Reports of the Technical Conference of the Institute of Image Electronics Engineers of Japan
Reports of the 264th Technical Conference of the Institute of Image Electronics Engineers of Japan
Showing 1-30 articles out of 30 articles from the selected issue
Reports of the 264th Technical Conference of the Institute of Image Electronics Engineers of Japan
  • Yoshiharu Nakamura, Masahiro Masuzoe, Shuichi Enokida
    Session ID: 12-03-01
    Published: February 21, 2013
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    By the pattern matching method generally used in a defects verification of a leadframe which are some semiconductor parts, an exact positioning process is required. However, it is not able to respond to the partial shift by individual specificity in matching with a temporal image of good condition. Therefore, in this research, a defects verification method of temporal images needlessness is proposed. In our proposal method, the linearity of the leadframe is assumed. And it is also assumed that the variance of oriented gradients in the local image which includes defect area is wide. In our method, the variance is calculated by Harris corner detection algorithm. In some experiments to assess our proposal method, a 30-micrometer defect was detected more than 95%.
    Download PDF (5611K)
  • Masakazu Sadano, Wataru Toyama, Shuichi Enokida, Toshiaki Ejima, Toshi ...
    Session ID: 12-03-02
    Published: February 21, 2013
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    In this paper, we have proposed a high-performance three-dimensional recognition method based on the idea of generation and verification of the reference point cloud, targeting the bulk of industrial components. The position estimation of reference points cloud is considered as posture recognition of the corresponding industrial component. Using the principles of the generalized Hough transform, the positions of the reference point group are estimated. In experiment of industrial components recognition, it is shown that both precision and recall increase. This indicates that interference suppression and verification derived from reference point cloud make the performance of object recognition better.
    Download PDF (1091K)
  • Keisuke NAKAZAWA, Kota AOKI, Hiroshi NAGAHASHI
    Session ID: 12-03-03
    Published: February 21, 2013
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Shape modeling is an important task for object recognition,detection,and deformation.In recent years,machine learning methods are significantly developed that find secret regularities and potential variables in given data.This paper proposes shape representation of object contours by using a cyclic Self-Organizing Map that is one of the unsupervised learning method.It is difficult for a traditional cyclic SOM method to represent odd-shaped object contours.We address this problem by using the order of points located on the order of SOM units.We also propose methods of extracting shape features from object contours.Our method is examined by the experiments of silhouette recognition.
    Download PDF (514K)
  • Shota Fujii, Shuichi Enokida, Toshiaki Ejima
    Session ID: 12-03-04
    Published: February 21, 2013
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Mean-Shift, one of well-known method as tracking algorithm, utilizes a color histogram as the amount of feature of the non-rigid object. The aim of this research is to give to the tracking method the capability of dealing with a luminance variation. Observed luminance values are changed linearly with respect to the variation of the luminance intensity. Thus, a framework which has the capability of transforming these changed values by the luminance variation into same amount of feature will realize a robust tracking against luminance variation. In this research, a subspace consists of samples which are obtained by observing the same reflectance (color) object at several luminance intensity conditions. In a similar way, typical plural colors are selected so as to construct color subspace, respectively. Then, the distances of projecting the input luminance value into each subspace respectively are utilized as the amount of feature. The experimental results showed that Mean shift tracking based on the histogram which is calculated by the distances to plural (color) subspace is effective to the luminance variation.
    Download PDF (4915K)
  • Mei KODAMA, ZhiZhi YU
    Session ID: 12-03-05
    Published: February 21, 2013
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    As video content applications, video search techniques are often used. The techniques of video identification methods are required for the various applications, such as, duplicate downloads and some copies of the same content. They are also used as the guarantee methods of correctness content from the aspect of copyright protection. Then, we focus on a video identification method as video retrieval techniques. There are some retrieval methods, which treat the spatial and the temporal information, such as, color, histogram of the signal, the shape information, and so on. They have the problem of the processing time of calculating the feature values. Accordingly, in this paper, we proposed the video identification system used motion vector information as video retrieval systems and explained them. As motion characteristics, we focus on the histogram of motion vectors. Moreover, we proposed two methods as feature matching methods. The one is the quantized mv histogram method. The other is the region extraction method using the histogram information, in which, we calculate the local maximal values in descending order. We present the effects of proposed video retrieval schemes using representative vectors by the simulation experiments and consider them.
