All issues
Predecessor
32.56
Displaying 1-23 of 23 articles from this issue
- |<
- <
- 1
- >
- >|
-
Article type: Cover
Pages Cover1-
Published: December 11, 2008
Released on J-STAGE: September 20, 2017
CONFERENCE PROCEEDINGS FREE ACCESSDownload PDF (12K) -
Article type: Index
Pages Toc1-
Published: December 11, 2008
Released on J-STAGE: September 20, 2017
CONFERENCE PROCEEDINGS FREE ACCESSDownload PDF (89K) -
Gentaro TAKEDA, Tomohiro YENDO, Toshiaki FUJII, Masayuki TANIMOTOArticle type: Article
Session ID: BCT2008-155
Published: December 11, 2008
Released on J-STAGE: September 20, 2017
CONFERENCE PROCEEDINGS FREE ACCESSIn this paper, we present a new image acquisition system. The proposed system can capture the dynamic scene from all-around views. This system consists of 360 convex-mirrors, a tilted flat spinning mirror, and a high-speed camera. 360 convex-mirrors are put around an object and these form virtual images of it. The size of this system is smaller than that of early system since convex-mirrors form reducing virtual images. High-speed camera can acquire multi viewpoint images by mirror scanning. Here, we simulated this system with ray tracing and confirmed the principle. And we succeed to compensate distortions on these pictures.View full abstractDownload PDF (1222K) -
Tehrani Mehrdad Panahpour, Akio Ishikawa, Shigeyuki Sakazawa, Atsushi ...Article type: Article
Session ID: BCT2008-156
Published: December 11, 2008
Released on J-STAGE: September 20, 2017
CONFERENCE PROCEEDINGS FREE ACCESSIn this research, we address the problem of 3-D model generation from disparity maps. Given an inaccurate 3-D model, dense multiview images are generated. Disparity maps between each pair are estimated using stereo matching algorithm. The disparity maps are projected into space as depth candidates. The kernel classifier is applied to the depth candidates for each layer of candidates from ground, and the candidates with higher probability values are selected for further fine-tuning process. In fine-tuning step, the best location of the border of the objects in each layer from ground is determined using dynamic programming. Finally the 3-D model is generated. Using the generated 3-D model, we regenerate the dense multiview images and disparity maps. We perform the same process for several iterations until the changes in the generated 3-D model are small. Experimental result shows we can eventually enhance the quality of the 3-D model within a few iterations in comparison with the starting 3-D model.View full abstractDownload PDF (543K) -
Sven FORSTMANN, Jun OHYAArticle type: Article
Session ID: BCT2008-157
Published: December 11, 2008
Released on J-STAGE: September 20, 2017
CONFERENCE PROCEEDINGS FREE ACCESSWe present an efficient SIMD optimized implementation of the parallel voxel-based forward projection algorithm. The algorithm rasterizes RLE encoded volume data in a front to back manner by utilizing a specialized view transform. In addition to the original method, our implementation achieves a significant speedup by utilizing a multi-segment culling mechanism in combination with a binary visibility map. We show that it is possible to uncompress and visualize large volume data directly from GPU memory without streaming data from slower CPU memory.View full abstractDownload PDF (1544K) -
Mikio SASAKI, Hideaki NANBAArticle type: Article
Session ID: BCT2008-158
Published: December 11, 2008
Released on J-STAGE: September 20, 2017
CONFERENCE PROCEEDINGS FREE ACCESSIn this paper, we propose a method for pedestrian detection and tracking which is based on MPEG video processing. Firstly AC power of frame difference for each block of pixels is calculated through DCT coefficients and is evaluated as a confidence measure of the pedestrian. Secondly, similarity of the motion between neighboring MPEG macroblocks is calculated and evaluated as a classification measure for moving objects. Finally these two measures are integrated and judged by the specified rule base to output recognition results. This first stage method has independently attained the maximum recognition rate of more than 90% per second. As the second stage algorithm for the moving camera, a multivariate time series analysis using the PCA (Principal Component Analysis) has been tried. Also future works on adaptation and behavior modeling are investigated.View full abstractDownload PDF (1519K) -
