Reports of the Technical Conference of the Institute of Image Electronics Engineers of Japan
Reports of the 260th Technical Conference of the Institute of Image Electronics Engineers of Japan
Showing 1-40 articles out of 40 articles from the selected issue
Reports of the 260th Technical Conference of the Institute of Image Electronics Engineers of Japan
  • Akinori SHINDO, Yuta GOTO, Osamu UCHIDA
    Session ID: 11-04-01
    Published: February 24, 2012
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    In recent years, two-dimensional code typified by the QR code is spread widely and used in various situations. These two dimensional codes generally attached to printed materials such like posters and magazines, but they are harmful to design of printed materials because it is inorganic pattern. To solve such problems the method is proposed which keep design quality by embedding information invisibly with technique of digital watermarking. For example, Moroo et al. proposed a data embedding method for printed material by changing the value of yellow component. However, the method has problem that users may not notice the fact that some information is embedded in printed materials. To solve the problems Goto and Uchida proposed a method that generating 2D code considering design and visibility that buried the information by changing the minute of colors what is used to generate 2D code in 2010. In this paper, we conduct verification experiments for extracting information from printed materials captured by cell phone attached camera. The results of the experiments indicate the usefulness of the method.
    Download PDF (383K)
  • Mei KODAMA
    Session ID: 11-04-02
    Published: February 24, 2012
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    We often use video content over the high speed network. As the data is larger than text information, the data delivery methods using data caching are studied. There are insufficient schemes for video content, and the content management methods are important. Then, we propose the management method using caching and delivery based on scalable data. We use the data transition process of scalable structure. This is based on elapsed time and its content quality. In this study, a content utilization service model using mobile and home terminal is de?ned. We consider the efficiency in this method from a viewpoint of multiple quality.
    Download PDF (141K)
  • Kohta Itoh, Atsushi Osa, Hidetoshi Miike
    Session ID: 11-04-03
    Published: February 24, 2012
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    In recent year, localized torrential rains develop every year, they bring an enormous toll. Localized torrential rains often have occurred in meso-scale convective systems organized by thunderclouds. In this case, it's difficult to predict occurring localized torrential rains because that thunderclouds develop in the short time. In this study, we address detection of atmospheric wind-shear, an important evidence of growing thunderclouds in the meso-scale convective system. Wind direction and speed of upper atmosphere were calculated by using echo intensity images, and we compared between wind direction of upper atmosphere and wind direction of lower atmosphere recorded by AMeDAS. "Torrential rain in Tyugoku and Kyushu on July, 2009" was taken up as an example. We could confirm atmospheric wind-shear when thunderclouds making torrential rain arise. The results of this study indicate the possibility that analysis of weather observation images can detect wind-shear in meso-scale convective systems.
    Download PDF (920K)
  • Kohei ARAI, Herman TOLLE
    Session ID: 11-04-04
    Published: February 24, 2012
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Japan Television (TV) Commercials program contains important information that displayed only in seconds. People who need that information has limited time for reading and noted that information. This research work focus on the method for automatically detects text, extract and recognized important information from TV commercials. We propose a method for product information extraction from TV commercial using a combination of hybrid text extraction method and knowledge based system with pattern matching rule based method. Implementation and experiments on 100 commercials screenshot images achieved a high accuracy result on text extraction and "phone number" information recognition.
    Download PDF (400K)
  • Hiroshi OCHI
    Session ID: 11-04-05
    Published: 2012
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Download PDF (423K)
  • Koki SATO, Masataka TOZUKA, Makoto OHKI, Kunihiko TAKANO, Yukio TAKIMO ...
    Session ID: 11-04-06
    Published: February 24, 2012
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    We are considering for transmission of computer hologram data. It is very important for realizing 3D-TV. In this paper we report the experimental results about transmission of hologram data and 3D image reconstruction using white LED light.
    Download PDF (1814K)
  • Takuya YOSHIDA, Akira KUBOTA, Yoshinori HATORI
    Session ID: 11-04-07
    Published: February 24, 2012
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    This paper discusses a view interpolation method for reconstructing an intermediate all-in-focus image from multi defocused different view-point images of a scene consisting of three depths.In previous work, it was found that this method was effective for three images of a three-depth scene. We examined whether this method were effective for multi view-points and a multi-depth scene.As the result, we found that an intermediate image could be generated. However, there are some problems in this case. The main problem is that many camera variables and the amplitude of the derived filters has big values in their higher frequency. Finally we discuss measures solving the problems.
