ABSTRACT. In this paper, a support tool for creating water flow animation from a single water flow image without requiring special skill of use is proposed. A method for creating a water flow animation by compositing water flow videos was proposed by Okabe et al. However, this method requires some degree of knowledge of image processing for specifying the direction of water flow and searching video database for similar images. If this problem is solved, it is expected that the range of user of this method may expand. As a solution to this problem, an interactive support tool for creating a water flow video based on Okabe's method is proposed and its usefulness is confirmed with some examples.

ABSTRACT. Areas on a landscape image were segmented using a method for semantic segmentation, and the depth of the road landscape was computed using the change in position and size of vehicles and pedestrians. To generate two images for our right and left eyes, the segmented regions of persons, the road, and buildings were translated left or right according to their depth. Our pseudo stereoscopic was realized using an anaglyph image made from the two images for our right and left eyes. Comparison of the depth accuracy with our proposed method and a light-field camera is discussed.

ABSTRACT. In lectures, the teacher sometimes needs to grasp the holistic state of all students not only each student. Therefore, the various methods to estimate the behavior of students by using videos have been proposed. However, taking high resolution videos of classrooms cause a violation of privacy. So, we examine the possibility of the analysis of the behavior of students in low resolution videos that are difficult to identify individually. At first, we prepare some smoothed videos that are low resolution videos. Next, we create a Haar-like classifiers for each smoothed video. Then, we examine the number of skin color pixels of the face area detected by this classifier. From the experiments, it was possible to correctly determine when students were facing the front from the number of skin color pixels in the smoothed videos.

ABSTRACT. Cardiovascular Disease (CVD) is considered to be the common cause of death in several counties while the necessity of experienced cardiologists is at its peak. The fate of the patients depends sorely on how well-equipped the personnel and the hospitals are to overcome the clinical issues. Still, it can take a substantial amount of time for the practitioners to perfect their skills, even more so for rookies who just entered this daunting field. Hence, an educational oriented tool will undoubtedly assist the newcomers in this critical profession. This paper aims to utilize a recent simulation technology namely "Position Based Dynamic" or PBD method to visualize the mechanism and phenomenon of the muscle and blood inside the human heart, including the heart muscle movement, the blood current and the interaction between them. Then an evaluation interview was conducted with a medical professor to review the simulated animation. As a result, the system had proven the concept of using PBD to create an educational oriented system for medical undergraduates has been proven plausible.

ABSTRACT. We discuss a small dot sequence called a ``dot pattern’’ on a 3D object generated by a 3D printer; 3D drawing is performed using a digital pen. To avoid projective distortion, each polygon of the 3D computer graphics data is converted into a rectangle. Although the inside of the polygon is flat, multiple polygons are connected in a complex that is not flat. A polygon including a pen tip is detected, and the position of the pen tip is computed by our method. We designed our dot pattern and an algorithm to compute it. We simulated our method and showed the feasibility of handwriting.

ABSTRACT. The color of objects captured in the sea is different from that of the observation on the ground due to the influences of scattering and absorption. We develop a method for restoring the color of underwater images considering the influences of scattering and absorption of light in water based on a physical approach. First, scattered light (in-scattering) is removed, and the color is restored based on the attenuation of light due to out-scattering and absorption. We examined the proposed method using synthesized underwater images and applied the method to the restoration of an actual image captured at the bottom of the sea.

ABSTRACT. Political participation in elections is interesting to study in the era of web 2.0. Various conversations and interactions can be recorded for analysis using one method in Artificial Intelligence. The research data is sourced from a campaign group in an Online Social Network (OSN). This campaign group actively interacts from before the official campaign period to the determination of the winner. The interactions in the group are collected to get patterns that occur in daily activities. Interactions in the OSN in the form of likes, comments and shares will be compared to existing post activities. This research proposes the pre-processing of data using the Sum, Mean and Geometric Mean. Furthermore, data clustering is done using the Self Organizing Map (SOM) method to find out the grouping of activities that occur. The goal is to get interesting visualizations in grouping data. The results obtained from the application of the Sum, Mean and Geometric Mean in the pre-processing followed by the Self Organizing Map can display good cluster visualizations.