Tomohito Hamanaka, Akira Kubota, Yoshinori HatoriArticle type: Article
Session ID: BCT2008-159
Published: December 11, 2008
Released on J-STAGE: September 20, 2017
CONFERENCE PROCEEDINGS FREE ACCESSRecently, many approaches have been studied for generating free-viewpoint image based on real captured images. This paper presents an improved method for generating an intermediate image from differently focused stereo images for a scene of two depths, The presented method is based on linear space-invariant filtering that is independent of the scene structure. The main advantage of the presented method is to reduce the number of the required images from 4 to 2 while keeping the quality compared with our previously presented method.View full abstractDownload PDF (2574K) -
Shunichi KITAHARA, Nari TANABE, Toshihiro FURUKAWA, Hideaki MATSUE, Sh ...Article type: Article
Session ID: BCT2008-160
Published: December 11, 2008
Released on J-STAGE: September 20, 2017
CONFERENCE PROCEEDINGS FREE ACCESSWe propose a Kalman filter based restoration algorithm for degraded images. The algorithm aims to achieve high quality image restoration for blur and noise from the canonical state space models with (i) a state equation of original image, and (ii) a observation equation of original image and noise. The remarkable feature of the proposed method is realization of high performance image restoration without sacrificing original image despite simple image restoration using only Kalman filter algorithm. We show the effectiveness of the proposed method, which utilized the Kalman filter theory for the proposed canonical state space models with the colored driving source (original image), using numerical results and subjective evaluation results.View full abstractDownload PDF (1504K) -
Takuya KURAISHI, Masashi ITO, Akinori ITO, Shozo MAKINOArticle type: Article
Session ID: BCT2008-161
Published: December 11, 2008
Released on J-STAGE: September 20, 2017
CONFERENCE PROCEEDINGS FREE ACCESSMultiple Description (MD) Coding is one of effective methods for concealing burst packet loss. This method divides source information into multiple streams, and adds them correlation using redundant information. Utilizing the redundant information, the source can be fairly recovered if packet losses occur during the transmission. In this paper, we propose a method of MD Coding for Flash Video (FLV) based on bitstream pattern reconstruction. The effectiveness of the proposed method is examined for actual video data with packet loss simulations. Our proposed method showed almost equal quality with related method, but only needed a little redundancy. This result supported the proposed method to be effective for concealing burst packet loss.View full abstractDownload PDF (938K) -
Keiichi SAWADA, Seok KANG, Yuji SAKAMOTOArticle type: Article
Session ID: BCT2008-162
Published: December 11, 2008
Released on J-STAGE: September 20, 2017
CONFERENCE PROCEEDINGS FREE ACCESSIn this paper, we propose a MPEG video scrambling method that have the ability to control the concealed level considering visual characteristics. In the proposed method, a target PSNR of the scrambled video is appointed, and the quantized DCT coefficients in I-frames are changed accodring to the target PSNR. The proposed method corrects the amount of changes considering visual characteristics, so the characteristics of the video don't influence the concealed level. In addition, the proposed method can suppress an increase of data size by restricting the range of cofficients scrambled. Subjective evaluation test show that the proposed method is effective.View full abstractDownload PDF (885K) -
Yoshitaka INOUE, Jiro KATTOArticle type: Article
Session ID: BCT2008-163
Published: December 11, 2008
Released on J-STAGE: September 20, 2017
CONFERENCE PROCEEDINGS FREE ACCESSBy the popularization of HDTV, opportunities of watching high-resolution images on large-screen are increasing. For enlarging images, bicubic interpolation is widely used, but blur occurs. In this paper, we proposed super-resolution using IM-GPDCT applied to the compressed video. First, we considered the effect of padding the extended area in the process of IM-GPDCT. Second, IM-GPDCT was improved by using the difference of frequency distribution between magnification methods. The occurence of ringing was also suppressed. Finally, in addition to the IM-GPDCT applied to compressed video frames, using motion detection and multiple frames improved image quality.View full abstractDownload PDF (1038K) -
