Reports of the Technical Conference of the Institute of Image Electronics Engineers of Japan
Reports of the 284th Technical Conference of the Institute of Image Electronics Engineers of Japan
Showing 1-23 articles out of 23 articles from the selected issue
  • Yoshiaki Yamamoto, Kouhei Omote, Toru Tamaki, Bisser Raytchev , Kazu ...
    Session ID: 17-03-01
    Published: 2018
    Released: July 01, 2020
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    The color of its objects captured in the sea is different from that of the observation on the ground due to the influences of scattering and absorption of light. It is important to restore the color of underwater images in order to estimate the material of an object and its corrosion state using optical information such as color and reflectance of the surface of an object. In this research, 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, scattering (in-scattering) light is removed from an image captured in water, and the color of an underwater image is restored based on the attenuation of light due to scattering (out-scattering) light and absorption. We examine the proposed method using underwater images rendered with scattering and absorption phenomena. We also investigate lighting effects considering scattering and absorption.
    Download PDF (6298K)
  • Mutsuo SANO, Shohei NAMISATO, Keiji YAMAMOTO, Yumiko INOUE
    Session ID: 17-03-02
    Published: 2018
    Released: July 01, 2020
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    We focus a mirror neuron system which reproduce another person’s body image in the oneself brain and propose a method which makes body image reconstructed through virtual body expression by using a first-person vision based immersive VR display. Specifically, we discuss how VR visual stimulation with electrical and vibration stimulus has a relation to sense of ownership and sense of agency and what embodied interactions have an effect to rehabilitation for hemiplegia patients.
    Download PDF (1001K)
  • Kazuma UENISHI, Jaime SANDOVAL, Munetoshi IWAKIRI, Kiyoshi TANAKA
    Session ID: 17-03-03
    Published: 2018
    Released: July 01, 2020
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    While the attention to 3D point cloud processing processing technology increases,   feature point extraction with high repeatability has been demanded in order to solve some problems. Conventionally, extraction method was attempting to ensure repeatability by searching for existing points in a salient position. On the other hand, the VKOP extraction method can obtain feature points at virtual position computed by planar surfaces within a point cloud. Experiments have confirmed that VKOP feature points are robust to occlusion and sensor noise. However, this method has a problem that it depends on the performance of pre-processing (plane estimation method). To improve, a method has been proposed that evaluate the likelihood of stability of estimated planes and select planes for feature point extraction. In this paper we advanced the discussion of likelihood, and propose a method that estimate the likelihood of stability with the area of the planar surfaces. Moreover, we extract VKOP feature points the order of the likelihood in order to obtain a lot of the number of points. From experiments, we confirmed that feature points with higher repeatability than the conventional VKOP extraction method can be obtained.
    Download PDF (822K)
  • Naoko SAWADA, Masanori NAKAYAMA, Makoto UEMURA, Issei FUJISHIRO
    Session ID: 17-03-04
    Published: 2018
    Released: July 01, 2020
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Blazars are attractive objects for astronomers to observe in order to demystify the relativistic jet. They need to detect characteristic temporal variation patterns and their correlations of multiple time-varying variables in observed datasets to uncover spatiotemporal features of blazars, whereas it is difficult for them to scrutinize the datasets with traditional animated scatterplots. TimeTubes is our visualization scheme that allows them to interactively explore characteristic temporal variation patterns in six representative time-dependent variables. This article introduces a new function which supports astronomers’ efficient data analysis by automatically detecting characteristic spatio-temporal subspaces for dynamic rotations of polarization and flares of blazar intensity.
    Download PDF (2347K)
  • Jaime SANDOVAL, Kazuma UENISHI, Munetoshi IWAKIRI, Kiyoshi TANAKA
    Session ID: 17-03-05
    Published: 2018
    Released: July 01, 2020
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    3D planes detection is a core task of robot navigation and human crafted objects recognition. However, robots operate with restricted computing power and memory, hence; there is a need of highly efficient algorithms with better accuracy. In a previous work, the Sliding Voxel algorithm outperformed state of the art methods such as RANSAC and the Randomized Hough Transform in precision and computational efficiency. However, one of its limitations was the amount of planes it can detect in one execution. In order to solve this limitation, we propose the Multiscale Sliding Voxel for 3D planes detection. A method that executes the Sliding Voxel algorithm at multiple scales and provides an efficient planes consensus and segmentation.