ABSTRACT. To minimize damage during disasters, the rapid collection and delivery of accurate information are essential. Therefore, the potentiality of social media, especially Twitter, in the case of a disaster has attracted worldwide attention. In this study, we conduct a cluster analysis to classify Unicode emojis used in the tweets posted during two disasters, the 2018 Osaka North Earthquake and the 2018 Typhoon Jebi. Moreover, we attempt to visualize the content of tweets containing emojis by creating the co-occurrence network diagrams.

ABSTRACT. Recently, we have used many videos by various image display devices. In particular, since an opportunity of viewing contents in a large size display or a mobile device has been increased, it is one of the important problems to prevent visually induced motion sickness (VIMS). Then, a determination method by using motion histograms had been proposed to solve the problem. However, there are some problems of the reduction of determination accuracy due to slow speed screen shake and the similarity of movement of gaze objects and screen shake information. In this study, this paper proposes the two following methods; the first method uses a quantized histogram and a division of direction, and the second one uses divided screen into background area and center area. The author considers the improvement of accuracy of the determination method with motion histograms by the simulation experiments, and indicates the effect of the proposed method (PM). By the results, the first method is improved at most 27.90% in accuracy, and the second method is improved at most 15.18% in accuracy, comparing with the conventional method.

ABSTRACT. This paper describes the two major expansion possibilities of future digital signage systems in regional area in addition to existing advertisement services. One is the guidance services for emergency or disaster situation such as earthquake, tsunami, typhoon, flood, etc., while another is the regional cultural information services for sightseeing or general education. Basic investigation of system architecture and the content management has been discussed. For emergency or disaster situation, organizational policy management of government is essential, while the content evaluation and the interaction feedback is important for regional cultural information services. For both cases, professional engineers of digital signage content management are necessary, then the regional human resource development will be strongly needed.

ABSTRACT. Communication among farmers and consumers is considered important in building active sustainable farm management and revitalizing agriculture in Japan. However, what types of communication is effective for farmers and youth people who carry the future agriculture still needs investigation. This study aims to introduce how the use of images on Social Networking Service (SNS) may create new sustained experienced value. Images on the existing website that use an agriculture platform are classified into several categories, and survey of impression effect will be conducted to target audience. The study will clarify the effectiveness of sharing images on the SNS to make sustained opportunity and interest of the youth in the field of agriculture.

ABSTRACT. In recent years, an unmanned construction system has been used at construction sites to ensure the safety of workers. However, there is a problem that the digging position shifts because the sight line is different from normal operation. In this paper, we propose a display system that can change the position and the direction of the camera by using the see-through HMD. In the experiment, a construction machine operator and others performed remote control. Moreover, the work efficiency and the work accuracy of remote control were measured. As a result of the experiment, the work efficiency and the work accuracy have been improved than the conventional remote controlling display system.

ABSTRACT. In the field of materials engineering, secondary particle area was common as a quantitative evaluation method used for particle morphology control. The microscopic aggregates are photographed with an electron microscope to perform multifractal analysis. The purpose of this research is to evaluate the morphology, dispersibility and correlation of aggregate structure by analysis. In the experiment, the conditions were set along the particles that make up the aggregates. From the experimental results, we considered the aggregate state obtained by multifractal analysis. In addition, comparing the results of consideration with the actual analysis image, it has been confirmed that the structure evaluation of aggregates by the proposed method is possible.

ABSTRACT. Remote control construction work machines are used at the place where workers are exposed to dangerous slopes or contaminants,etc. Therefore, various studies are being conducted to make the task of remote control more efficient and reduce the burden on workers. In the conventional evaluation, feedback by a questionnaire is general in the evaluation of burden reduction on workers. In this study, eyestrain was measured using an Auto Ref/Keratometer that can monitor the activity of the ciliary muscle. Then, the quantitative evaluation was performed based on eyestrain data obtained by experiments of three display systems. As a result, it was possible to confirm the system that is least tired from the measured eye strain value. Therefore, it has been confirmed that the proposed eyestrain evaluation method can be used for worker burden evaluation.