Lei LV, Tomohiro YENDO, Masayuki TANIMOTO, Toshiaki FUJIIArticle type: Article
Session ID: BCT2008-164
Published: December 11, 2008
Released on J-STAGE: September 20, 2017
CONFERENCE PROCEEDINGS FREE ACCESSIn this paper we focus on EPI (Epipolar-Plane Image), the horizontal cross section of Ray-Space, and we propose a novel method that chooses objects we want and edits scenes by using multi-view images. On the EPI acquired by camera arrays uniformly distributed along a line, all the objects are represented as straight lines, and the slope of straight lines are decided by the distance between objects and camera plane. So we propose a scheme to detect line with a specific slope. This enables an effective removal of obstacles, object manipulation and a clearer 3D scene with what we want to see will be made.View full abstractDownload PDF (1314K) -
Masaki NAKAJIMA, Ichiro MATSUDA, Susumu ITOHArticle type: Article
Session ID: BCT2008-165
Published: December 11, 2008
Released on J-STAGE: September 20, 2017
CONFERENCE PROCEEDINGS FREE ACCESSCurrently, motion compensation (MC) based on precise motion vectors, typically with half-pel or quarter-pel accuracy, is commonly used in video coding schemes. On the other hand, we have proposed a novel MC method which employs motion vectors with integer-pel accuracy and adaptive interpolation filters to compensate various motions. Recently, the method has been extended to use motion-compensated multiple frames at once. In this paper, we combine our method with an overlapped block MC technique using a smoothed window function. Simulation results indicate that the proposed method provides a coding gain of up to 1.0 dB in SNR compared with our previous method.View full abstractDownload PDF (705K) -
Shuma OKAZAKI, Ichiro MATSUDA, Susumu ITOHArticle type: Article
Session ID: BCT2008-166
Published: December 11, 2008
Released on J-STAGE: September 20, 2017
CONFERENCE PROCEEDINGS FREE ACCESSWe are developing an efficient lossless video coding scheme based on motion compensation and block-adaptive 3D prediction. In this scheme, multiple 3D predictors which use both the current and motion-compensated previous frames are switched block-by-block. In the previous work, these predictors are designed for each frame. However, if textures and/or motions in a sequence vary spatially rather than temporally, designing the predictors for each area given by intra-frame dividing and inter-frame merging would be reasonable . In this paper, therefore, appropriate conditions for such spatial-temporal partitioning are explored. Simulation result indicate that the proposed method can provide bit-rate savings of 0.01-0.06 bits/pel for tested images.View full abstractDownload PDF (1081K) -
Tomohiro SUGAYA, Takahiro SASAKI, Kazuhiko OHNO, Toshio KONDOArticle type: Article
Session ID: BCT2008-167
Published: December 11, 2008
Released on J-STAGE: September 20, 2017
CONFERENCE PROCEEDINGS FREE ACCESSThe video coding standard H.264/AVC succeeds in providing both high quality and high compression efficiency with the widely used MPEG-2, but it requires large amount of computational power mainly consumed motion estimation. Many method has been proposed for the motion estimation. We also proposed a new method using Multiple Expanded Templates (MET) based on the hierarchical motion search with both high speed motion estimation capability and search accuracy quality as well as Full Search for MPEG-2. In this paper, we here proposed the new MET algorithm that uses a search area prediction, which provide more efficient motion search with high accuracy quality by limiting a search area when the MET motion vector is similar to the median predictor. We also implemented this method to a software encoder of H.264/AVC, and evaluated it by using several test sequencies. As a result, it is shown that proposed method is about 2 times faster than EPZS in spite of no degradation of the video quality.View full abstractDownload PDF (875K) -
Masakazu YUZAWA, Akira OHMORI, Ryoichi KIMURA, Gaku HIBIArticle type: Article
Session ID: BCT2008-168
Published: December 11, 2008
