-
Pages
11A-0-
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
-
Hiromasa FUKUI, Tomonobu YOSHINO, Sei NAITO
Pages
11A-1-
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
In the case of subjective quality assessment on high resolution video, an observer has to vote by viewing the limited area due to the short viewing distance. In order to analyze the effect of the viewing condition on the evaluation result, we conducted two experiments: First, eye tracking data was acquired during viewing 4K video sequences. Second, viewing area was divided into two regions, and observer evaluated the subjective quality of the divided region.
View full abstract
-
Hirotaka SAKAMOTO, Hideki YOSHII, Koji MINAMI
Pages
11A-2-
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
We have developed a rear-projection display system, which has various screen shapes for automobiles to fit the shapes to the design in a cabin. Its projection optical system is applicable to various screen shapes without redesigning the system. In this paper, a signal processing to correct its distortion is described.
View full abstract
-
Shota Nakahara, Satoru Okagaki, Keiji Nakamura, Kuniko Kojima, Masahik ...
Pages
11A-3-
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
We have developed a rear projection display in a car. In rear projection system, an optical engine and a screen are separated. Therefore it appears that an image shakes by the vibration of a car. We measured tolerance of image shake by a subjectivity evaluation examination.
View full abstract
-
Masaru Nakano, Daisuke Kubota, Ryo Hatsumi, Yasuhiro Niikura, Tetsuji ...
Pages
11A-4-
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
For mobile(e-book) applications, we fabricated a prototype reflective liquid-crystal display panel with c-axis-aligned crystal indium-gallium-zinc oxide field-effect transistors. The panel shows a very small reflectance change of less than 1.22% in low-frequency driving, enabling low-frequency driving where flickers are barely perceived by human eyes.
View full abstract
-
Taku OONO, Hirohiko FUKAGAWA, Takahisa SHIMIZU, Toshihiro YAMAMOTO
Pages
11A-5-
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
Top-emitting iOLEDs were successfully fabricated by optimization of electrode structures to reduce sputtering damage for organic layers. The device characteristics of the top-emitting iOLED are almost the same as that of conventional bottom-emitting iOLED employing the same emitting material.
View full abstract
-
Pages
11B-0-
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
-
Akifumi MOCHIZUKI, Kousuke MATSUSHIMA, Keiichi UCHIMURA, Gou KOUTAKI, ...
Pages
11B-1-
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
The silicon carbide filter is used to suppress the particulate matter emitted by diesel vehicle. However performance of the filter is changed by deposition of particulate matter. Therefore it is necessary to observe the inside of the filter to develop the filter optimally. This work proposes method of 3D-Neutron Computed Tomography using I-MAP and TV method to observe the inside of the silicon carbide filter. We compared the proposed method with the conventional method by simulation.
View full abstract
-
Takahiro TANAKA, Norihiko KAWAI, Yuta NAKASHIMA, Tomokazu SATO, Naokaz ...
Pages
11B-3-
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
Applying a conventional image inpainting method only once does not always generate plausible textures, and this problem is usually solved by repeatedly applying image inpainting with manual reselection of failure regions. This paper proposes a method that automatically detects failure regions in a resulting image using a convolutional neural network.
View full abstract
-
Kohei Inoue, Shunsuke Hama, Kenji Hara, Kiichi Urahama
Pages
11B-4-
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
We propose a method for inpainting images by an extrapolative filter which outputs at each damaged (or missing)
pixel a weighted average of extrapolated pixel values from neighboring undamaged and inpainted pixels. Experimental
results show that the proposed method can inpaint damaged regions with smaller blur than conventional methods.
View full abstract
-
Shiori YAMAGUCHI, Keita HIRAI, Takahiko HORIUCHI
Pages
11B-5-
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
Smoke removal in images are the useful technology for presenting clear appearances. In this study, we propose a novel video smoke removal method based on dark channel prior and frame alignments. Our method can provide effective smoke removal results in dynamic scenes.
View full abstract
-
Pages
11C-0-
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
-
Masayoshi TOMIZAWA, Yoshitsugu MANABE, Noriko YATA
Pages
11C-1-
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
Stereoscopic image device expanded the color gamut has been introduced. However it can’t perform the three-dimensional presentation with the faithful color reproduction. This paper proposes an accurate color reproduction of a three-dimensional image display device with wide color gamut.