    Download PDF (1208K)
  • - Challenge from Hiroshima-
    Tomoyuki NISHITA
    Session ID: 17-03-06
    Published: 2018
    Released: July 01, 2020
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
  • Jos Stam
    Session ID: 17-03-07
    Published: 2018
    Released: July 01, 2020
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    In this talk I present my work on fluid dynamics for the entertainment industry. The talk will introduce basic concepts of fluids and a brief history of computational fluid dynamics. Subsequently I will talk about my contributions of applying computational fluid dynamics to the entertainment industry like games and movies. I will also discuss our implementation of this technology into our MAYA animation software. In 2008 I received a Technical Achievement Award from the Academy of Motion Picture Arts and Sciences (``tech Oscar") for this work. I will also mention my work on bringing fluid dynamics to mobile devices like the Pocket PC in 2001 and the iPhone in 2008. In 2010 we released FluidFX and MotionFX for iOS and MacOS. The talk will feature many live demonstrations and animations. The talk is basically a condensed version of my book ``The Art of Fluid Animation".
    Download PDF (34K)
  • Hidehiko OKUBO, Masaki TAKAHASHI , Masanori KANO , Tomoyuki MISHINA
    Session ID: 17-03-08
    Published: 2018
    Released: July 01, 2020
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    We propose a new on-air graphics system to be utilized in live sports broadcasting. The graphics system has the object tracking function, it is possible to track the ball which is shot in actual sports field, and get the 3D position accurately, in real-time. Our system utilizes multiple Pan-Tilt-Zoom cameras to be applied to broadcasting cameras. The cameras with zoom lens are mounted on platform equipped rotary encoders, and we apply the encoder data to get the real-time camera information. Our key technologies are consisted from real-time object tracking, camera calibration for Pan-Tilt-Zoom cameras, triangulation, and 3DCG graphics system. We put these technologies into together, and experimented system accuracy with 4-camera system. Our experimental results show high accuracy enough for on-air graphics system.
    Download PDF (1073K)
  • Tsubasa SASAKI, Takafumi SAITO
    Session ID: 17-03-09
    Published: 2018
    Released: July 01, 2020
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Designing aesthetically appealing models is vital for the marketing success of industrial products. Tangential Angle Parameterization Curves (TAP Curves) are new curves having both position control and the curvature control. TAP curves in its quadrature form are not compatible with current CAD systems. In this paper, we present a method that approximates one TAP curve segment by one rational cubic Bézier curve segment. We have analyzed the behavior of approximate errors when the radius of curvature of the TAP curve has changed. We have implemented our algorithm and verified that one TAP curve segment can be approximated by on rational Bézier curve segment when the change of tangential angle is less than 90 deg.
    Download PDF (705K)
  • Masataka KUDOU, Hiroki TAKAHASHI
    Session ID: 17-03-10
    Published: 2018
    Released: July 01, 2020
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Recently,researches on automatic driving and driving assistance are actively discussed all over the world.In such researches, realization of safety driving is the most important issue.This paper, therefore, proposes a method to extract street gutter regions in residential area of Japan.In the proposed method,preprocess is applied to an frame on a road photographed from on-vehicle movie,then,street gutter regions are extracted by using Probabilistic Hough transformation and geometric linear feature.Finally, quantitative accuracy evaluation is performed on the obtained result.
    Download PDF (1086K)
  • Mei KODAMA
    Session ID: 17-03-11
    Published: 2018
    Released: July 01, 2020
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    The opportunities of viewing a TV program in a large size display and a high-definition video have recently increased. When a video has screen shake information, it is one of the important issue to prevent a viewer from VIMS (visually induced motion sickness). Then, a screen shake extraction method using histogram of motion had been proposed to prevent a viewer from screen shake, which was a factor of VIMS. However, it is a problem not to be able to detect the feature of screen shake due to small swing width or slow speed of shake. Then, we classify the situation of change of the histogram caused by screen shake in this method, and the efficiency is evaluated.