ABSTRACT. Recently, research for increasing the thermal conductivity by adding a high thermal conductivity material to a low thermal conductivity material and controlling the dispersion and aggregation state of additives is noticed. Although many studies have been done to evaluate the aggregation state from the shape of particles or particle groups, but a method for quantifying the state of particle groups by adding conditions other than the shape has not been proposed. Therefore, we apply multifractal analysis to particle groups and propose a method to quantify the distribution of particle groups from the value of information dimension. We successfully did quantitative evaluation of the distribution of the particle group generated by SN/SUS316L composites.

ABSTRACT. In this work, we explored the possibility of video analysis for animal husbandry through the approach of video summarization. Several raw videos with specific length of time about the daily behaviors of cattle in cattle barn are applied as the experimental object. In the approach, deep leaning, statistics and temporal segmentation are used to extract key-frames and make an optimal summary from the frames. And it can be confirmed that this approach achieves the purpose of video summarization based on cattle's movement through analyzing the results of the experiment.

ABSTRACT. This research is to develop a support system for producing resin works. Resin work is one of handmade works and has various production methods. It is difficult for beginners to produce a resin work with desired design. In order to solve this problem, we simulate resin work with computer graphics. As visual simulation by transparent objects, volume rendering by ray casting method is used. Ray tracing method is combined to express reflection and refraction of light.

ABSTRACT. In this study, we have conducted a cold pressor test to investigate the relationship between autonomic nervous activity and pupillary fluctuation. As a result, it was found that the LF / HF component of the pupillary fluctuation increases at sympathetic dominance. In addition, it was found that the slope of the power spectrum of pupil fluctuation tends to decrease when sympathetic is dominant. From the above, it was shown that autonomic nerve activity can be evaluated by analyzing pupillary fluctuation.

ABSTRACT. Light source separation, which separates an original image lit by multiple lights into decomposed images lit by each light source, is an ill-posed and challenging problem. Especially, source separation from a single image without manual inputs is a more challenging task. In this paper, we tackle this novel task. We present a two-step approach: 1) to generate candidates of the decomposed images through the network conditioned by arbitrary light position, and 2) to select the decomposed images from the candidates on physical constraints. The evaluation using synthetic images demonstrates the effectiveness of our method.

ABSTRACT. General radiography is an imaging examination, performed in many facilities. In this imaging, clearly grasping fractures and lesions are required. Before radiation technologist refer to target ranges or positioning during imaging from the past general radiography images. The aim of this paper is to clarify problems of current X-ray image reference by performing questionnaire investigations to 100 medical radiation technologist. We gained the following answers why they didn’t refer images: Being busy. Taking too much time. The images are not automated. Based on the answers, we develop a patient specific image reference supporting system which makes imaging more quickly and automatically without increasing the procedures. This system is expected to be useful for the image information-sharing among multiple facilities, and classification tables of image reference for each patient are created, which are based on Ministry of Health, Labour and Welfare Standard JJ1017 for frequently used codes of radioactive rays. In order to operate this system in a clinical manner, we create simply patient information / order entry systems of electronic medical chart, radiology information system, and a picture archiving and communication system functions, and asses average time of image references, positioning during imaging /setting imaging conditions, and the usability by two medical radiology technicians, and clarify the usefulness.

ABSTRACT. For standardization of guidelines to prevent visual induced motion sickness (VIMS). It is required to clarify the relationship between the visual environments and VIMS. In this paper,we focused on relationships between VIMS and ambient illuminance using subjective VIMS scores and body movements as objective index experimentally. As a result subjective scores shows that the recovery from VIMS is quicker in bright condition than in dark one. From the body movement analysis subjects who complains VIMS tends to have large body movement in bright condition.

ABSTRACT. Early detection of pancreatic cancer is very difficult, and prognosis of the cancer is poor. In addition, as the morbidity are increasing in Japan, early detection and establishment of the treatment are urgently needed. However, the carcinogenic mechanism of pancreatic cancer has not been completely elucidated. Detailed 3-dimensional anatomical models may help to elucidate the carcinogenic mechanism. We therefore examined 3-dimensional anatomical reconstruction from microscopic images of pancreas serial section of mouse with pancreatic cancer. The size of the microscopic image is huge and manual segmentation is not realistic. Therefore, we automated the segmentation by using a support vector machine and texture features of co-occurrence and run-length matrices. We also performed step-wise variable selection for higher precision, and performed 3D Gaussian smoothing for noise removal.