View full abstract
-
Takuya KATO, Yoshitugu MANEBE, Noriko YATA
Pages
11C-2-
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
When we digitize pictures, texture information that accrued by such as the gloss or the roughness is lacked. This paper proposes a method of image synthesis and texture retaining of big picture by video image. Also the texture is reproduced by displaying a picture matched with the angles.
View full abstract
-
Daisuke KOBAYASHI, Noriko YATA, Yoshitsugu MANABE
Pages
11C-3-
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
Image features for material recognition has developed manually. We constructs neural network for material recognition by deep learning, and attempt to acquire image features directly.
View full abstract
-
Tadashi MATSUMOTO, Noriko YATA, Yoshitsugu MANABE
Session ID: 11C-4
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
Texture of object depends on the physical properties of the object surface. However, the relationship between the texture and physical properties have not been clarified. This paper proposes a method to quantitatively represent the metallical value of the metal balls.
View full abstract
-
Kenjiro SAITO, Leo MIYASHITA, Yoshihiro WATANABE, Masatoshi ISHIKAWA
Session ID: 11C-5
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
BRDF measurement by nonlinear optimization needs a lot of time for prototyping. As a rapid BRDF measurement, there is the method based on algebraic solution. However, this method assumes to use only Beckmann distribution as microfacet distribution function. We propose the extension of this method by using GGX distribution.
View full abstract
-
Session ID: 12A-0
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
-
Mayumi KAWANA, Mitsunobu OKUDA, Yasuyoshi MIYAMOTO, Eiichi MIYASHITA, ...
Session ID: 12A-1
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
We have proposed new storage devices utilizing current-driven domain wall motion along magnetic nanowire, in order to achieve ultra-high data transfer rate. From simulation for formation of magnetic domains in magnetic nanowire medium with a soft magnetic underlayer(SUL), it was shown that magnetic domain is formed stably when wide SUL is located under magnetic nanowire. Furthermore the use of SUL is an effective method of achieving a high speed recording.
View full abstract
-
Tomomi TAKAGI, Yosuke HORI, Toshikatsu SAKAI, Hiroshi OHTAKE
Session ID: 12A-2
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
We have been studying a novel RGB-staked image sensor using organic photoconductive films. Electron-accepter doped organic accepter films and electron blocking layers were fabricated. External quantum efficiency (EQE) of these films were evaluated, and doping concentration such as 8:2 was 80％ of the EQE.
View full abstract
-
Ryo HASEGAWA, Keita YASUTOMI, Hiroki KAMEHAMA, Yoon Hungjune, Sumeet S ...
Session ID: 12A-3
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
This paper suggests the lock-in pixel using the fully-depleted pixel of the backside-illumination type, which enable the high sensitive detection of near infrared light or X rays . We present simulation results for this structure and evaluation of the test chip.
View full abstract
-
Tomohiro NAKAMURA, Takahiro YAMASAKI, Ryohei FUNATSU, Hiroshi SHIMAMOT ...
Session ID: 12A-4
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
The 133-Megapixel CMOS image sensor, which is fabricated with 1D-stitching, has sensitivity difference between each
exposure area. We have found that the boundary area has sensitivity difference with other exposure area. We have tested correcting sensitivity difference using different correction functions for each exposure area, and have confirmed obscuring the stitched boundary.
View full abstract
-
Takeo FUJITA, Toru AOKI, Narihiro MATOBA
Session ID: 12A-5
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
Imaging devices using fish-eye lens have been in demand recently for surveillance or on-vehicle, because of wide-angle of view. However, fish-eye images need distortion correction to grasp shape or position of objects precisely. We studied high-precision distortion correction method for fish-eye image, using index value combined with height from camera and distance from optical axis.
View full abstract
-
Session ID: 12B-0
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
-
Mitsuhiko KANAMORI, Ichiro MATSUDA, Yusuke KAMEDA, Susumu ITOH
Session ID: 12B-1
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
Recent video coding standards such as H.264/AVC employ intra prediction to exploit inter-block correlations of video signals. In this paper, we investigate a method to estimate prediction modes at the decoder side to reduce the amount of side information needed for the intra prediction.
