-
Article type: Cover
Pages
Cover1-
Published: July 25, 2013
Released on J-STAGE: September 21, 2017
CONFERENCE PROCEEDINGS
FREE ACCESS
-
Article type: Index
Pages
Toc1-
Published: July 25, 2013
Released on J-STAGE: September 21, 2017
CONFERENCE PROCEEDINGS
FREE ACCESS
-
Yuuki Okada, Katsumi Konishi, Katsuyuki Shinohara
Article type: Article
Session ID: AIT2013-129
Published: July 25, 2013
Released on J-STAGE: September 21, 2017
CONFERENCE PROCEEDINGS
FREE ACCESS
This work focuses on a vehicle license plate detection from defocuses and blurred images and proposes a detection method. The proposed method provides a number based on the marginal distribution of each digit image. Experimental results show that the proposed method can detect digit numbers on vehicle license plates efficiently.
View full abstract
-
Yusuke TEZUKA, Kazuma SHINODA, Madoka HASEGAWA, Shigeo KATO
Article type: Article
Session ID: AIT2013-130
Published: July 25, 2013
Released on J-STAGE: September 21, 2017
CONFERENCE PROCEEDINGS
FREE ACCESS
Classifying images into meaningful categories is an important problem for image retrieval. This paper addresses an automatic distinction of images including natural and artificial object. Tajima et al. proposed the automatic distinction technique of natural object images and artificial ones. Their method uses boundary of the object obtained from the edge information of each image. However, this method is not applicable to distinguish images including both of natural objects and artificial ones. In this paper, we propose a block-based distinction technique which divides an image into blocks and judges whether a natural object area, or artificial one by utilizing the directional difference information of the edge connection.
View full abstract
-
Sue Han LEE, Youhei AKIMOTO, Hernan AGUIRRE, Kiyoshi TANAKA
Article type: Article
Session ID: AIT2013-131
Published: July 25, 2013
Released on J-STAGE: September 21, 2017
CONFERENCE PROCEEDINGS
FREE ACCESS
This paper describes the attempt to use extended Local Binary Pattern (E-LBP) to extract both texture and edge features for low resolution head pose classification. Gradient magnitude (GM) descriptor and texture (TX) descriptor are extracted using E-LBP and trained by a multi-class Support vector Machine (SVM). Experimental results demonstrate that the combination of both GM and TX descriptor could achieve a higher accuracy rate compared to previous methods.
View full abstract
-
MinZhi CHUA, Youhei AKIMOTO, Hernan AGUIRRE, Kiyoshi TANAKA
Article type: Article
Session ID: AIT2013-132
Published: July 25, 2013
Released on J-STAGE: September 21, 2017
CONFERENCE PROCEEDINGS
FREE ACCESS
This paper describes a preliminary study of the Asian male face classification based on personal attractive preference. Face classification based on attractiveness in male has been lacking attention. Some of the conventional methods are gender and ethnicity specific and are not able to handle male faces and faces from other ethnicity group without modification. In this paper, we experiment on Asian male face classification based on personal attractive preference using separate eigenface reconstruction method. Result shows that the separate eigenface reconstruction method is able to handle male faces and achieves averaged accuracy ranging from 6.8% to 18.8% when compared to conventional methods Eisenthal and Turkmen, respectively, with 95% confidence level according to Wilcoxon signed-rank test.
View full abstract
-
Toshiro SATO
Article type: Article
Session ID: AIT2013-133
Published: July 25, 2013
Released on J-STAGE: September 21, 2017
CONFERENCE PROCEEDINGS
FREE ACCESS
Since East Japan earthquake in 2011, there is currently strong need not only for renewable energy development but also for efficient systems with more and more energy/power saving. The purpose of this study is to realize a novel power delivery to various devices in the System in Package having plural LSIs (analogue/digital), sensors, antennas and others, which will be composed of many DC-DC converters integrated in package, the authors call it "Package-level DC power grid" or "Micro smart power grid controlled through ICT scheme". This paper describes the current status of the Package-level DC power grid for System in Package.
View full abstract
-
Yasuki Yokoyama, Takashi Miyazaki, Mitsuo Sone, Hiroaki Yamamoto
Article type: Article
Session ID: AIT2013-134
Published: July 25, 2013
Released on J-STAGE: September 21, 2017
CONFERENCE PROCEEDINGS
FREE ACCESS
Multi-directional switching median filter (Multi-directional SMF) is a method to remove impulse noise from an image effectively. When we use this method, we need to set a threshold for each image. However, a decisive method has not been established yet. We considered that the values of the edge data of the image would be the key factors in deciding the threshold. We studied the relationship between the statistic values of the edge data and the threshold for each image. Our experiments confirmed that there is some relationship between them, and we propose one threshold calculating method here.
