All issues
Predecessor
39.8
Displaying 1-33 of 33 articles from this issue
- |<
- <
- 1
- >
- >|
-
Article type: Cover
Pages Cover1-
Published: February 21, 2015
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS FREE ACCESSDownload PDF (11K) -
Article type: Index
Pages Toc1-
Published: February 21, 2015
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS FREE ACCESSDownload PDF (97K) -
Kaito ISHIDA, Yasuyuki SAITOArticle type: Article
Session ID: ME2015-35
Published: February 21, 2015
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS FREE ACCESSThe ensemble means that two or more players perform music at the same time. The ensemble needs plural players, but players may not be gathered easily. The automatic accompaniment system is useful to solve this problem. However, a simple automatic accompaniment is unsuitable to use at a concert because it has some problems. Therefore, this study examines to improve expression to be able to use automatic accompaniment at a concert.View full abstractDownload PDF (469K) -
Riku Sato, Yasuyuki Saito, Suguru Agata, Shigeki SagayamaArticle type: Article
Session ID: ME2015-36
Published: February 21, 2015
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS FREE ACCESSCurrent mainstream music sheet is staff notation not limit to piano. Staff notation is used for many types of instruments and that has large information volume. However, it is difficult to understand staff notation for beginner player who are not used to reading it because of its complexity. Vertical line notation is a new type sheet that is easy to understand and put forward by Suguru Agata who is collaborator of this research. It has simple and intuitive design. In this study, we examined a method of creating vertical line notation from MIDI file.View full abstractDownload PDF (559K) -
Sae NEMOTO, Yasuyuki SAITOArticle type: Article
Session ID: ME2015-37
Published: February 21, 2015
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS FREE ACCESSThis paper shows to create a music piece which matches with impressions of an input color image. The picture and the music have a strong influence on the emotion of the person. When a person watches an image such as a picture or a photograph, it is assumed that they can feel deeper the impression of the image if a music which matches the image impression is played.View full abstractDownload PDF (3826K) -
Toshimitsu Hara, Norihiko Suzuki, Sae Nemoto, Yasuyuki Saito, Jun Okuz ...Article type: Article
Session ID: ME2015-38
Published: February 21, 2015
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS FREE ACCESSCurator has described by word of mouth to students who visit Shibayama ancient-tomb cluster which is nationally designated historic site. However, it is difficult to understand it deeply only by explaining. So, we are superimposed picture of relics and excavation situations by AR. In addition, we create music piece which matches atmosphere of the place.View full abstractDownload PDF (9883K) -
Sho KATO, Mie SATOArticle type: Article
Session ID: ME2015-39
Published: February 21, 2015
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS FREE ACCESSIn this study, we limit an appreciated object to a painting or picture, and parts of the appreciated object are covered with obstacles. By feature matching between RGB images and a template image of the appreciated object, we search an appreciated object in RGB image. Because the searched object is partially invisible, the template image is perceptively transformed and the searched object in RGB images is overlaid with the transformed template image. However, the overlaid images cause inconsistency about the positional relationship. Therefore, we make translucent images of the overlaid images for perceiving correct positional relationships. In addition, we evaluate effectiveness of the overlaid and translucent images.View full abstractDownload PDF (11801K) -
Keita MINAMIYAMA, Sota SUZUKI, Mie SATOArticle type: Article
Session ID: ME2015-40
Published: February 21, 2015
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS FREE ACCESSIn this study, we examine shadow casting of a virtual object for an AR system focused on grabbing with a bare hand. In this system, we obtain depth information of user's thumb and forefinger fingertips and those shadows with a three-dimensional measure. We estimate the position of a light source in the real space by finding two straight lines each of which links a pair of fingertips and its shadow. In addition, by setting a virtual light at an estimated position we draw a shadow of a virtual object that corresponds to the real environment.View full abstractDownload PDF (3746K) -
Aya KAMEI, Takuya TAKANOHASHI, Mie SATOArticle type: Article
Session ID: ME2015-41
Published: February 21, 2015