View full abstract
-
Takayuki SHIKAKURA, Ichiro MATSUDA, Susumu ITOH, Yusuke KAMEDA, Shin-i ...
Session ID: 12B-2
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
For efficient coding of holographic images used for fluid flow measurement, we have proposed a two-stage coding
technique utilizing a JPEG framework. In this paper, we introduce a technique of brightness compensation into bi-directional motion compensated prediction to exploit temporal correlations of the JPEG images.
View full abstract
-
Motohiro KIZAKI, Yusuke KAMEDA, Ichiro MATSUDA, Susumu ITOH
Session ID: 12B-3
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
We previously proposed a lossless coding method using block-adaptive prediction for light eld camera data. In this parper, formation of blocks used in the adaptive prediction is investigated by considering periodic structure of a micro lens array placed in front of an image sensor.
View full abstract
-
Shota ORIHASHI, Masaru TAKEUCHI, Rintaro HARADA, Yasutaka MATSUO, Jiro ...
Session ID: 12B-4
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
In this paper, we propose a method to control H.265/HEVC encoding for 8K videos. Our method estimates motion complexity by local feature matching between frames, selects an optimal prediction mode and determines search ranges of motion vectors. Using our method, we achieved highly efficient and low computation encoding.
View full abstract
-
Tomoki IGARASHI, Yusuke KAMEDA, Ichiro MATSUDA, Susumu ITOH
Session ID: 12B-5
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
We previously proposed a lossless coding method for color quantized images. In this method, a 3D position of each representative color in a RGB color space is directly predicted from neighboring pels. This paper compares three kinds of the prediction techniques from the viewpoint of trade-off between prediction efficiency and the required side information.
View full abstract
-
Kazuki SEKINE, Satoshi ITO, Yoshifumi YAMADA
Session ID: 12B-6
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
A new image reconstruction method using equi-spaced under-sampled signal is proposed. Our method has an advantage over standard compressed sensing in that the quality of the image does not depend on the randomness of the sampling trajectory. Simulation studies and experiments show that the proposed method using the Curvelet transform has almost the same peak signal-to-noise ratio as that of a compressed sensing reconstruction.
View full abstract
-
Session ID: 12C-0
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
-
Hidetaka MIYAJIMA, Takuzi SUZUKI, Noriko YATA, Yoshitsugu MANABE
Session ID: 12C-1
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
This paper proposes a system for augmented reality exhibition without handheld smart devices. The system detects the point of view and display information at the position appropriate to each viewer. Inconvenience of the visitors who have to watch the contents and its description in turn will be solved.
View full abstract
-
Tomomi YOKOTA, Takuzi SUZUKI, Noriko YATA, Yoshitsugu MANABE
Session ID: 12C-2
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
The early modern Kosodes are preserved carefully because of their historical value, so we cannot wear these Kosode. However many images of Kosode are available. This paper proposes a method to reproduce a form of dressing virtually by the pattern extraction from the image of Kosode.
View full abstract
-
Hajime SASANUMA, Yoshitsugu MANABE, Noriko YATA
Session ID: 12C-3
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
AR technology is used often in applications that simulate the placement of furniture. This paper proposes a
method that can remove real object from the real environment (DR) and add virtual object to the real environment (AR)
including the region of removed object by using depth camera.
View full abstract
-
Koki KOSAKA, Naoki HASHIMOTO
Session ID: 12C-4
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
Radiometric compensation methods have been proposed. However, these methods do not support various viewpoints because specular deflection is not considered. In this study, we observe appearance of screen which has specular reflection from various angle and discuss the method to reduce appearance changes.
View full abstract
-
Naoki SHINOZUKA, Yoshitugu MANABE, Noriko YATA
Session ID: 12C-5
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
In the observation of human’s situation, it is important to consider their privacy. This paper proposes a cup mounting some sensors. It can acquire data of temperature and pressure which are affected by change of contents conditions. We try to estimate indirectly user's situation based on these data.
View full abstract
-
Pages
13B-0-
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
-
Chikara MIYAJI
Pages
13B-1-
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
It is very important to perform slow motion, multi-cameras, and step-by-step for sports specific browsing on the
Internet video. The author implements slow motion, and multi-camera capabilities using serialize method. Also combination of
streaming and downloading still frames make possible to perform forward/backward step-by-step and flexible thumbnail action.