    Download PDF (2489K)
  • Takuya TANOUE, Yoshiaki TANIGUCHI, Go HASEGAWA, Hirotaka NAKANO
    Session ID: 12-03-06
    Published: February 21, 2013
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    In recent years, video processing-based methods for counting pedestrians in a crowded scene have been proposed. In these methods, the relationships between various features in video sequences and the number of pedestrians are obtained using training data. Then, the number of pedestrians is estimated from the video features and the pre-learning relationships. However, for each method, different video features are used and its performance is evaluated under different environments. Therefore, it is difficult to compare the effectiveness of video feature for estimating the number of pedestrians. In this report, we investigate the relationships between various video features and the number of pedestrians using multiple regression analysis under same evaluation environments. We use the size of moving objects, the form of moving objects, the number of optical flows, and the number of clusters of optical flows as video features. Through analysis using actual video sequences, we show that combining the number of optical flows and form of moving objects is effective for estimating the number of pedestrians under crowded situations.
    Download PDF (4259K)
  • Makiko NAGASAWA, Yoshinori HATORI, Mitsugu KAKUTA, Tadao HAYASHI, Yosh ...
    Session ID: 12-03-07
    Published: February 21, 2013
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    A method of analyzing a smash motion of a badminton player from an image is proposed. The proposed method is constituted by KL transformation and mapping. First, the proposed method expresses a motion of a part with the 1st, the 2nd, and the 3rd principal axes obtained by KL transform in time. Next, mapping is performed in velocity space. This proposed method roughly divides the motion of the part to the whole body into a closed loop, a curve, and a point (or a straight line). Analysis of time change of the motion is conducted. In this paper, computer simulation shows the theoretic verification result of the proposed method. Next, an experiment using an image by a Japanese elite player shows an application possibility of the proposed method. Furthermore, the method of analyzing for discovering a new motion from a motion of the sports player of international match level is considered from a viewpoint of sports science.
    Download PDF (509K)
  • Kyohei HOSONO, Kota AOKI, Hiroshi NAGAHASHI
    Session ID: 12-03-08
    Published: 2013
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    In this paper, we propose a method for dance analysis of TV programs or videos on the Internet. These videos have various lighting, angles, dancers, or camera movements. And most of them contain a variety of scenes. To analyze incomplete and insufficient dance information, we use spatio-temporal features. We focus on periodicity of dance motion, extract spatio-temporal interest points, and synchronize the dances of different videos. Experimental result shows that the proposed method can analyze the dance movements, and compare with each other.
    Download PDF (798K)
  • Ouji YAMAGUCHI, Takayoshi SUKETOMO, Yoshiki MIZUKAMI, Katsumi TADAMURA
    Session ID: 12-03-09
    Published: February 21, 2013
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    In this paper, we reconstruct 3D structure by stereo matching using segmentation. We optimize the segments in order to improve the accuracy of disparity based on a cost function. In particular, by comparing the 3D reconstruction result of penalizing discontinuity of the segment and that of the disparity, we discuss their accuracy and computing time.
    Download PDF (1352K)
  • Yusuke UCHISAWA, Akio KIMURA
    Session ID: 12-03-10
    Published: February 21, 2013
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    This paper describes several methods for extracting 3D geometric information such as a height of an object, an angle formed by two space lines, and a distance between two points on a plane, from a single uncalibrated image using various scene constraints. First, we estimate a camera projection matrix using the homography of a reference plane and two imaged points whose heights from the reference plane are both already known. Second, we extract some kinds of 3D entities from the image with the estimated matrix. Specifically, we propose that a plane that intersects the reference plane with an arbitrary angle can be determined from an imaged segment whose length on the plane is already known, or from two imaged space lines whose intersection angle is already known. Furthermore, we show a procedure for determining planes adjacent to the reference plane sequentially. Finally, we show that satisfactory results are obtained in the experiments conducted to verify our proposed methods using simulated data and real images.