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS FREE ACCESSRecently, Image projection on a moving object have been studied. In this study, we register a 3D point cloud of the entire 3D shape of an object as a template. By matching the templates with a 3D point cloud of the object that is measured in real time, we fit Images to the position and direction of the moving object. For speed-up of the matching, we extract feature points from the 3D point cloud of the entire 3D shape. In this paper we experiments three methods of extracting feature points and we determine the nearest method which can extract feature points that human choose.View full abstractDownload PDF (11673K) -
Takahiro TANAKA, Mie SATOArticle type: Article
Session ID: ME2015-42
Published: February 21, 2015
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS FREE ACCESSImmersive image projection has been studied that gives an immersive feeling by projecting images around a viewer. In the preceding study, a surrounding image projection environment was built with two projectors attached with fisheye lenses. However, this environment had a problem that the luminance distribution was not uniform in the projected surrounding image. In this paper, we decrease the dispersion in luminance for the surrounding image projection.View full abstractDownload PDF (10670K) -
Kentaro YAMADA, Masashi OKUDAIRAArticle type: Article
Session ID: ME2015-43
Published: February 21, 2015
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS FREE ACCESSIt is difficult for visually impaired persons to recognize crosswalks, since they couldn't find correct locations of crosswalks. In this paper, a method is proposed to recognize crosswalks with a smartphone to notify them with voice guidance. The location is to be decided using a mobile photo with a smartphone and an aerial photo fetched with GoogleMapAPI based on smartphone's GPS information. Lines of crosswalks in both photos are extracted by image processing procedures such as binarization, edge detection and Hough transform. Then, the appropriate crosswalk location is decided with the lines by matching mobile photo lines and aerial photo ones. Using the additional aerial photo, recognition rate is improved compared with mobile photo based recognition, even if there are some occlusions in the mobile photo due to roadside trees, shrubberies, guardrails, cars, and so on. Experimental results with 125 images show 63% correct detection rate. It takes about 6 seconds for processing totally.View full abstractDownload PDF (13967K) -
Min GAO, Daisuke TAKAHASHI, Noriyoshi OKAMOTOArticle type: Article
Session ID: ME2015-44
Published: February 21, 2015
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS FREE ACCESSRecently, purposing of prevent the crimes, monitor's function were developed. It could also detect special person. But the overlap in a few of people is a problem. These researches investigate that used depth image and color image can be taken camera to detect the person in the overlapped situation. In this research, based on the preceding test we detected person in the real situation. Furthermore, on the base of preceding research we make the person determine standard that can make results better.View full abstractDownload PDF (4786K) -
Kensuke Sakata, Daisuke Takahashi, Noriyoshi OkamotoArticle type: Article
Session ID: ME2015-45
Published: February 21, 2015
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS FREE ACCESSThis paper presents a writer authentication method is for handwriting by fingers. In particular, our methods study corresponds to the forged handwriting, because that is one of the important problems in writer authentication. Our technique get 25 test subject's writing information that is shape, degree and path of fingers by motion sensor. Moreover, our technique examine that there are differences among individuals between each test subjects to compare each test subject's features. As a result, subjects have same features as target subject are, 6 subjects in case of comparing degree, 9 subjects in case of comparing shape, 9 subjects in case of comparing path of finger. Comparing all, we can't find that test subjects have same feature as target test subject's feature.View full abstractDownload PDF (3494K) -
Hitoshi MIYAZAKI, Tomoaki NAKAMURA, Masahide KANEKOArticle type: Article
Session ID: ME2015-46
Published: February 21, 2015
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS FREE ACCESSThis paper proposes a method to generate robot's behaviors representing emotions reminded from the given speech. The robot learns the relationship between words contained in speech and behaviors by using the machine learning technique. This makes the robot possible to generate its behaviors automatically that correspond to spoken words. The robot's behaviors will become more human-like and this will widen the application fields of the robot. The proposed method has several advantages, that is, the generation of robot's behaviors from the speech containing unknown words, the variety of generated behaviors even for the same speech, and the mutual estimation from speech to behavior and vice versa. The effectiveness of the proposed method has been shown through the subjective evaluation experiments of robot's behaviors generated by the proposed and previous methods.View full abstractDownload PDF (10132K) -