These make to see sports movements quickly and precisely on the Internet.
View full abstract
-
Takanori Abe, Chikara Miyaji, Kazuto Yoshida, Hiromasa Oku
Pages
13B-2-
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
Immediate feedback of sports performance to players would be important factor for their technique im-
provement. This presentation reports a realtime feedback system of a table tennis ball rotation for service practices.
View full abstract
-
Yuta KOIKE, Toru IZAWA, Tomohira TABATA, Masaru TAKAI, Hiromasa OKU
Pages
13B-3-
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
In this paper, a new tracking algorithm for 1ms auto pan-tilt is proposed. The algorithm can handle the
case when the target is a white object, or the case when the target and the background have the same color.
View full abstract
-
Airi SAKAUSHI, Aiko UEMURA, Shingo MURAOKA, Jiro KATTO
Pages
13B-4-
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
It is difficult for beginners to play the guitars with correct sounds. This paper suggests a guitar playing support system using image processing. In detail, our system recognizes string, fret, fingering, stroke, and outputs sound corresponding ingering. We provide beginners the system of fingering practice while listening to the correct sound.
View full abstract
-
Shunki YASAKA, Takashi KOMURO, Hiroshi KOYASU, Hitoshi MAEKAWA
Pages
13B-5-
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
In 3D nger tracking for supporting piano performance, Kalman Filter is a useful way of
tracking. This study improves the tracking precision in Kalman Filter by calculating errors using
epipolar constraints, and improves tracking speed by restricting the detection area.
View full abstract
-
Fumiya KOH, Masaya KUROKI, Yoshihiro FUJITA
Pages
13B-6-
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
We have studied a new template matching method to measure a soccer ball rotation-speed for motion locus analysis
and showed effectiveness of our proposed method.
View full abstract
-
Pages
13C-0-
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
-
Koki IKEDA, Yoshitugu MANABE, Noriko YATA
Pages
13C-1-
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
In virtual shoe fitting, it is difficult to
simulate
for all k
inds of shoe
s
using
one
3D data of foot
.
T
o
realize such
kind of simulation, this paper proposes an estimation method of the skel
e
t
on position
based on
3D
data of foot in order to
deform the 3D data
.
View full abstract
-
Junya MATSUKI, Noriko YATA, Yoshitsugu MANABE, Hirohito HIROHASHI
Pages
13C-2-
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
We
try
to
estimate
shoes
size
that fitted to
the consumer
’s feet and
to visualize result of three
-
dimensional
data
using
aligning and collision detection by
processing
three
-
dime
nsional point clou
d
.
In the end , we aim at
implementation of
virtual fitting system of shoes and feet.
View full abstract
-
Ryosuke ABO, Yoshihiro FUJITA
Pages
13C-3-
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
W
e propose a
script
ing
method for making
both Web
and
CG
video content
s
.
Hybrid
scripting
rules were defined by HTML(Hyper Text Langua
ge) and TVML(TV
p
rogram
M
aking Language
).
We have
verified an
experimental
system a
nd
some results were shown.
View full abstract
-
Pages
14B-0-
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
-
Yu MATSUI, Tatsuki MURAKAMI, Yoichi KAGEYAMA, Chikako ISHIZAWA, Makoto ...
Pages
14B-1-
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
In order to manage river bank, we proposed a method to classify vegetation using Support Vector Machine
(SVM). In this paper, we analyzed the result of classification to make this method become more high performance. The
results show that it may improve a system to reclassify the domain indicating the middle values in the R-element.
View full abstract
-
Naoto FUJITA, Yoichi KAGEYAMA, Chikako ISHIZAWA, Makoto NISHIDA
Pages
14B-2-
Published: 2015
Released on J-STAGE: May 31, 2018
CONFERENCE PROCEEDINGS
OPEN ACCESS
In a previous method, we have proposed a method for detecting a signboard area from the scene images. In order
to improve processing speed and accuracy, this paper proposes an algorithm for extracting the signboard area in scene images,
using image features. The experimental results showed clearly that the proposal technique can extract correctly two or more
signboards in a scene image.
View full abstract