    Download PDF (3464K)
  • Keisuke ISHIMARU, Toshio KAWASHIMA, Kengo TERASAWA
    Session ID: 11-04-08
    Published: February 24, 2012
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    In recent years, digital archives become popular and is open to the public on the Internet. The digital archive is a very effective means, in order to utilize historical documents safely. By exhibiting digitized data on the Internet, degradation of documents can be prevented. By revising digitized data, deterioration of documents can be restored. Scrupulous attention is paid in the photography of the image of the digital archive. In this paper, we remove the shadings which occur because of buckling up with keeping a texture from the scroll picture digital image that shadow is included in. This showed the technique that removed the shadings.
    Download PDF (872K)
  • Yusuke Tsuda, Kazuhumi Kaneda, Toru Tamaki, Bisser Raytchev
    Session ID: 11-04-09
    Published: February 24, 2012
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    We propose a method for storing High Dynamic Range (HDR) images compactly and displaying them quickly taking into ccount spectral distribution. It is important to display images quickly to various kinds of display with different resolutions. To accelerate the display process, we store a 2D spatial component and a 1D spectral component separately in a discrete wavelet transform process. When displaying the stored images, we firstly convert a 1D spectral component into RGB components. In the RGB conversion, we make use of the property of orthonormal-basis functions. Then, we restore RGB images using a part of a spatial component (converted RGB coefficients) epend on the resolution of a display device. Our method makes it possible to efficiently restore and quickly display spectral HDR images in the case of lower resolution.
    Download PDF (2330K)
  • Hidekazu Sekizawa, Masahiro Suzuki, Kazutake Uehira
    Session ID: 11-04-10
    Published: 2012
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Higher resolution of color image input device such as a digital camera is progressing. Improvement of color noise in low light environment has become an issue. Result we have developed a process of the Low Noise Level of a color image using the (correlation of the luminance signal and color signal) limited color space, was applied to the sensor spectral characteristics of LMS, low-light spectral characteristics RGB inspired visual model before with high sensitivity from the sensor, shown in the simulation that the noise reduction is possible. In addition, this low noise processing showed that there is a potential advantage of the fact that the luminance signal and color signals can be exchanged under certain conditions, to be close to par with the resolution of the linear monochrome color scheme color sensor juxtaposition. This time, this a step further, in a two-dimensional array is an array close to the visual characteristics was attempted (L: M:S = 4:4:1: M). This is potentially obtain a higher resolution signal than conventional Bayer array.
    Download PDF (543K)
  • Kunio Ohno
    Session ID: 11-04-11
    Published: February 24, 2012
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Bertrand Russell said that human knowledge should be based on both description and acquaintance. The former seems to need writing skill while the latter seems suitable to be expressed by graphical information. This paper describes the classification to apply various technical document and the reasonable result were proven.
    Download PDF (378K)
  • Yohko ONO, Yoshiki MIZUKAMI, Katsumi TADAMURA
    Session ID: 11-04-12
    Published: February 24, 2012
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    We propose a novel method for the lighting simulation system which can treat a luminaire locating different spectral distributions on the portion of the luminescent surface. The feature of the proposed method is to do the lighting simulation using multiple virtual small light sources with individual distribution of luminous intensity and spectrum. The virtual small light sources are obtained through subdividing the luminescent surface of the luminaire. The conventional lighting simulation method has a problem caused by the limitation of the projection direction of light, that is, a point on the illuminated surface and the direction of light projected from the luminaire is one to one correspondence. We can solve this problem through employing the proposed concept of the virtual small light sources. We confirmed the usefulness of the proposed method through a prototype which implemented the proposed method.
    Download PDF (417K)
  • Kohei Arai, Tran Xuan Sang, Nguyen Thi Uyen
    Session ID: 11-04-13
    Published: February 24, 2012
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    In this paper, we present a multi agent-based simulation model for helping disabled persons in emergency situation. This simulation model allows the interactions among victims and volunteers during the evacuation process. Disabled persons and volunteers are modeled as agents with different behaviors. When the decisions to help disabled people are made, the volunteers will apply flexible technique by making choices between the shortest route and the least congested route according to route network conditions for every simulation time-step. The shortest path algorithm of Dijkstra and the reinforcement learning method of Q-learning algorithm are used to find the appropriate path to reach to the disabled people. In order to represent the route network in more realistic way, the GIS route network model is used. The simulation results indicate that the simulation using the flexible technique has less rescue time than these using two methods respectively.