Jun Hui PANG, Tomoaki NAKAMURA, Masahide KANEKOArticle type: Article
Session ID: ME2015-47
Published: February 21, 2015
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS FREE ACCESSA number of research topics have been proposed on the mechanism of face cognition. One interesting topic involves the identification of types of facial features that attract more attention and stay longer in memory when humans observe faces. This paper examines this topic using subjective evaluation experiments that facilitate the judgment of resemblance between two face images. This approach not only focuses on particular facial parts such as eyebrows, eyes, mouth and facial contour, but also it examines combinations of multiple facial parts. We also compare parallel and sequential presentations of two faces to identify the peculiar shape features of facial parts that stay longer in our memory. Experimental results indicate that there is a certain tendency between facial features and their cognition by humans.View full abstractDownload PDF (784K) -
Nei KOU, Tomoaki NAKAMURA, Masahide KANEKOArticle type: Article
Session ID: ME2015-48
Published: February 21, 2015
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS FREE ACCESSThe authors have developed an automatic caricature system using the principal component analysis for facial feature points. Because of the various influences caused by differences in individual persons and shooting conditions, automatically extracted facial feature points sometimes contain the wrong results. This paper proposes a new method to probabilistically evaluate the wrong extraction of feature points of mouth based on the shapes represented by each of principal components and statistical distribution of principal component scores. Furthermore the misdetections of feature points are corrected by reflecting the characteristics of principal components. The experimental results show the effectiveness of the proposed method for the mouth.View full abstractDownload PDF (7350K) -
Yuki HAMAGUCHI, Naoki HASHIMOTOArticle type: Article
Session ID: ME2015-49
Published: February 21, 2015
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS FREE ACCESSRecently, an image representation method called projection mapping has a lot of attention. This method can change the appearance of real objects by projecting textures or shadings to real objects. In general projection mapping, distorted image are sometimes observed because observer's viewpoint is spacially-fixed. Some of previous researches are considering actual observer's viewpoint, but observers have to wear a sensor or 3D glasses for the viewpoint measurement. Therefor, in this research, we propose a projection mapping system by using dynamic viewpoint tracking without any wearable sensors and glasses. This method can achieve real-time and exact tracking of viewpoint, and provide motion parallax effects with a quite natural feelings.View full abstractDownload PDF (8849K) -
Koki KOSAKA, Naoki HASHIMOTO, Hiroshi SANKOH, Sei NAITOArticle type: Article
Session ID: ME2015-50
Published: February 21, 2015
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS FREE ACCESSAs a technique for projecting an ideal appearance to a surface whose reflection properties are non-uniform, radiometric compensation methods with a projector-camera system have been proposed. These techniques modify the colors of a projection image in order to display our desired appearance on the non-uniform surface. However, they have problems which is that unexpected luminance would be projected in the region where the reflection characteristic are changed steeply, and projected image would gradually become to be noisy whenever correction process is repeated. Therefore, in this research, we propose accurate radiometric compensation of a projector-camera system by using phase shift method. It is possible to apply the relationship between the projector pixels and the camera pixels, with sub-pixel accuracy. We also evaluate our method by comparing with the conventional method using a gray-code projection.View full abstractDownload PDF (12729K) -
Haruka JIBIKI, Toshikazu SHIMIZU, Yasuyuki SAITOArticle type: Article
Session ID: ME2015-51
Published: February 21, 2015