ABSTRACT. Real paper gets deformed anisotropically due to a uni-directional placement of the inner fibers. Few existing CG researchers allow for such fibers to represent virtual papers, and thus anisotropic paper deformation has not been well represented. In this study, we present a two-dimensional visual simulation model for anisotropic papers, which abstracts mutual mechanical relationships among intersecting fibers by a network of filler and hinge springs, and incorporates connection points for keeping the shape of each bended fiber. By releasing the network connections when the filler and hinge spring extend above a certain limit, the paper provides a plausible tear. We succeeded in generating a different appearance of the torn-off line of papers according as the pulling direction.

ABSTRACT. Deterioration serves as an imperative factor of visual simulation when expressing the fine temporal change of clothes as well. Considering the complexity of their microscopic structures, a simple and easy way to damage them is required. This paper, therefore, presents a novel method for modeling clothes with fiber–level details and expressing their damages on a ray-casting basis. Our particular focus was placed on damaged jeans, where representative kinds of local deterioration, including fading and cutting of fibers, affect the entire appearance of the clothes. The expressiveness of the proposed method was empirically proven with several modeling and rendering experiments.

ABSTRACT. Fire is very familiar and important to us so that a lot of studies have been performed. Some of them are the methods to control the fire shape for animations, and others are techniques to visualize realistic fire in real time. In fact, the flame color depends on the combustion state such as complete combustion, incomplete combustion, and non-combustion; however, the most previous researches consider only incomplete combustion state. Even the study considering the color of complete combustion does not treat the combustion state that depends on physical property of the fuel gas. Therefore, this paper proposes a method to simulate and to visualize a candle flame by estimating the physical property of the fuel gas and discriminating the combustion states. The result of the simulation has shown that the color of a candle flame changes according to the combustion states.

ABSTRACT. Generating aesthetic curves and surfaces is an important part of industrial and graphical designs. Aesthetic quality of curves and surfaces depends on its curvature profiles. Therefore, freely manipulation of curvature profile is in demand. The TAP Curve: Tangential Angle Parameterization Curve is one of the solutions for generating aesthetic curves. This method provides freely manipulation of curve-shape and curvature profiles simultaneously. One of our proposals is a method to extend TAP Curves to space curves. This method inherits enough features to generate aesthetic curves from the TAP Curve. Furthermore, we propose several methods to generate surfaces with TAP Curves.

ABSTRACT. We always live under 1G on earth. However, gravity suddenly is lost when we fall from a high place and stumble on walk. Animals make defensive reactions with loss of gravity as a life crisis. It is important knowledge to learn the defense response by the decrease of gravity in the aging society. That knowledge will help in prevention of fall down for the elderly. We developed a simple Low gravity generator and observed the defense response of rodents under rapid low gravity.

ABSTRACT. This paper shows the results of applying machine learning methods to data collected from the pressure exerted to a gamepad’s button with the purpose of finding patterns that can help predict the player’s: fun, frustration, boredom, valence, arousal and dominance at a determined time. We gathered data from players while playing a 2D shooting game and used it to classify player’s perception related parameters using Neural Networks and Support Vector Machines. As a result: boredom showed the highest accuracy of prediction (83.20%), frustration, fun, difficulty and arousal were predicted with more than 75% of accuracy.

ABSTRACT. This paper describes the way of making a priority list in a city’s alliance partners for early restoration from natural disasters. Since Japanese government encouraged cooperation of cities for restoring from the disasters, the total number of the alliances has been increased and each city requires to know the most helpful alliance partners based on the actual damages. In order to support that, a prototype of prioritizing these cities has been developed, using the existing data and the satellite images. It can be used not only for the future disasters but also for mutual preparations or drills by municipal governments.