View full abstract
-
Ryota HOSHINA, Toshinori KIRA, Yasuo TAKEHISA, Yohei AKIMOTO, Aguirre ...
Article type: Article
Session ID: AIT2013-135
Published: July 25, 2013
Released on J-STAGE: September 21, 2017
CONFERENCE PROCEEDINGS
FREE ACCESS
Fiberscope is expected to lead the new possibility in diagnosis and operation since it is easy to make it thinner than endoscope and it can be penetrated in very narrow gap of 1〜2mm. However, the quality of images acquired by fiberscope is seriously deteriorated due to the shadow caused around the perimeter of each optical fiber that transmits optical signal. In this work, we study on two noise removal methods that central point of all optical fibers in fiberscope. Through the experiments using the images acquired by fiberscope actually, the effectiveness of these methods is compared.
View full abstract
-
Toshinori Kira, Ryota HOSHINA, Yasuo TAKEHISA, Yohei AKIMOTO, Aguirre ...
Article type: Article
Session ID: AIT2013-136
Published: July 25, 2013
Released on J-STAGE: September 21, 2017
CONFERENCE PROCEEDINGS
FREE ACCESS
Endoscope using fiberscope is expected to lead the new possibility in diagnoses and operation since it is easy to make it thinner than endoscope equipped CCD like gastro camera. However, the resolution of images acquired by fiberscope is equivalent (limited) to the number of optical fibers bundled, and particular noise so-called honey-comb noise is caused among fibers. Thus, the image quality captured by fiberscope is generally very low. In this work, we try to solve this problem and improve the resolution of images acquired by fiberscope by employing noise removal using Voronoi tessellation as well as stochastic super-resolution.
View full abstract
-
Ryoji Suzuki, Yoshiki Tanaka, Yohei Akimoto, Aguirre Hernan, Kiyoshi T ...
Article type: Article
Session ID: AIT2013-137
Published: July 25, 2013
Released on J-STAGE: September 21, 2017
CONFERENCE PROCEEDINGS
FREE ACCESS
The color compensation method, which recovers the color perception of elderly people making similar to the one of young people, entirely amplifies color signals to compensate lenticular yellowing and pupil diameter reduction by aging. Thus, the color signals in the original image sometimes exceeds the range (0〜255) of RGB color system, which causes undesirable effect on the output (corrected) image by rounding operation of such pixels into the displayable range. From this point of view, in this work, we examine which color exceeds the displayable range and how the color distribution of the input image affects on the output image in this color compensation method by using 464 colors in the data set and several natural images.
View full abstract
-
M. Azmil Azan Jani, Kang Leng Chiew, KokSheik Wong
Article type: Article
Session ID: AIT2013-138
Published: July 25, 2013
Released on J-STAGE: September 21, 2017
CONFERENCE PROCEEDINGS
FREE ACCESS
In this work, a steganalysis method is proposed to identify half-tone stego images. The proposed steganalysis eliminates the need to perform inverse half-toning as in the conventional methods, and it detects the stegos generated by DHST and DHED. A suspicious half-tone image is first filtered to generate a pseudo-grayscale image and converted into a binary image. The dark regions in the output binary image are utilized as the mask to generate an intermediate image on which the proposed pattern localization analysis is carried out. 25 signal-to-noise ratios are collected to form the discriminating feature vector. This feature vector is then fed into the support vector machine for training and validation purposes. Results suggest that the proposed steganalysis technique can distinguish half-tone stegos from cover images with precision of 0.9603 and 0.8407 for DHED and DHST, respectively, even when the secret message length is merely 10% of the maximum payload.
View full abstract
-
Kazuki Minemura, Koksheik Wong
Article type: Article
Session ID: AIT2013-139
Published: July 25, 2013
Released on J-STAGE: September 21, 2017
CONFERENCE PROCEEDINGS
FREE ACCESS
In this paper, we study on hiding external information into JPEG-XR compressed image. Our proposed method is based on flipping sign of the selected FCT (Forward core transform) components. Specifically, 4 × 4 blocks are first categorized into smooth and complex classes based on block energy. Then, FCT from the complex blocks are selected to hide information. To evaluate the proposed method, we consider three criteria, namely, image quality of the output image, available payload, and output image file size. Experimental results show that the proposed method achieves satisfactory payload and high image quality while maintaining the original file size.