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS FREE ACCESSThis paper examine automatic turning over music sheets by gesture recognition. Sometimes player is not turned up music sheets due to occupied both hands in a musical performance. So player asks turning staff especially in piano. However, to practice with staff is rare case. Precise timing of page turning and suitable opening page depend on staff skill. Nodding is a typical signal for page turning in actual playing.View full abstractDownload PDF (10580K) -
Kohei FUJII, Yasuyuki SAITOArticle type: Article
Session ID: ME2015-52
Published: February 21, 2015
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS FREE ACCESSIn recent years, dance has become compulsory unit in the compulsory education organization. However, the leading system of dance education is not perfection. The purpose of this study is to assist choosing music to dance. And development of a system that converts an arbitrary existing music into a dance music. Then we contribute to the dance education. First, we develop an application of arrangement for dance music from MIDI file.View full abstractDownload PDF (454K) -
Misato NISHIZAWA, Asako SOGA, Kunihiko ODAArticle type: Article
Session ID: ME2015-53
Published: February 21, 2015
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS FREE ACCESSWe have developed a visualization system of angle information for classical ballet motion. The system calculates pelvic obliquity and hip joint angles per frame using 3-dimensional coordinate data of markers captured by an optical motion capture system and visualizes them by plane CG objects and fan shape CG objects. These CG objects are shown in conjunction with human body animation. Plural human body animations are arranged and reproduced. To evaluate an optimum method to visualize the characteristics, 10 examinees tested our proposed system.View full abstractDownload PDF (816K) -
Mio TOKUDA, Asako SOGAArticle type: Article
Session ID: ME2015-54
Published: February 21, 2015
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS FREE ACCESSIn this research, we developed a system that retrieves motion data including postures to the user inputs using Kinect. Similar poses to the user input are retrieved by comparing key poses extracted from motion data. As feature values of human body postures, the distances for each axis between both wrists and both ankles and the distance for Y axis from the pelvis to the knees and the positional relationship of the wrist and the center of the spine are calculated. Those patterns are used to retrieve poses, and key poses of motions in accord with the inputted patterns are extracted from the database in real time.View full abstractDownload PDF (6534K) -
Akiro NAKAHIGASHI, Asako SOGAArticle type: Article
Session ID: ME2015-55
Published: February 21, 2015
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS FREE ACCESSWe developed a walk-through system in 3D space using Kinect to support exhibitions at museums. This study reproduces the arrangement of the sliding doors of a temple in a 3D space and expresses the perspective of a person who is virtually walking using Kinect in 3D space. This system uses the position data of the head, the feet, the hands, the shoulders, and the spine that are acquired by a Kinect and recognizes five poses and three motions. Depending on the recognition results, the user can change the perspective, open or close the sliding doors, roll up the bamboo blinds, and light or extinguish a candle in 3D space.View full abstractDownload PDF (7558K) -
Hiroaki Murakami, Shogo TokaiArticle type: Article
Session ID: ME2015-56
Published: February 21, 2015
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS FREE ACCESSWe acquire figures as textures from multiview videos, and do free view point browsing with them by a billboard method. The figures acquired are selected and switched according to the location of one's eyes. We explain the above method with application example on a sports scene.View full abstractDownload PDF (10555K) -
Chihiro Tsutake, Yutaka Nakano, Toshiyuki YoshidaArticle type: Article
Session ID: ME2015-57
Published: February 21, 2015
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS FREE ACCESSAlthough the intra prediction in H.265/HEVC standard offers 35 prediction modes including the planar, DC, and 33 angular modes to improve the coding efficiency, the calculation cost in the intra prediction is accordingly increasing. This paper proposes an estimation technique for the optimal intra-prediction mode based on the reliability metric for motion vectors. The experimental results given in this paper have indicated the advantage of the proposed technique.View full abstractDownload PDF (5880K) -
Takatoshi MATSUKURA, Shoji YAMAMOTOArticle type: Article
Session ID: ME2015-58
Published: February 21, 2015
