution between three-dimensional feature points obtained by a Structure-from-Motion technique. Next, plane sweep is applied to the images along the normal direction of the detected planes and multiple disparity space images (DSIs) are generated. Then, disparity images are obtained by finding disparities with the minimum cost through the DSIs. Finally, these disparity images are unified into a whole disparity image by using the cost as a criterion.
View full abstract
-
Haruka JIBIKI, Yasuyuki SAITO, Eita NAKAMURA, Shigeki SAGAYAMA
Article type: Article
Session ID: ME2016-31
Published: February 13, 2016
Released on J-STAGE: March 30, 2018
CONFERENCE PROCEEDINGS
FREE ACCESS
This paper discusses discrimination of player's nodding between page turning cue and finding a rhythm for an automatic page turning system for music score. Nodding is a typical signal for page turning in actual playing the piano. So we developed an automatic page turning system by player's nodding. However nodding include in not only turning cue but also finding a rhythm. Then the system needs to recognize only nodding for page turning cue. Nodding of finding a rhythm is shallow than that of page turning cue for a result of surveying a sampling of these. Therefore using nodding deepness for a threshold can be capable of discriminating two nodding.
View full abstract
-
Tomoko NISHIJIMA, Asako SOGA
Article type: Article
Session ID: ME2016-32
Published: February 13, 2016
Released on J-STAGE: March 30, 2018
CONFERENCE PROCEEDINGS
FREE ACCESS
This study supports understanding Shomyo scores and to stimulate interest in Shomyo by people who are unfamiliar with it. We developed a creation-and-play system for Shomyo scores for tablet computers. In our system, Shomyo scores are reproduced on a tablet based on the input of users who create scores by dragging ink line images and lyrics. Sound can also be generated that corresponds to the score to confirm that the user is actually chanting a Shomyo.
View full abstract
-
Xiaohua ZHANG, Ning XIE, Heming HUANG, Yuelan XIN
Article type: Article
Session ID: ME2016-33
Published: February 13, 2016
Released on J-STAGE: March 30, 2018
CONFERENCE PROCEEDINGS
FREE ACCESS
In this report, several methods are introduced for extracting character strokes from a document image with an irregular background. It is known that the irregular background is the biggest obstacle for restoring text strokes from document image. This paper proposes several approaches for removing the uneven shading by using piecewise polynomial model and Gaussian kernel model to estimate shading surface. However, these two models cannot well estimate more complicated background surface such as bleed-through, creased paper, smear and stains. A progressive method for restoring text strokes is proposed for these kinds of document images. First, a weighted constant time median filter is used to compute the irregular background and remove it as completely as possible on retinex theory. Next, hierarchical thresholding is established at different scales. Unclassified pixels in regions close to text strokes are further binarized. Finally, after computing the skeleton of each text stroke, a geometric approach is applied to preserve the connectivity of the strokes. Intensive experiments were conducted on various document images. The results demonstrate that our method yields better results than other methods.
View full abstract
-
Article type: Appendix
Pages
App1-
Published: February 13, 2016
Released on J-STAGE: March 30, 2018
CONFERENCE PROCEEDINGS
FREE ACCESS
-
Article type: Appendix
Pages
App2-
Published: February 13, 2016
Released on J-STAGE: March 30, 2018
CONFERENCE PROCEEDINGS
FREE ACCESS
-
Article type: Appendix
Pages
App3-
Published: February 13, 2016
Released on J-STAGE: March 30, 2018
CONFERENCE PROCEEDINGS
FREE ACCESS
