ABSTRACT. The Ministry of Health, Labor and Welfare (MHLW) indicates that Japanese people should increase social activities that involve physical activity and mental health in order to continue living life without requiring care. We know many　 methods of exercise for health, but it is difficult to adjust exercise load. We tend to make exercise too strong and we give up the continuation of it. Therefore, we focused on worshiping temples and shrines that have a positive impact on mental health. people can increase physical activity due to walking in the precincts of temples and shrines. It is a low intensity exercise. Further, they walk through many trees while seeing historic buildings that have good mental impacts. In addition, it is considered of one of the regional activities, because it can be related to the local residents such as the priests of the temples and shrines. In this study, we measured activity amount to show that temple-worshiping is one sufficient exercise. It has shown that it is exercise enough for health maintenance. The activity of receiving "GOSHUIN" is also social activities. It leads people to increase the amount of physical activity, feel enjoyable, and being related to local residents. It is important knowledge that health is important not only for the body but also for the mind. In this study, we understood that worshiping and getting "GOSHUIN" have advantages over other exercise methods.

ABSTRACT. Muscle strength is widely measured in various scenes and various ways; for example, the human, the grip and back strength dynamometers are often used in medical institutions and sports. It is crucial to measure not only static muscle strength but ballistic muscle strength such performed in jumps, fall avoidance movement, vertical jumps, etc. Measuring ballistic muscle strength can contribute to evaluate not only subject’s muscle strength but whole-body exercise capacity and physical functions. In this study, we developed a device and analysis method to measure dynamic muscle movement such as ballistic muscle strength on rodents’ by using electric shock and 3D digital image analysis.

ABSTRACT. We focused on the relationship between beta motion in peripheral vision and gestalt collapse. The experiments were performed using a point sequence and a rod-like beta motion. As a result, it was found that there is a difference speedup and shape between point sequence and rod-like beta motion.

ABSTRACT. One of the fixational eye movement, the microsaccade is an involuntary movement and a fast saccade. It is known that its amplitude and frequency change in case of gazing. In this paper, we experimented how micro saccades change according to the size and color of the texture image (checkered pattern).

ABSTRACT. In this paper, under the research of visual phenomena in edge blurring, we propose a method to insert edge blur suppression line as a method to reduce the contrast deterioration caused by edge blur, and quantitatively examine the effect of suppression line on the contrast reduction.

ABSTRACT. We are working on a project that classifies the degree of error proneness based on how a person draws a line when a point cloud is shown on a tablet device. To identify what kind of point clouds are meaningful for classifying the degree of error proneness, we have computed 9 kinds of graph-theoretic measures for point clouds and found that the monotonicity measure (the squared Spearman’s correlation coefficient) is meaningful for classifying the degree of error proneness.

ABSTRACT. A thermal image taken by a thermal camera and a RGB color picture were arranged in a dataset pair; the datasets were learned using a deep learning algorithm called pix2pix. After the sufficient training in the machine learning, it was possible to generate a thermal image from a color image taken with a digital camera. A thermal camera is required when we create teaching materials, however the thermal camera is not necessary for normal shooting using our method.

ABSTRACT. A three-dimensional computer graphics models was reconstructed based on a method called Structure from Motion using historical old videos taken by an in-vehicle camera after World War II in Japan.

ABSTRACT. In this study, we examined a method of imaging all the teeth in the oral cavity simultaneously by using a 360° camera. By attaching a light-emitting diode tape to the RICOH THETA S, which is not equipped with a lighting device, it was possible to shoot all teeth in almost uniform brightness.

ABSTRACT. This paper aims to improve the visibility of 3DCG (three-dimensional computer graphics) video games as seen from a third-person point of view. In this paper, we propose and evaluate techniques that make transparent obstacles when the obstacles block the view of a camera. As a result of the subjective evaluation experiment, it was found that the proposed method of making semi-transparent obstacles can improve the visibility of the video game and thus was evaluated to be suitable for inclusion in gaming content.

ABSTRACT. Collaboration managers of medical companies have been seeking scientists from image processing domains as well as medical doctors who are well-informed in both life sciences and informatics in order to conduct research and development. As a preliminary step for joint research agreements, they have been examining the researchers' achievements in a database of the literature, as well as their personality and competency at academic conferences. Using a questionnaire survey, this paper showed that the requisite character types for research collaboration could be classified into the following: (a) Cooperative Type, (b) Leader Type, and (c) Balance Type.

ABSTRACT. We investigated the viewpoint dependency of perceived gender differences by evaluation tests using artificially generated 3Df aces by the impression transformation vector method. For a set of faces that included two classes (male and female), we acquired color images and 3D coordinates with a RGB-D sensor (Kinect V2) to build a morphable 3D face model. We obtained impression transformation vectors to synthesize gender differences and applied then to novel faces. We obtained impression transformed face images with switched viewpoints by rotating the 3D face model. The gender difference ratings were made by a paired comparison method that randomly displayed two images to select the one that gives a more masculine (or feminine) impression.

ABSTRACT. Our experiment recognized finger numeric characters 0-9 through depth images obtained by Kinect V2 RGB-depth sensors. We applied principal component analysis to a learning sample set of the depth image data of the normalized fingerspelling, identified the fingerspelling expressed by test samples using both a class-featuring information compression (CLAFIC) method and convolutional neural network (CNN), and evaluated their identification performances.

ABSTRACT. The temporal segmentation of a time-series sequence of facial expressions into utterance units remains a critically unsolved problem in facial expression analysis. Facial expressions must be analyzed based on 3D motion data because utterance generation is made by complex deformation of the face’s 3D shape. In this study, we examined the possibility of segmenting facial expressions into utterance units through recognition experiments using the 3D position data of facial feature points recorded by a motion capture system.