View full abstract
-
Yutaro AOKI, Osamu UCHIDA, Kazumi TERADA, Kiyonari FUKUE, Kohei CHO
Article type: Article
Session ID: AIT2013-140
Published: July 25, 2013
Released on J-STAGE: September 21, 2017
CONFERENCE PROCEEDINGS
FREE ACCESS
The Great East Japan Earthquake that occurred on March 11, 2011 brought serious damage in the Tohoku district in Japan. There have been many examples of utilization of satellite image analysis to grasp the situation of damage. However, henceforth, it is more important to monitor how the destructed environment will recover for a long time. From such a background, our research group is working on a project such as "A project on the recovery monitoring of the destructed environment in the disaster area by satellite data analysis and field work and the practice of earth environmental education." The main purposes of the project are to monitor the status of the environment recovery from both aspects of satellite image analysis and field work with high school students in the disaster area, and to practice earth environmental education for young people through such work. Moreover, to construct Web-based system to present the status of the environment recovery and the reconstruction of the disaster area clearly and to transmit this information all over the world are also purposes of the project. In this paper, the part of the progress of the project, especially the topic of the prototype construction of the Web-based system to present the status of the earthquake disaster revival is described.
View full abstract
-
Shun KITAHARA, Osamu UCHIDA
Article type: Article
Session ID: AIT2013-141
Published: July 25, 2013
Released on J-STAGE: September 21, 2017
CONFERENCE PROCEEDINGS
FREE ACCESS
In recent years, in sporting event such as baseball and tennis, quantifying various data such as the movement of the ball and the players is utilized to improve the players' skill and the tactics of the match. Though there exists equipment that can record various kinds of data of sports automatically, they are expensive in general and not everyone can use them easily. In this study, we focus on tennis and propose a method to record the players' movements, the trajectories of the ball, and the positions of the bounds and strokes from tennis videos taken by commonly-marketed digital video camera automatically, and to provide players with useful information comprehensibly. The inter-frame difference method is used to extract the ball and a particle filter is used to extract and track the player. The experimental results show that the proposed method is sufficiently practicable for analyzing tennis matches.
View full abstract
-
Mei KODAMA
Article type: Article
Session ID: AIT2013-142
Published: July 25, 2013
Released on J-STAGE: September 21, 2017
CONFERENCE PROCEEDINGS
FREE ACCESS
Due to the popularity of mobile terminals, a large number of people can also use the contents over the network. The cached delivery system of video content is useful in the network environment in which many people use a lot of video contents. In addition, the content management method based on quality is one of the important issues to provide various quality of the content. Then, we had proposed the content management method using scalable video data, and the cached content delivery system. In this study, we focus on the variation of the video quality in the non-caching content at the top. We consider the situation for the variation in the access model.
View full abstract
-
Eiji WATANABE, Takashi OZEKI, Takeshi KOHAMA
Article type: Article
Session ID: AIT2013-143
Published: July 25, 2013
Released on J-STAGE: September 21, 2017
CONFERENCE PROCEEDINGS
FREE ACCESS
In this report, the dominant behaviors of the speaker and audience in the lecture are analyzed by using the time-series model. First, we detect as a feature for behavior the number of skin-colored pixels in face region of the speaker and audience. Next, we construct time series models for these features for the speaker and audience. Furthermore, we analyze the synchronization phenomena in the speaker and audience by comparing time series models.
View full abstract
-
Msafumi KOSUGI, Osamu UCHIDA
Article type: Article
Session ID: AIT2013-144
Published: July 25, 2013
Released on J-STAGE: September 21, 2017
CONFERENCE PROCEEDINGS
FREE ACCESS
An effective combination of music and video images generates efficacious interaction between auditory and visual processing. Then, these techniques are used at concert halls, on promotional videos, or on TV commercials. However, the professional knowledge and/or tools are necessary to create video images considering the interaction between auditory and visual processing. From this kind of circumstances, we have proposed a method to create a video image well-suited to the music based on its characteristics automatically. However, the peak such as chorus section of the music has not been taken into account in the method. Then, in this study we tried to create a video better-suited to the music by extracting chorus section based on both the music activation level that is obtained by using the root mean square of the volume of the low-tone and the high-tone and the skewness of the middle-tone. The results of the verification experiment show the effectiveness of the method.
View full abstract
-
Article type: Appendix
Pages
App1-
Published: July 25, 2013
Released on J-STAGE: September 21, 2017
CONFERENCE PROCEEDINGS
FREE ACCESS
-
Article type: Appendix
Pages
App2-
Published: July 25, 2013
Released on J-STAGE: September 21, 2017
CONFERENCE PROCEEDINGS
FREE ACCESS
-
Article type: Appendix
Pages
App3-
Published: July 25, 2013
Released on J-STAGE: March 30, 2018
CONFERENCE PROCEEDINGS
FREE ACCESS