    Download PDF (1216K)
  • Hidehiko OKUBO, Mahito FUJII
    Session ID: 12-03-11
    Published: February 21, 2013
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    We propose a new video segmentation method for the sake of interactive object extraction in video compositing. For interactive extraction, the number of segmentation regions should be lesser to reduce the computation. Our method is based on mean shift segmentation, and has improvements to separate similar-colored adjacent regions belong to different objects, even with smaller number of segmentation regions. In order to achieve such segmentation, our method analyzes transition of all pairs of preliminary segmented adjacent regions. We demonstrate the effectiveness of our segmentation method by extraction examples.
    Download PDF (857K)
  • Takuya Okamoto, Shuichi Enokida, Toshiaki Ejima
    Session ID: 12-03-12
    Published: February 21, 2013
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Using bundle adjustment, we have tried to restore structure from images captured by cars running straight. However, due to relation between optical axis and car direction, some corresponding feature pairs, which are extracted from the images, are not suitable to restore the 3D point. Taking geometrical relation between optical axis and car direction into consideration, we have designed a filter which is able to exclude unsuitable corresponding pair. It has experimentaly shown that the designed filter makes restored structure quality better.
    Download PDF (2968K)
  • Yuan Qu, Yue Bao, Fei Gu, Haruki Komura
    Session ID: 12-03-13
    Published: February 21, 2013
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Integral Photography expression is most promising in the autostereoscopic display technology. REGZA 3D television with nine parallaxes was developed by Toshiba. However the nine parallaxes are not enough to display 3D image with high quality. In order to increase the number of parallax, a 3D video display system using two high-resolution panel was also proposed. However, the configuration of this system is complicated, and the utility is low. In this paper, we propose a three-dimensional flat-screen video display system using transparent display without using a high-resolution display panel. It is a method based on the IP using sparse lenticular lens. The depth of the object to be displayed is depending on the division of the region. The effectiveness of the proposed method was confirmed. That the three-dimensional display can provide a deep depth is confirmed by the experiment using a tablet PC and an OHP film instead of a transparence display.
    Download PDF (561K)
  • Xiaodong Qu, Yue Bao
    Session ID: 12-03-14
    Published: February 21, 2013
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    In the three-dimensional display technique, integral photography method using a light reproducing and lenticular lens method using stereoscopic images using parallax images are well known. However, they have the problem of feeling visual fatigue. The holography technology can solve these problems; however, the system is complex and large. This paper proposes a method to downsize the holographic system using a bullet type LED and a high-definition LCD panel.
    Download PDF (677K)
  • Ryo Harashima, Yue Bao
    Session ID: 12-03-15
    Published: February 21, 2013
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    The Head-Mounted Display (HMD) is a method to enjoy big screen 3D. A problem of HMD is that the motion sickness is induced visually. For reducing the visually induced motion sickness of HMD, this paper proposes a method to fix a virtual TV in real space. To fix the virtual TV in real space, a template matching is used for matching between feature points extracted from the image of right camera and left camera. By the experiment results, the proposal method can reduce the visually induced motion sickness are confirmed.
    Download PDF (550K)
  • Osamu KOGA, Keisuke NONAKA, Tomonobu YOSHINO, Yoshinori HATORI
    Session ID: 12-03-16
    Published: February 21, 2013
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    For content aware convert from a rectangular image to an arbitrary-shaped image, an image retargeting method is proposed that consists of two phases; moving important regions into the watching region and reconstructing the rest of the image (background), where the latter one is extremely slow. In this paper, we propose a fast method for the reconstruction. Our method is based on the fact that the background is continuous in the original image and it can be determined from the nearest important region. We use graph cut to estimate the background around the important region, and it realizes fast reconstruction.
    Download PDF (2097K)
  • Keisuke HAYASHI, Koki FUJIMOTO, Hirohisa FUJISAKI
    Session ID: 12-03-17
    Published: February 21, 2013
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Ultrasonic diagnostic equipment can observe internal organs in the part of in vivo by the noninvasive. The image processing of the ultrasound imagery did the contrast emphasis and the local pixel neighborhood searching. The area value of right and left ventricles was measured from this processing image. The cross correlation between right and left ventricles was requested from this area value. As result, a short amount time phase lag was admitted for the area value between right and left ventricles.