    Download PDF (153K)
  • Kohei ARAI, Steven Ray SENTINUWO
    Session ID: 11-04-14
    Published: February 24, 2012
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Obstacle avoidance is currently treated by methods that fall into two broad categories, global and local. In this paper, we consider the case that the environment contains obstacles whose speed and direction cannot be predicted. To avoid such kind of obstacles, we introduce a local path planning methodology for a robotics by using Cellular Automata model. This local path planning was combined with Dijkstra shortest path algorithm as global path planning to obtain an optimal route. Using the proposed methodology, a scene in a typical hospital corridor has been created. By using CA model, we evaluated two kinds of obstacle avoidance movement. First, the robotic use the "stop and go" method, which is the robotic will decrease its speed while encounter a transient obstacle. The second one is detour model, in which the robotic will make a detour motion to avoid a transient obstacle. The simulation results show that in dynamic environment that having transient obstacles, the "stop and go" method produced minimal collision with shortest-distance path. While, using the detour model generated minimal collision with time-minimal navigation path.
    Download PDF (734K)
  • Koki FUJIMOTO, Tomohisa YAMAGATA, Hirohisa FUJISAKI
    Session ID: 11-04-15
    Published: February 24, 2012
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    The time-series imagery of right and left ventricles was acquired in ultrasonic diagnostic equipment. The sampling frequency of the ultrasound imagery is 40Hz. The picture 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 a result, a short amount time phase lag was admitted for the area value between right and left ventricles.
    Download PDF (530K)
  • Yuki MIYAZAKI, Kota AOKI, Hiroshi NAGAHASHl, Yoshiki KUBOTA, Shinichi ...
    Session ID: 11-04-16
    Published: February 24, 2012
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Radiation therapy is one of the effective cancerous treatment method,because of less side effects to the patient. This treatment irradiates radiation beam into cancer cell. Since the body organ is moving while the therapy,cancer must be tracked quantitatively with the motion of the organ. Motion quantification maybe performed by tracking the pixels of the Ultrasound image. Ultrasound imaging is a noninvasive method to analyze the motion of the organ. However,due to the noise and less texture information,it is hard to perform tracking for a longtime. Therefore,this work proposes a robust method to track the motion of the organ by utilizing online learning mechanism.
    Download PDF (1066K)
  • Syunichi Watatani, Hiratosi Kaneda, Tomohisa Yamagata, Hirohisa Fujisa ...
    Session ID: 11-04-17
    Published: February 24, 2012
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    The balance function is an important factor for the sports player. Therefore, the sports player's balance function test was done by using the Wii balance board. The balance function measurement requested foot pressure center of gravity coordinates at open eyes and closed eyes from the right leg and the left leg for ten seconds. The balance function evaluation to this measurement value used the balance ratio and the weight transfer distance as the objective rating. In addition, the vector locus of the barycentric coordinate used it as a subjective criterion.
    Download PDF (598K)
  • Kohei ARAI, Rosa ANDRIE
    Session ID: 11-04-18
    Published: February 24, 2012
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Human Gait as the recognition object is the famous biometrics system recently. Many researchers had focused this issue to consider for a new recognition system. One of the important advantage in this recognition compare to other is it does not require observed subject’s attention and assistance. Human gait estimation for handicapped person is one important point for the psychiatrist to give a conclusion about their patients' condition. The analysis has to be done periodically whether the gait is better or worse after some therapy done. This research will help psychiatrist give the parameter measurement of the handicapped gait person and make some diagnostic evaluation. This research will use image processing technology and pattern recognition to replace some of expensive scanner devices use by psychiatrist.
    Download PDF (250K)
  • Shizuka FUJISAWA, Go HASEGAWA, Yoshiaki TANIGUCHI, Hirotaka NAKANO
    Session ID: 11-04-19
    Published: February 24, 2012
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    The demand for automatic counting of pedestrians at event sites, buildings, or streets has been increased. Existing systems for counting pedestrians in video sequences have a problem that counting accuracy degrades when many pedestrians coexist and occlusion occurs frequently. In this paper, we introduce a method of clustering optical flows extracted from pedestrians in video frames to improve the counting accuracy. The proposed method counts the number of pedestrians by using pre-learned statistics, based on the strong correlation between the number of optical flow clusters and the actual number of pedestrians. We evaluate the accuracy of the proposed method using several video sequences, focusing in particular on the effect of parameters for optical flow clustering. We find that the proposed method improves the counting accuracy by up to 25% as compared with a non-clustering method. We also report that using a clustering threshold of angles less than 1° is effective for enhancing counting accuracy.