ABSTRACT. Recently, many image recognition technologies are applied to detect abnormalities such as deterioration of infrastructure and lesions of human bodies. In practical, because missing the abnormality is quite risky, recall rate is much more important than false detection rate. In this paper, we propose novel training method of CNN to minimize false detection as much as possible while keeping high recall rate by maximizing a specific narrow area under the ROC curve according to the target recall. We show the effectiveness of our method in the experiments of semantic segmentation task.

ABSTRACT. In this study, we propose a supply chain management method that accurately tracks the logistics in the supply chain process without compromising data integrity. Supply chain is typically used a RFID (Radio Frequency IDentifier) or barcode for tracing objects. However, this method is not protected against malicious activities like someone peeling off an RFID tag and replacing it with a fake. These types of security problems can be solved using PUF (Physical Unclonable Function) and blockchain. Our proposal is to tamper resistant as it tracks and logs the actions of all users using advanced technologies.

ABSTRACT. Histopathology determines treatment of cancer and pre-cancer through classifying into a diagnosis the patterns of microscopic organization of cells in tissue sections from biopsy or surgical specimens. This can be done using images derived from chemical staining techniques using hematoxylin and eosin (H&E) so that we can determines the three part of the cell, which are nucleus, cytoplasm, and fibre. Visual counting of nucleus tissue takes time, therefore, reducing the time consuming sample processing by capturing its intrinsic spectral information has proved promising. In this project, we focus on introduce a method to identify the nucleus in pancreas cell image by using template-matching and axislength-based. In defining the accuracy of the process, the nucleus in the original image are manually counted and treated as a benchmark. The number identified by the process is then compared to the benchmark. The conclusion shows that the accuracy of the arranged template-matching and axislength are 37% and 54%, respectively, yet to be called established; therefore, further effort should be undertaken. Nevertheless, the template matching and axislength-based process show a possibility to be used in automated nucleus identification.

ABSTRACT. To improve exercise motion such as sprinting motion, it is necessary to identify which abilities lacks from the exercise motion. This paper focuses on a method to propose an improvement strategy that can supplement the lacking abilities. First, we evaluate the sprinting motion from human joints movement to find the lacking abilities based on evaluation items. From the evaluation results, we predict whether a learner can perform some training related to sprinting motion effective or not. And, we find an important item to supplement the lacking abilities with analyzing the prediction model. Experiments show promising results for achieving our proposed system that identifies lacking abilities from sprinting motion and find a training suitable to improve the abilities.

ABSTRACT. Although the validity of various items (e.g., arm swing) to evaluate sprinting motions were clarified in recent studies, items such as “the interlock between upper and lower limbs”, which are said to be important factors for good sprinting, are not included in the recent studies. In this paper, we clarify the criteria for what kind of motion is the one that upper and lower limbs interlock enough, using an automatic method that secures fairness and objectivity. That enables to augment the methods to improve motions. We extract the features of our item by using the sprinter's joint coordinates and clarify the relationship between the features and the items of Suzuki et al. and Kaji et al. using machine learning. The results of experiments show that our item is highly correlated with some of the recent studies’ items. Thus, it is expected that the criteria of our item get clear.

ABSTRACT. This paper proposes e-learning applications specially designed for classical Japanese literature classes in the Information Science and Engineering Faculties. These applications are developed on the HTML-5 or Android platform, and effectively use image content to help get unmotivated students interested in literature classes. The content includes illustrated handscrolls and paintings on a folding screen. By using these content data, a groupware allowing students to put comments on the handscroll images and a jigsaw puzzle game using classical literature images have been developed. The key technical points behind these applications are event-driven programming, image handling, and coordinate matching. Students who have actually used these applications say they have been helpful in learning literature and that they have enhanced their interest in programming techniques as well.

ABSTRACT. Expressing the differences in features of each person is important to create cartoons when we use them at the same time. We propose a system which visualizes feature values of a set of cartoons. We also discuss cartoon parameters that make it easier to express differences in features.

ABSTRACT. We present an efficient method of simulating hair interaction with turbulent flow. Hair interaction is one of the critical problems for hair animation, and many methods have been proposed to solve this problem in computer graphics field. We focus on the hair-air interaction including turbulent flow because it drastically affects hair streaming. In this paper, we integrate a wavelet analysis with vorticity confinement to control the turbulent flow. It realizes that users can easily control the hair streaming with only one parameter. Also, our method solves both hair-hair and hair-air interaction with a unified particle-based method that enables fast, natural two-way interaction.