ABSTRACT. When holding the events such as a lecture, organizer should make many stage plans. However, when we apply venue for preparation activities, the time is limited to use. In addition, when we simulate stage direction, we often waste the time of using venue because of making mistakes. The purpose of this research is to develop a stage event support system using virtual reality techniques. By using this system, organizer can simulate stage direction of the venue and schedule.

ABSTRACT. This paper proposes a rendering technique that makes semi-transparent silhouettes from three-dimensional obstacles when the obstacles block the view of a main character. This technique improves the visibility of a main character in 3DCG video games.

ABSTRACT. Underwater world is filled with blue light, producing color gradient from blue to green in the scenes. These colors can be reproduced by simulating multiple scattering in the water. Although this can be performed stochastic rendering methods such as path-tracing, they are computationally too expensive to apply to interactive applications. This paper reports a fast rendering method that utilizes the plane-parallel scattering theory.

ABSTRACT. Monitoring of land cover (LC) and land use (LU) are essential tools for countries to achieve sustainable food production. However, LC/LU visualization and mapping is generally labor intensive and requires high budgets and preparation time. In this study, we investigate the potential of deep neural networks in classification and visualizing of LC/LU maps. The study takes advantage of open access Sentinel-2 satellite data and Google Earth Engine (GEE) processing power for image acquisition and preprocessing.

ABSTRACT. A system for fluorescence microscopy consisting of a CCD Camera along with 24 LEDs ring sources is being developed. The System contains three different colored LEDs as illuminators and CCD camera as a detector to magnify and capture image from microscope slide speciment. Multiple images are taken using this system with 4 illumination conditions i.e. red, green, blue and white illumination sources. Central wavelength of the developed system are 466nm, 480nm, 522nm and 621nm.

ABSTRACT. In this research, we study on a method to make a user feel like jumping higher than actual by visual image manipulation using VR technology. In the proposed method, a method of exaggerating the actual height of jumping motion in VR space and a method of increasing the relative height of jumping motion by reducing its own scale in VR space are combined. In this way, we aim to realize jumping height that is impossible in the real world without feeling discomfort in VR space.

ABSTRACT. The purpose of this rserach is to develop a method to analyze long-term video recorded by a camera attached to the halter of a calf, and to analyze and visualize the behavior characteristics of the calf. Observing, recording and analyzing the behavior of cattle is important for studying the growth and behavior of cattle. The most classical method of recording is the direct observation and continuous recording by the observer. This method is essential to investigate basic calf behavior in detail, but the method takes a lot of time, is labor intensive, and is sometimes technically difficult. video is commonly used as a more effective cow behavior record. However, image analysis techniques for behavioral analysis have not been fully developed and have been analyzed manually by human observers. In this research, we extract the movement of the mouth in the video from the camera video, and visualize the characteristic of "rumination" which is the characteristic behavior of the calf. The principle of slit camera is used as a visualization method. As a result, long-time features are visualized as a single still image, and feature extraction and enhancement become possible using existing image processing techniques.

ABSTRACT. According to the spread of mobile camera devices, imaging applications have been enriched. In document shooting, a highlight or a shadow used to be in the target images. Binarization method are used in most of imaging application for visibility improvement. However, it is not applicable to color documents. Contrast and gamma correction can control the overall visibility of the image, but it is hard to remove local noises. It is difficult to remove only shadows and highlights from images without altering the target information such as the color design or color tone. In this paper, we will propose the technique to achieve improved visibility without losing target information by integrating multiple captured images.

ABSTRACT. Japan has a lots of greenery, and 2/3 is a forest area. It is a very high percentage of developed countries. But, it makes difficult about the forest monitoring that is decline of forestry and the aging of forestry workers. We have many earthquakes, which cause many landslides in the forest. And there are problems such as illegal dumping of garbage and disease of trees. Japan has maintained beautiful forests under human management. In this study, we developed a method to monitor the mountain forest using UAV. Multicopter is popular for aerial photography. The forest monitoring area is wide and the multicopter has short flight time, so it is not suitable for forest monitoring. We developed forest observation with coordination of fixed wing and multicopter.

ABSTRACT. The present study proposes a new method to diminish the sensation of mismatch in the 3D image when the binaurally recorded sound is displayed without tracking the head movement. The experimental evaluation showed that the additional sound object improved the perceived sound quality and the degree that the sound was three dimensional. The result indicates that the proposed method is capable to be applied to low cost 360-degree video contents such as virtual reality.

ABSTRACT. In this paper, we focused on distributed video coding (DVC) technology to realize low-operation high-efficiency compression coding for hyperspectral images. In DVC, an original image is restored from side information (SI) by using an error correction code. In this research, the image quality of SI is improved by generating SI by the parameter used for weighted prediction to the interband interpolation method by linear interpolation. In experiments with Hamming code (15, 11) and key band interval 5, the PSNR of the conventional method is about 42.87 dB and the PSNR of the proposed method is about 46.02 dB, and the image quality is improved by more than 3 dB by the proposed method.

ABSTRACT. We developed a personal computer system for assessing pedaling responsiveness to visual stimuli and examined the effects of aging. We observed that false responses (wrong side pedaling) and long lapses >1000 ms occurred twice and 4 to 5 times more frequently, respectively, in the elderly than in the young and middle-age groups. Also, the increase in reaction time when the stimuli-pedal relationship was reversed was twice grater in the elderly groups.