    Download PDF (3754K)
  • Koichi MIYAGAWA, Kota AOKI, Hiroshi NAGAHASHI
    Session ID: 11-04-20
    Published: February 24, 2012
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    This paper proposes a technique of moving objects segmentation from time-series images taken by a moving camera. Firstly, we detect motion of any pixel with optical-flow technique. Secondly, Optical-flow images are segmented with reaction-diffusion systems. Finally, moving object regions are detected from segmented regions with reaction-diffusion systems. It is possible to perform pixel parallel processing by the proposed technique. So it is expected as an arithmetic processor of the energy-saving and compact vision sensor.
    Download PDF (668K)
  • Kohei ARAI, Ronny Mardiyanto
    Session ID: 11-04-21
    Published: February 24, 2012
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    The eye-based domestic helper is proposed for helping patient self-sufficient in hospital circumstance. This kind of system will benefit for those patient who cannot move around, it especially happen to stroke patient who in the daily they just lay on the bed. They could not move around due to the body malfunction. The only information that still could be retrieved from user is eyes. In this research, we develop a new system in the form of domestic robot helper controlled by eye which allows patient self-service and speaks remotely. First, we estimate user sight by placing camera mounted on user glasses. Once eye image is captured, the several image processing are used to estimate the sight. Eye image is cropped from the source for simplifying the area. We detect the centre of eye by seeking the location of pupil. The pupil and other eye component could be easily distinguished based on the color. Because pupil has darker color than others, we just apply adaptive threshold for its separation. By using simple model of eye, we could estimate the sight based on the input from pupil location. Next, the obtained sight value is used as input command to the domestic robot. User could control the moving of robot by eye. Also, user could send the voice through text to speech functionality. We use baby infant robot as our domestic robot. We control the robot movement by sending the command via serial communication (utilizing the USB to serial adapter). Three types of command consist of move forward, turn left, and turn right are used in the system for moving the robot. In the robot, we place another camera for capturing the scenery in the front of robot. Between robot and user, they are separated by distance. They are connected over TCP/IP network. The network allows user control the robot remotely. We set the robot as server and user’s computer as client. The robot streams the scenery video and receives command sending by the client. In the other place, client (user) receives video streaming from server and control the robot movement by sending command via the network. The user could control the robot remotely even in the long distance because user could see the scenery in the front of robot. We have tested the performance of our robot controlled over TCP/IP network. An experiment measuring the robot maneuverability performance from start point avoiding and passing obstacles has been done in our laboratory. By implementing our system, patient in hospital could self-service by themselves.
    Download PDF (150K)
  • Keigo TAKAZAWA, Masahiro SUZUKI, Kazutake UEHIRA, Hiroshi UNNO
    Session ID: 11-04-22
    Published: February 24, 2012
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    We construct a direct interaction system between a 3-D picture and an observer's body using prediction of the visually perceived location, and evaluate it. The system must execute the process of interaction when the body is directly at the visually perceived location of the picture. Although conventional techniques assume that the visually perceived location is the same as the location defined by binocular disparity, both locations are often different. Therefore, for example, the observer can not interact although he or she sees the body touch the picture. Also, the observer can interact although he or she does not see the body touch the picture. In our previous studies, to achieve the interaction when the observer sees the body touch the picture, we proposed that the system predicts the visually perceived location using an observer's action and executes the process when the body is directly at the predicted location. We also evaluated a new prediction technique, and demonstrated the feasibility. In this study, we implement a prediction function into the system, and evaluate it. The results indicate the usefulness of our technique.