ABSTRACT. In Manga, Animation and Video-Game Contents, there is a luminous expression that covers a character's body. It expresses a strength of the character and is used as a kind of visual effect of the scene. In recent year, in Video-Game the effects are more expressive, but it almost needs to prepare for to make textures and to set timing of appearance and disappearance of texture. Our goal is to render these effects in real time.

ABSTRACT. Most of commercial Anime films in Japan made with 3DCGI are rendered in non-photorealistic look rather than in photorealistic look. Then these rendered elements will be merged with hand-drawn characters or environments in composite software such as Adobe After Effects or Foundry Nuke. However, this traditional pipeline is not always efficient, especially in complicated shots because it is quite difficult for directors to judge whether we construct scene with 3DCGI software or with composite software. This often causes many troubles at composite pipeline. We propose a new, efficient scene construction and composite system for Anime production, which intergrades 3DCGI and composite elements, utilizing Realtime-Graphics solutions of Game Engines. We have developed a scene construction system plug-in software for UnrealEngine4, which we can construct and composite Anime shots at the same time. We verified through our examination at some commercial projects and found that we can improve scene construction and composite pipeline with our method.

ABSTRACT. For realistic and comfortable museum experiences, a virtual museum which can be viewed with a head mounted display is developed. Based on an original screen-based virtual museum, virtual reality functions related to a head mounted display are implemented. Unity Game Engine and Oculus Systems are used for the implementation. Four aspects of "Freedom of operation", "Immersion", "Comfort of play", and "Picture quality" are examined to clarify which type of virtual museum would be suitable for any specific user's needs. As a result, a good evaluation was obtained for "Immersion"

ABSTRACT. Drawing is an art that children can enjoy freely, and in recent years a number of digital content combining drawing and CG technology has been developed and has gained great popularity at events. In this paper, we developed digital contents to create 3DCG models of trains and cars from their drawings and run them in 3DCG dioramas. The CG diorama and vehicles running in it can be observed in three dimensions by the principle of trick art. I exhibited the developed contents at events, and many children enjoyed creating 3DCG of vehicles by drawing them running them in CG dioramas.

ABSTRACT. Reduction of manual work of annotation is an essential part of sign language recognition research. This paper describes one of weakly-supervised learning for continuous sign language recognition. The proposed method consists of forced alignment based on DTW and Isolated word HMM adjustment using embedded training. While the proposed forced alignment only requires one manual annotation for each isolated sign language word, it can generate sufficient quality of the annotation to initialize isolated word HMMs. Embedded training adjusts initial HMMs to recognize continuous sign language words only using ordered word labels. The evaluation shows the proposed methods achieves significant improvements compared to the previous weakly-supervised learning.

ABSTRACT. Although most of hand pose estimation methods are good enough for estimating joints on hand, there are some still challenges on noisy poses. Since these missing or duplicated poses are neglected to handle or assumed as complete poses, it leads to effective on 3D pose estimation, action recognition and other skeleton-based systems. In this paper, we propose a method that consists of three parts of hand pose estimation. First, a hand area is detected by You Only Look Once (YOLO) hand detector model. Second, the previous hand area is inputted to the OpenPose to detect hand joints. Third, since some hand joints obtaining from the second step are incomplete or noisy, we use a model to predict these joints using Variational Autoencoder (VAE). Some improving accuracy of hand joint detection and generating noisy joints are shown in our experimental results.

ABSTRACT. Information about bus passenger flows is necessary to optimize bus routes and schedules.In this paper, we propose a fully automatic method for accurate estimation of bus passenger flows by analyzing videos recorded by cameras installed in buses. Our method has two parts: (1) collecting images of passengers boarding and disembarking the buses by using methods for human pose estimation based on CNN and human tracking with kernel correlation filter (KCF) and (2) estimating identical passengers' images by matching the images taken at the entrance and the exit of each bus.The results of experiments show that our method improves the accuracy of matching the same person's images by about 16% compared with a conventional method.