    Download PDF (591K)
  • Shinya HATSUDA, Lin MENG, Tomonori IZUMI
    Session ID: 17-03-12
    Published: 2018
    Released: July 01, 2020
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Raccoons are invasive aliens in Japan and damage ecosystem, crops, and buildings, especially, cultural heritages made of wood. We investigate vermin recognition for intelligent surveillance cameras to protect cultural heritages and we evaluated image recognition using convolutional neural networks (CNNs) with images we took in a wildlife breeding facility as training and test image sets. In this paper, raccoon images actually invading temples or shrines taken by surveillance cameras are used for evaluation. Three sets of images are used for training CNN: a set of raccoon images from a general training image set named CIFAR-100, a set taken in the facility, and a combination of the both. While CNN models trained with images from CIFAR-100 and from
    Download PDF (824K)
  • Kenya NOZAKI, Takahumi SAITO
    Session ID: 17-03-13
    Published: 2018
    Released: July 01, 2020
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    There are various flow fields on the earth, and it is necessary to visualize them to understand the influence. It is important to visualize many information on an image in visualization researches. In this research, we propose a method which can visualize orientation, velocity and structure of flow fields on an image. The outline of the method is combining oriented streamlines with brightness change and the image of line integral convolution. And we conducted a questionnaire-based evaluation experiment to compare visibility of orientation, velocity and structure on our method, arrow plot, streamline and LIC. As a result, only our method has good visibility about tree information.
    Download PDF (709K)
  • Wataru OGATA , Katsumi TADAMURA
    Session ID: 17-03-14
    Published: 2018
    Released: July 01, 2020
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    In this paper, a method which allows us to carry out a large scale spread of fire simulation in the three-dimensional space under a uniform square mesh having sufficient resolution and a fixed interval is proposed. The feature of the proposed method is that it allows us to simulate spread of fire in the three-dimensional space through considering the shape and arrangement of the building in the vertical direction. Specifically, the influence caused by the difference of location of a burning portion in the three-dimensional position can be considered by preparing the elevation value of the basement of each building which is obtained by referring the Digital Elevation Model data and by updating combustion state of each floor individually. The proposed method had been implemented in a PC and its usefulness was evaluated through applying it to an actual urban area.
    Download PDF (965K)
  • Makoto J. HIRAYAMA
    Session ID: 17-03-15
    Published: 2018
    Released: July 01, 2020
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Low vision persons need optical magnifiers to read books or newspapers. However, holding a magnifier with an exact focal distance stably and moving along with textual lines smoothly are difficult. Thus, an application of a book reading magnifier was developed for smartphones.
    Download PDF (274K)
  • Shoi WATANABE, Lin MENG, Tomonori IZUMI
    Session ID: 17-03-16
    Published: 2018
    Released: July 01, 2020
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Oracle bone inscriptions are the character used in the ancient Chinese dynasty more than 3000 years old. Deciphering OBIs is very important for study of ancient China and Kanji history and so on. Recently, various methods of OBI recognition by computer are proposed, but they assume a region of oracle bone image including exactly one OBI character is given. We investigate extraction of character regions from an oracle bone image and has proposed a method using general noisereduction filters and a Gabor filter so far. In this manuscript, the effect of the median filter and the local binarization filter is evaluated in addition to our previous method. Using Gaussian and oval Gaussian as region extraction filters in addition to Gabor filter, false reject rate (FRR) and false acceptance rate (FAR) of character regions are measured for each noise-reduction filters. Noise reduction using local binarization achieves about 10% lower FAR compared to previous one.
    Download PDF (1192K)
  • Shaolun SUN, Takafumi SAITO
    Session ID: 17-03-17
    Published: 2018
    Released: July 01, 2020
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    According to my paper, the purpose of this study was to show the cars passed in a certain time. So we make a static picture to show the vehicle information by using the video taken from the road surveillance camera. For visualization technology, the author has two proposals. Firstly, the author tried the method based on spatio-temporal cross section image. In the x y plane, time axis is t, x - t cross section image is made from the video. And we attempt to use a method by cutting and pasting parts of cars. Setting a line in the image, cutting out the part of the car where the car overlaps with this line, and sticking it to the road represented by the still image. And we want to extract the information of moving images, so we implemented functions such as extraction of foreground and background of image, vehicle speed calculation, color recognition etc. Lastly, this paper will give an efficiently way to view still images.