    Download PDF (323K)
  • Yoshimasa IWASAKI, Takeshi IKENAGA
    Session ID: 11-04-23
    Published: February 24, 2012
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    A digital camera becomes a commodity as an image capturing device and is widely used in many events in daily life. However, since the current camera needs to take out from a bag or something each time, there are many situations to miss important scenes. In order to solve this problem, we proposed a wearable digital camera with HMD and its hand-gesture interface. Although it works precisely in normal situations, it still causes misdetection for complicated background scenes. This paper proposed a finger point detection algorithm based on finger area estimation using amount of hand motion. Evaluation results using a prototype system show that the proposed algorithm can detect finger point precisely even in a complicated background containing skin color.
    Download PDF (771K)
  • Hayao MORI, Jong G. PARK, Eiji NUNOHIRO
    Session ID: 11-04-24
    Published: February 24, 2012
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Tokyo University of Information Sciences receives MODIS (Moderate Resolution Imaging Spectroradiometer) data, one of the sensors equipped by NASA's Terra and Aqua satellites, and researches of the analysis on change of environment as part of the academic frontier project. In this research, we are developing satellite image data analysis system (SIDAS)1.2 using the previously developed SIDAS 1.1 as the basic system. The main feature of SIDAS1.2 is the function of the stream data processing to deliver the information which the influence of East Japan great earthquake and the reproduction situation of vegetation for the stricken area in Tohoku region. This paper presents the overview of system configuration, the stream data processing and analytical information of SIDAS 1.2.
    Download PDF (1501K)
  • Yusuke OKUBO, Keisuke NONAKA, Yoshinori HATORI
    Session ID: 11-04-25
    Published: February 24, 2012
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    A mobile camera motion identification method for personal authentication is proposed. The video which is taken during sweeping in front of user’s face is used in this method. Furthermore, it shows that two contrasting motions can be identified with a high degree of accuracy via operational feature reflected in the video. In this paper, we identified three motions which are swept horizontally, vertically and obliquely as a function enhancement of conventional method. Dictionary data for motion identification is necessary in conventional method, and then we proposed a new method which replicated oblique motion’s dictionary data by using data chosen from horizontal or vertical motion in order to reduce amount of the data. The simulation results show that the proposed method can be identified with almost same accuracy as conventional one.
    Download PDF (683K)
  • Shota KAWANABE, Tomonobu YOSHINO, Yoshinori HATORI
    Session ID: 11-04-26
    Published: February 24, 2012
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    This paper proposes a method of making high resolution interpolatior-viewpoint image by images which is taken by circle-placed cameras. If we slide show these multi-viewpoint images, an object's locus will be formularized by sine and cosine. First, we estimate all objects' coordinates by block-matching on epipolar restriction. Then, we make an interpolatior-viewpoint image on Image Based Rendering by the locus. And generally speaking, free-viewpoint image's resolution will down when making forward image. To solve this problem, we include super-resolution algorithm to define pixel value.
    Download PDF (2955K)
  • Kouitirou Eiyama, Hiratosi Kaneda, Tomohisa Yamagata, Hirohisa Fujisak ...
    Session ID: 11-04-27
    Published: February 24, 2012
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Coordinates of pivoting foot center of gravity were measured aiming at the baseball player's batting analysis. The video camera and the Wii balance board were used for the measurement. The swing time was requested from the video image. Moreover, pivoting foot center of gravity coordinates were requested from the Wii balance board The evaluation to the swing used the vector locus of pivoting foot center of gravity coordinates.
    Download PDF (312K)
  • Tatsuya TAKAHASHI, Osamu UCHIDA
    Session ID: 11-04-28
    Published: February 24, 2012
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    The smartphone users routinely browse and download confidential document files, and therefore higher security level is needed for smartphones than conventional mobile phones. For Android smartphones, there exist three authentication methods; the PIN authentication method, the password authentication method, and the pattern authentication method which takes advantage of characteristics of touch screen. In the pattern authentication method, to cancel lock users trace four or more of nine points appeared on a screen in a predetermined order, and there exist smudge attacks for it. In this study, we propose a user authentication method for smartphones having the tolerance to smudge attacks. In the proposed method, users can cancel lock by selecting the pass-image from a number of images and flicking it. The method has also the tolerance to observation attacks because the latest image shot by user is used as the pass-image, that is, the pass-image is updated constantly in the method.