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS FREE ACCESSRecently, a surveillance camera system is widely introduced in public space, street, and so on, since safety and security of life is important for everyone. In order to capture more detail information, a High-Definition camera is useful for the next-generation security system. However, long periods of surveillance image by HD camera are necessary enormous storage mediums, and it causes an increase in cost. Therefore, we propose an applicable surveillance system by using multi cameras, which can perform differently-scaled observation. This paper describes a simple approach for the calibration of multi camera system. We use non-parametric matching for lens distortion and rigid-body transformation.View full abstractDownload PDF (12705K) -
Akihiro MATSUFUJI, Hideaki HONDA, Shoji YAMAMOTOArticle type: Article
Session ID: ME2015-59
Published: February 21, 2015
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS FREE ACCESSFor the wide spread of domestic robot, an omnidirectional camera is expected to be effective tool for recognizing the structure and equipment of indoor scene. Since this camera has an advantage of detecting both disposition and profile which are shape and texture, it is very useful information for identifying objects. However, it is difficult to acquire the actual arrangement, because the omnidirectional camera can capture only directional information. Therefore, we propose a solution approach to specify the relative arrangement of structure and equipment by using all circumferences laser line. Our approach challenges auto identification of 4 corners from captured image, and image mapping to be generally consistent with actual indoor scene.View full abstractDownload PDF (15188K) -
Kentaro HIRABAYASHI, Tomoaki NAKAMURA, Masahide KANEKOArticle type: Article
Session ID: ME2015-60
Published: February 21, 2015
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS FREE ACCESSAn accurate detection of facial features is crucial to analyze them and to synthesize impressive caricatures by computer. A lot of studies on automatic detection of facial features have been done until now, however, accurate detection methods which can be used in above-mentioned applications have not been developed yet. This paper proposes automatic extraction methods of hair and contours bf eyebrows, which are robust for influences by variations in individual persons and shooting conditions. Especially, an automatic extraction method of hair under strong illumination and that of eyebrows on which hair is partially overlapped are examined. These cases have not been handled properly by previous methods. The proposed methods are more practical than the previous methods since they can extract hair and eyebrows correctly and robustly under various conditions.View full abstractDownload PDF (12042K) -
Ryotaro TAKAOKA, Naoki HASHIMOTOArticle type: Article
Session ID: ME2015-61
Published: February 21, 2015
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS FREE ACCESSRecently, techniques using a depth camera measuring the distance to target objects attract attention. By using a depth camera, we can simplify recognize depth and three-dimensional objects having difficulty in distinction only with color images. However, the resolution of range image provided from the depth camera is lower than the color image which is generally provided, and there is a limitation that the information of the color image is reduced when we use that with the depth image combined with a color image. In this study, we realize the super-resolution of the depth camera by using a high-resolution color information.View full abstractDownload PDF (13275K) -
Ryo KOIZUMI, Daisuke KOBAYASHI, Naoki HASHIMOTOArticle type: Article
Session ID: ME2015-62
Published: February 21, 2015
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS FREE ACCESSRecently, spatial augmented reality (SAR) have been used as a method of editing appearance of real objects by using a projector. In addition to SAR with stationary objects, SAR with moving objects is paid attention to in case of entertainments, work supports, design study, and so on. Although the dynamic SAR requires pose estimation and tracking, it cannot accomplish enough accuracy and processing speed by using non-contact measurement by cameras. Therefore, in this research, we propose optimization of point cloud data used for the tracking at off-line and online processes, and accomplish acceleration of the SAR system which we have achieved with a depth camera.View full abstractDownload PDF (14257K) -
Article type: Appendix
Pages App1-
Published: February 21, 2015
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS FREE ACCESSDownload PDF (81K) -
Article type: Appendix
Pages App2-
Published: February 21, 2015
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS FREE ACCESSDownload PDF (81K) -
Article type: Appendix
Pages App3-
Published: February 21, 2015
Released on J-STAGE: September 22, 2017
CONFERENCE PROCEEDINGS FREE ACCESSDownload PDF (81K)
- |<
- <
- 1
- >
- >|