    Download PDF (670K)
  • Kento WATANABE, Hiroki TAKAHASHI
    Session ID: 17-03-18
    Published: 2018
    Released: July 01, 2020
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    A survey by the Cabinet Office, Labor and Welfare in Japan reported that the number of people with visual impairment is 315,000 in 2013. Tactile paving, therefore, take a very important role. Many tactile paving are, however, inappropriate. This paper describes a deficient tactile paving detection algorithm from a pedestrian view point video. At first, tactile paving surfaces regions are extracted by color thresholding in the first frame. The regions of the following frames are traced based on location of the regions extracted in the previous frame. In the following frames, I search only for the tactile paving surfaces area and its surroundings which is detected in the previous frame. After restricting the area for the tactile paving surfaces area which is detected, track the outline of the tactile paving surfaces area. As a result of the experiment, the accuracy of the deficient judgment was 67%, the accuracy of judgment of the straight line part was 65% and the judgment precision of the terminal part was 67%.
    Download PDF (866K)
  • Changlei JIANG, Xu WANG, Masanori KAKIMOTO
    Session ID: 17-03-19
    Published: 2018
    Released: July 01, 2020
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    In this research, we introduce a shading support method for animators or illustrators who draw 2D characters. Our method utilizes machine learning using TCDCN neural network. Through image recognition ability of machine learning, our system obtains position information of facial feature points of an animated character. Next, the system analyzes the acquired position information and the position of the pre-defined light source, and determines the shadow or shade generation area of the character and finally applies the shading on the 2D characters. This research could be expected to improve hand-drawn animation production efficiency.
    Download PDF (1421K)
  • Xu WANG, Taichi WATANABE, Masanori KAKIMOTO
    Session ID: 17-03-20
    Published: 2018
    Released: July 01, 2020
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    We introduce a method to automatically synthesize a city night view image from a day time photo. Our approach relies on the features of scotopic vision to adjust the image luminance in order to make the image become darker and looks like a real night scene. Then we considered the characteristics of the daytime’s city image which took a photo from overlooking views. We demonstrate some experimental results of our plausible night view image generation.
    Download PDF (3528K)
  • Kazuki ISHIKURA, Takafumi SAITO
    Session ID: 17-03-21
    Published: 2018
    Released: July 01, 2020
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Route panorama is a still image that can show all the scenes from a route. The route panorama is formed by connecting each frame of a captured a video sequence. It is used as a navigation tool and a town information search tool. In existing research, however, the target landscape was limited to urban areas. In this study, We aim at generating route panoramas of scenes from train window. We make the route panorama of all Chuo Line. When creating a root panorama from the railroad, We solved the problem of fast speed, the mixture of distant view components and near view components. Also, summarizing method of the root panorama is developed using the user questionnaire.
    Download PDF (604K)
  • Junshi Kimura , Katsumi Tadamura
    Session ID: 17-03-22
    Published: 2018
    Released: July 01, 2020
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    In recent years, people requiring assistance at a disaster such as elderly people and persons with disabilities account for more than 60 percent of casualties due to natural disasters. Consequently, the number of those casualties can be reduced through applying highly effective disaster prevention reduction measures to them. Therefore, early realization of both frameworks, safety confirmation which is carried out immediately after the occurrence of a disaster and the guidance which leads people to an appropriate evacuation facility for each of them, is desired because of their high effectiveness. A system to choose an evacuation facility which works on a mobile terminal is a candidate as a solution of this problem. However, no system providing functions and user interface design which consider about individual handicap of people requiring assistance at a disaster has been development yet. In this paper, a prototype system for choosing an evacuation facility in consideration of individual handicap of people requiring assistance at a disaster and for carrying out safety confirmation had been developed and its usefulness was confirmed through being evaluated by future users.
    Download PDF (1363K)
  • Satoshi KOUTSUHARA, Toshiaki YAMANOUCHI, Kazuhisa YANAKA
    Session ID: 17-03-23
    Published: 2018
    Released: July 01, 2020
    CONFERENCE PROCEEDINGS RESTRICTED ACCESS
    Google Earth is a wonderful application that allows everyone to browse satellite photos around the world on the Web. Urban buildings and mountains are expressed in 3D, and the user can move the viewpoint up, down, left and right by mouse operation. However, there is no autostereoscopic display function yet. Therefore, in this research, we developed a Python script which moves viewpoint as programmed in advance by using PyAutoGUI which is a Python GUI automation library and saves the screen shot in file. By synthesizing the acquired images with another program, a stereoscopic images of integral photography (IP) method can be obtained. Autostereoscopic images with both horizontal and vertical parallax are displayed. Anywhere in the world covered by Google Earth can be displayed by IP.
    Download PDF (541K)
feedback
Top