    Download PDF (459K)
  • Takahiro KAKEI, Osamu UCHIDA
    Session ID: 11-04-29
    Published: February 24, 2012
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    In recent years, the quantity of video contents on the Internet has increased explosively. Therefore, it becomes difficult to retrieve favorite video contents. In this study, we propose a method to support video content retrieval using time series characteristic of facial expression. The amount of emotional expression is that how much the emotion of "disgust", "sadness" and "happiness" is appeared on the face. We use the expression change map by dividing up the shift of facial expression of feeling between impassiveness and maximum emotional face into several stages. The amount of emotional expression is estimated by the comparison between user’s face and the expression change map. The proposed method supports video content retrieval by using the similarity of user preference conducted by the time-series variation in the amount of emotional expression.
    Download PDF (600K)
  • Shin WATANABE, Shota KANAMORI, Bisser RAYTCHEV, Toru TAMAKI, Kazufumi ...
    Session ID: 11-04-30
    Published: February 24, 2012
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Our goal is to develop an effective algorithm for handling the spectral distribution in photon mapping. Today, most of photon mapping methods handle the light in RGB, a 3-component representation. The spectrum of light should be taken into account to render wavelength-dependent phenomena such as dispersion. On the other hand, a spectral rendering increases the computational cost. In this paper, we propose two methods using importance sampling in photon casting and photon tracing processes. These methods help developing an effective photon mapping algorithm taking into account spectral distributions.
    Download PDF (2450K)
  • Kohei ARAI, Yuji YAMADA
    Session ID: 11-04-31
    Published: February 24, 2012
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    The data hiding technique based on wavelet multi-resolution analysis (MRA) is already proposed. This research aims at improvement in the degree of key picture visual confirmation difficult in the circulation picture created using data hiding based on MRA. The concealment method which made the key picture the concealment method based on the steganography technology which is a kind of data hiding technology, and the multi-value picture is proposed. Moreover, it asks for a wavelet suitable for data hiding of a picture.
    Download PDF (516K)
  • Feiling HE, Hiroki TAKAHASHI
    Session ID: 11-04-32
    Published: February 24, 2012
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Recent years, book management systems with mobile phone are developed. Those systems are inputted data by manual instructions or barcode scanning. It, however, takes much time and efforts by manual input. Moreover, it is not sufficient for all books to be managed by barcodes. This paper proposes a new approach based on content-based image retrieval by using a picture taken by a mobile phone. Multiple features and two-phase matching strategy are employed to improve retrieval accuracy. The experimental results show 5% improvement compared with the traditional SURF.
    Download PDF (530K)
  • Kohei ARAI, Rosa ANDRIE
    Session ID: 11-04-33
    Published: February 24, 2012
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Human Gait as the recognition object is the famous biometrics system recently. Many researchers had focused this issue to consider for a new recognition system. One of the important advantage in this recognition compare to other is it does not require observed subject’s attention and assistance. There are many human gait datasets created within the last 10 years. Some databases that widely used are University of South Florida (USF) Gait Dataset, Chinese Academy of Sciences (CASIA) Gait Dataset, and Southampton University (SOTON) Gait Dataset. This paper classifies human gender using the energy of 2D-Discrete Wavelet Transform in CASIA Gait Database of Infra Red Camera. By using Backpropagation, the classification result is 92,9% accuracy.
    Download PDF (334K)
  • Ayako KATOH, Yui HOSHINO, Naoki KOBAYASHI, Hideki KOMAGATA, Fumikazu ...
    Session ID: 11-04-34
    Published: February 24, 2012
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    An image analysis algorithm was developed and studied for automatically segmentation of hepatic fat microvesicles. Fat microvesicles were preliminarily segmented by a self-organizing map neural network using a multiple pieces of information, such as color and morphologic features. The circularity measurement and the separation technique for combined fat microvesicles were effective to increase accuracy of segmentation. The experimental results of applying the proposed method to typical NASH HE stained images shows basic ability for quantification of steatosis areas.
    Download PDF (313K)
  • Wentao TANG, Mitsuji MATSUMOTO
    Session ID: 11-04-35
    Published: February 24, 2012
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Long in the history of learning support system, e-learning, mobile learning and ubiquitous learning system created a convenient system can be accessed by anyone easily, at anytime and anywhere. However, there are no known researches that focus on improving the learning experience by inducing the motivation of the learner. In literatures, it was found that learning activities involving the real world and teaching materials that closely related to the learner are the keys to improve the learning efficiency and motivation. In this research, we propose the GC-LULS learning support system. Instead of focusing on building a powerful learning support system, the GC-LULS focused on the practicality. By using of sensors, positioning system, augmented reality together, we can create educational games embedded with real world context, thus emerging the virtual contents with the real world learning world model. Overall, GC-LULS will increase chance in triggering the learner’s curiosity and thirst of knowledge, improving efficiency in teaching and learning.