ABSTRACT. In this paper, we propose an image retrieval method for living rooms. Images are retrieved by kansei words that are expressed impressions by adjectives such as "fashionable", "warm" for intuitive search. The proposed method correlates kansei words and image features in a kansei search space by using canonical correlation analysis. Then, generate a kansei model that maps variables on the kansei search space using regression analysis. We employ several features and evaluate quantitatively retrieval accuracy. The proposed method retrieves 59.73% similar impression images among the top five retrieval results from 277 images.

ABSTRACT. For attractive video presentations for food such as in cooking programs, museum exhibitions, or food samples in restaurants, not only visual but also audio presentations are needed. Especially with visual devices used from relatively near viewing positions like table top displays, stereo speakers placed outside a display does not work properly, that is, sounds are heard from left or right sides, not from displayed objects. To solve this problem, an audio projection method by a parametric speaker onto the glass of display is proposed. The sound beams from the parametric speaker are hyper-directional and reflections from the glass reach viewer's ears directly. Thus, the sounds are felt to be heard from displayed food objects on the table top display. The proposed method was used at the Korea-Japan food culture exhibition at first few years ago and continuously used for audio-visual food contents creations.

ABSTRACT. Web-based signage, that is digital signage using web technologies, has potential as a next generation signage architecture. However, some doubt that web-based signage is practical, especially when comparing the quality of digital signage terminal service on a web browser with that of proprietary digital signage. The quality requirements for a commercial use digital signage system are high and they have to be applied to web-based signage as well. In this study, we investigated how web-based signage satisfies the requirements through an implementation focused on content presentation, and we verified the effectiveness of our implementation by applying it to demonstrations. In addition, we describe the expectations for standardization of web-based signage implementation.

ABSTRACT. The digital signage services, widely installed in the town, are effective to notify the information. The interactive type of the digital signage such as a KIOSK terminal is also used for showing the detailed information interactively, but it might cause some privacy issues for users who do not want some others to watch their behaviors. In this regard, we consider a method for the personal service by pairing with users’ personal devices such as smartphones. Since avoiding users burdened for the service, the method should not require users to install additional application on their personal devices. In this paper, we propose a method for the personal signage service and its implementation using Wi-Fi redirection and web technologies, and we report verification results on its effectiveness from the demonstrations and the trial service.

ABSTRACT. In Japanese grammar, it is generally said that spatio-temporal deixis ``KORE (this)’’ and ``SORE (that)’’ refer to something close to the speaker’s and the hearer’s space, respectively. In this study, we analyze video data of the preparatory work for a traditional festival which have been recorded for 7 years and reveal the usages of the deixises in cooperative activities within intricate material environments are also sensitive to the following 3 additional factors. 1. The postural configuration of the speaker and the hearer (such as vis-à-vis, side-by-side etc.) 2. The activities in which the speaker and the hearer are engaging (same or different activity) 3. The co-occurring verbs (referring to speaker’s or hearer’s action)

ABSTRACT. In this paper, we report an effective design system for Japanese character “Kanji”. We set the type, thickness, curvature and color of the lines that make up the character as parameters. By using the Interactive Genetic Algorithm (iGA) and changing the parameter selection probability according to the user's preference, 90% of "Kanji" can be judged as a favorite before the tenth generation. Thorough these process, it turned out that the user cares about the different parameters of the design. It was assumed that the number of generations required to obtain the optimal solution would increase enormously because the number of options increased to the power as the number of parameters increased, but it was suggested that this method could be reduced to a multiple. When designing “Kanji” by the proposed method, the load required can be estimated by increasing the number of parameters.

ABSTRACT. Conventional registration methods like Iterative Closest Points (ICP), due to their local minima convergence cannot successfully register point clouds that are low overlapping (<40%). Moreover, state-of-the-art methods rely on difficult to tune descriptors to make point correspondences and are not able to register low overlapping point clouds. In this work, we present a novel iterative method which does not entirely rely on point descriptors and leverages the local minima convergence of the registration between supervoxels, which leads to the correct alignment of the full point clouds with an overlapping ratio as low as 19.88%.