    Download PDF (615K)
  • Takuya YAMAMOTO, Hirohisa FUJISAKI, Toyoshi SOGABE
    Session ID: 12-03-18
    Published: February 21, 2013
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Recently, the medical image such as X-ray CT, MRI image and etc. is essential to the clinical diagnosis. The MRI image can be scanned from various angles. On the other hand, X rays CT image can only scan the image of the horizontal cross-section. The doctor observes X-ray CT image, and is judging the three-dimensional architecture. The interpolation image data was inserted in X-ray CT image data. The constructed system can display an arbitrary cross-section from three-dimensional data including the interpolation image.
    Download PDF (536K)
  • Katsutoshi MIYAZAKI, Yoshiki MIZUKAMI, Katsumi TADAMURA
    Session ID: 12-03-19
    Published: February 21, 2013
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    The large-size digital signage of high-intensity is expected to be located in every show window and shop front along a street in urban areas in the near future. There is a possibility that this becomes the optical stimulus to a walker at night. However, for the present, since the regulation on both the quantity of luminous flux of a display and the content of output does not exist, it may escalate in the tendency which is going to draw a walker's attention. More specifically, the content of display is expected to become a frequent-change movie and a strong-contrast image. This suggests that the digital signage will cause the occurrence of danger of harmful optical stimulation, so-called Pokemon shock. In this paper, we made a basic study about the matters required in order to avoid such a situation and to hold the safe illumination environment in a urban area, and considered about the validity of it.
    Download PDF (493K)
  • Eiji koshida, Katsuya Kashiwagi, Hirohisa Fujisaki
    Session ID: 12-03-20
    Published: February 21, 2013
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    The living body keeps the homeostasis to the protective response for stresses. The present study measured the palma manus temperature at smoking, the beverage, and sphygmomanometry. The palma manus temperature changes were compared according to the stress. Processing of the difference between images was done, and the temperature change of the hand palm was observed.
    Download PDF (483K)
  • - The Changes of Pupil Parameters with Cold Stimulus -
    Ayako KATOH, Taiki SHIMADA, Naoki KOBAYASHI, Yasuhiro FUKUI
    Session ID: 12-03-21
    Published: February 21, 2013
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    We have developed a non-wearing type of pupillography for estimation of stress level. The pupillography can measure pupil area with a resolution of 5.63μm/pixel at 200fps. And we proposed a blink noise rejection method to analyze pupil fluctuation in frequency domain or time-frequency domain. With the video-type pupillography, blink noise is overlapped with pupil area signal. After blink intervals are detected with derivative value of pupil area, the values of the intervals are rejected and interpolated with cubic spline interpolation. We conducted cold stimulus test. As the results, discrete Fourier transform of time-variant signals of pupil area show that the overall values change in response to changing autonomic nerve activity.
    Download PDF (689K)
  • Kazunori SHIMAMURA
    Session ID: 12-03-22
    Published: February 21, 2013
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    The internet has penetrated as the most important social infrastructure. It brings the convenient world called the wired city by way of combining with the mobile technologies like WiFi and GPS system. But the circumstances of the current communication technology are still limited. This article describes the system scheme called as an off-lined network which enables to visualize the un-visible covered cables and to sense the location information where the GPS satellites are difficult to be found out. The numbers of passive RFID tags are attached to the current internet as the virtual off-lined network layer components. This technological architecture might bring the tipped city infrastructure.
    Download PDF (550K)
  • Msafumi KOSUGI, Osamu UCHIDA
    Session ID: 12-03-23
    Published: February 21, 2013
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    An effective combination of music and video images generates efficacious interaction between auditory and visual processing. Then, these techniques are used at concert halls, on promotional videos, or on TV commercials. However, the professional knowledge and/or tools are necessary to create video images considering the interaction between auditory and visual processing. Therefore, it is difficult to create such videos easily. In this paper, we propose a method to create a video image well-suited to the music based on its characteristics automatically. In this study, we suppose that the low-pitched and the high-pitched volume of the music indicate the heaviness and the intensity of it, respectively. In the proposed method, the color and the shape of the object show up in the video are decided by the music characteristics (the heaviness and the intensity of the music). The results of the verification experiment show the effectiveness of the method.
    Download PDF (454K)
  • Makoto J. HIRAYAMA
    Session ID: 12-03-24
    Published: February 21, 2013
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Implementations of audio descriptions into movies for visually impaired persons are explained. Trends on audio descriptions to television programs in U.S., U. K., and Japan are reviewed. Then, audio descriptions mainly in theaters are explained. Issues on some methods for presentation, transmission, audio and video synchronization, and authoring are discussed. Finally, a presentation method using bone-conducted headphones is proposed.