    Download PDF (1068K)
  • Kohei ARAI
    Session ID: 11-04-36
    Published: 2012
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Eye based computer key-in system (Eye Based Human-Computer Interaction: EBHCI) with high success rate which works for a long time period by using moving keyboard and works under severe conditions such as illumination condition is proposed. Also its applications to Electric Wheel Chair: EWC control, having meal aid, communication aid, information collecting aid, robot utilized helper robot, e-learning system for improving learning effectiveness are proposed. Through experiments with normal and able persons, it is confirmed that the proposed EBHCI and EBHCI based application systems does work well in terms of functionality and performance.
    Download PDF (718K)
  • Yuichi Yamada, Kota Aoki, Hiroshi Nagahashi
    Session ID: 11-04-37
    Published: February 24, 2012
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    In this paper, we try to address the problem of "Can we recognize a human activity from a single image?". For real-life scenes, humans can communicate a concise description in the form of a sentence relatively easily. Such descriptions might identify the most interesting objects, what they are doing, and/or where this is happening. In computer vision, this problem is a highest-level operation, and it is a very difficult and largely unsolved problem. This paper presents a method for recognizing a scene based on co-occurrence of global feature and local feature. Experimental result shows that the proposed method can obtain good classification performance by using feature co-occurrence for each category.
    Download PDF (2338K)
  • Atsushi SUZUKI, Osamu UCHIDA
    Session ID: 11-04-38
    Published: February 24, 2012
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    An immense amount of digital image is distributed and stored over the Internet due to the spread of cloud services and smartphones. Under the circumstance, there is an increasing need for image retrieval using not only keyword query but also image query, that is, similarity-based image retrieval. In this study, we propose an image retrieval method considering similarity of impression. The similarity of impression between two images is computed by the eight dimensional emotional vector defined by the emotional words which represent the impression of the images. The emotional vector of an image is determined by the color information and the composition of the image. In particular, we derive emotional vector of an image by using the proportion of 180 emotional words computed by the color information of the image and the proportion of 6 emotional words derived by the location of objects on the image. We verified the validity of the proposed method by the experiments using Caltech 101 database.
    Download PDF (545K)
  • Kohei ARAI, Cahya RAHMAD
    Session ID: 11-04-39
    Published: February 24, 2012
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    In this paper, a novel method for content base image retrieval is proposed. Using The Wang`s dataset, the proposed method is evaluated and is compared to the other method. we develop a novel image retrieval method based on wavelet transformation to extract the local feature of an image, the local feature consist color feature and texture feature. Once an image taking into account, we transform it using wavelet transformation to four sub band frequency images. It consists of image with low frequency which most same with the source called approximation (LL), image containing high frequency called horizontal detail (LH), image containing high frequency called vertical detail (HL), and image containing both horizontal and vertical detail (HH). In order to enhance the texture and strong edge, we combine the vertical and horizontal detail to be other matrix. The next step is we estimate the important point called significant point by threshold the high value. After the significant points have been extracted from image, the coordinate of significant points will be used for knowing the most important information from the image and convert in into small regions. In particular, 2D discrete wavelet transform (DWT) and Haar base function is used to detect the significant points position. Base on the significant points position then extract the feature vector. The experimental results demonstrate that our method on standard dataset are encouraging and outperform the other existing methods.
    Download PDF (385K)
  • Yasunori ISHIKAWA, Kazutake UEHIRA, Kazuhisa YANAKA
    Session ID: 11-04-40
    Published: February 24, 2012
    Released: March 31, 2019
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Optical watermarking is a novel technology with which the images of real objects that are usually difficult to modify physically can contain invisible digital watermarking information, by using spatially modulated illumination that contains invisible watermarking information. We have proved that the detection accuracy near 100% is acquired with weak intensity of embedding watermark of which embedded watermarking image cannot easily be seen. In our previous study, real time processing of detection was difficult, although the improvement in the speed and automation is required when the practical forms of use are considered. In this report, we explain that fast and automated processing technique of detecting optical watermarking and the results of experiments in applying to moving picture that are embedded with time domain sequential information.
    Download PDF (296K)
feedback
Top