    Download PDF (305K)
  • Sezan ISEDA, Shohei TAKAHASHI, Takeshi OKUYA, Jun OHYA
    Session ID: 12-03-25
    Published: February 21, 2013
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    This paper analyzes causes of determining whether people go to theaters to appreciate movies, based on narratology, focusing on trailers. That is, this paper proposes a hypothesis plot based on narratology and explores relationship between people's determination of movie appreciations and consistency of trailers and the hypothesis plot, using subjective tests. The subjective tests let subjects give six-level scores, which represent importance for determining movie appreciations, for each of nine causes for the determination, including trailer, for each of 10 movies. As a result, if the organization of a trailer fits well to the hypothesis plot tends to let people decide to go to the theater. On the other hand, in case of very well-known movies, this tendency is not observed; future works such as more detailed and statistical analyses are needed.
    Download PDF (210K)
  • - The metadata grant technique to a baseball live broadcast -
    Hiroyuki SAKAI, Yoshinori HATORI, Nobuko AKAI
    Session ID: 12-03-26
    Published: February 21, 2013
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Metadata grant for multi-use of images by image recognition technical applications
    Download PDF (356K)
  • - Representation of the view motion taking the state of snowy road into account -
    Naoto SUMIYA, Yoshiki MIZUKAMI, Katsumi TADAMURA
    Session ID: 12-03-27
    Published: February 21, 2013
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    We propose an expression method of the slip motion on the snowy road on the 3DCG-based driving simulator. Specifically, we developed the following two methods for expressing the slip motion on the frozen road surface: (1) A simulation model which realizes the state transition of road surface in winter by using the temperature and the weather condition as parameters. (2) A method for expressing the view motion change in a frame for the driving simulator based on the result of the slip situation on every tire. We implemented the proposed method and confirmed the usefulness of it.
    Download PDF (580K)
  • Yuya ARAI, Osamu UCHIDA
    Session ID: 12-03-28
    Published: February 21, 2013
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    In recent years, opportunities of taking photographs have been increasing owing to the popularization of digital cameras, smart phones, and SNS that we can post photos easily. Because of the convenience of taking photos using digital cameras, we often take many photos of a single object. However, it is difficult to select the best shot when the number of photos is large. Therefore, in this study we propose a method to select the best shot from a group of photos that show same object. In the proposed method, first we estimate the color of the photographic subject from the center region of one of the photos selected from the group of photos. Then, we conduct region extraction having the color estimated as the photographic subject color for all photos within the photo group. Next, we compute the sharpness of the extracted region on every photo that is assumed to be photographic subject, and select the best shot by comparing the values of the sharpness. We conducted the experiments to verify the effectiveness of the proposed method and got good results.
    Download PDF (753K)
  • Hidekazu Sekizawa, Masahiro Suzuki, Kazutake Uehira
    Session ID: 12-03-29
    Published: 2013
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Color fringing at the edges of the character is an issue in the color reading device. If you do not have a problem with black text is printed in solid black in particular, is very difficult to see the image, such as being due to misalignment of the printing process. Precision ones without chromatic aberration is required for an optical system such as a lens. When the edge enhancement processing is performed to compensate for a decrease in the response characteristics of reading systems, I will further emphasize the color fringing. I tried to report that the new low color noise image enhancement processing can be reduced at the same time be carried out color bleeding edge enhancement in this report.
    Download PDF (700K)
  • Shota SAZAWA, Atsushi OSA, Hidetoshi MIKE
    Session ID: 12-03-30
    Published: February 21, 2013
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Solar power generation has many advantages over other renewable energy, while there are many problems in order to spread it. In addition to the improvement of power generation efficiency and the reduction of manufacturing cost, power generation prediction is important for spreading it. Although an amount of power generation is approximately proportional to the volume of sunshine duration in the region, prediction accuracy would be significantly reduced in areas such as AMeDAS data is not accumulated. In this study, we developed a method to estimate the volume of sunshine duration using weather satellite images. We devised a formula of the estimation model for Miyazaki city as an example. As a result, we confirmed the possibility of estimating the volume of sunshine duration using weather satellite images.
    Download PDF (604K)
feedback
Top