Released on J-STAGE: September 20, 2017
CONFERENCE PROCEEDINGS FREE ACCESSField Trial report for 13GHz multichannel transmitted link. The transmission line model introduction by optical fiber and radio convergence.View full abstractDownload PDF (882K) -
Norihiro NISHIO, Yoichi TAKEBAYASHI, Takahiro SUGIYAMAArticle type: Article
Session ID: BCT2008-169
Published: December 11, 2008
Released on J-STAGE: September 20, 2017
CONFERENCE PROCEEDINGS FREE ACCESSIn this paper, we disccused a comparative analysis of workflows for staff works in a new compact studio for internet video programs and a conventional studio for TV programs. When we produce internet video programs to lower the cost of shooting in a compact studio. We analyzed the workflow of staffs work in the compact studio and coventional studio. Comparative analysis of this data, and support staff in a compact studio working on the guidelines was obtained.View full abstractDownload PDF (1812K) -
Yusuke ARAI, Shusei ITAGAKI, Jiro KATTO, Shinichi SAKAIDA, Kazuhisa IG ...Article type: Article
Session ID: BCT2008-170
Published: December 11, 2008
Released on J-STAGE: September 20, 2017
CONFERENCE PROCEEDINGS FREE ACCESSThis paper presents a new denoising method which uses motion estimation and quantization restriction for Motion JPEG After describing two denoising methods, mixing multiple frames using motion estimation and correction of DCT coefficients using quantization restriction, we explain our proposal which combines the both. We show several experimental results using Motion JPEG sequences and compare performances of our proposal with the existing methods.View full abstractDownload PDF (981K) -
Shusei ITAGAKI, Yusuke ARAI, Shinichi SAKAIDA, Kazuhisa IGUCHI, Seiich ...Article type: Article
Session ID: BCT2008-171
Published: December 11, 2008
Released on J-STAGE: September 20, 2017
CONFERENCE PROCEEDINGS FREE ACCESSIn this paper, we apply the fact that the image quality can be improved by synthesizing images to design of a new video coding technique. In intra-coded video, we can expect improvement of the image quality by referring to neighborhood frames at a decoder. Moreover, even for an area where the motion estimation effect is not achieved, applying two or more intra-coding methods in a periodical manner brings image quality improvement.View full abstractDownload PDF (2027K) -
Taro IIZUKA, Yue Yonghao, Yoshinori DOBASHI, Tomoyuki NIHSITAArticle type: Article
Session ID: BCT2008-172
Published: December 11, 2008
Released on J-STAGE: September 20, 2017
CONFERENCE PROCEEDINGS FREE ACCESSFrom the psychological point of view, it is said that people feel comfortable if the rhythms of sound and video are synchronized. Therefore, works for the synchronization are performed frequently in the process of content making. We are aimed at developing a computer assisted system for the synchronization, which was done manually. In our system, we detect rapid variations in the sound and the video as accents, and adjust the playback speed of the video to make the accents of the video and the sound matched. To detect the accents, we first calculate the time-varying local variances of the variation in the signals, and determine the local thresholds for the detection. Next, we extract signals which exceed the thresholds as accents and assign the exceeded amount as the weight of each accent. In the synchronization process, we first take into account both the weights and the time difference between the occurrences of the accents of the sound and the video, and search a best matched accent of the video for each accent of the sound. Then, we adjust the playback speed of the video to ensure that the difference between the occurrences of the pair of the accents is small enough that people would not feel uncomfortable.View full abstractDownload PDF (994K) -
Article type: Appendix
Pages App1-
Published: December 11, 2008
Released on J-STAGE: September 20, 2017
CONFERENCE PROCEEDINGS FREE ACCESSDownload PDF (81K) -
Article type: Appendix
Pages App2-
Published: December 11, 2008
Released on J-STAGE: September 20, 2017
CONFERENCE PROCEEDINGS FREE ACCESSDownload PDF (81K) -
Article type: Appendix
Pages App3-
Published: December 11, 2008
Released on J-STAGE: September 20, 2017
CONFERENCE PROCEEDINGS FREE ACCESSDownload PDF (81K)
- |<
- <
- 1
- >
- >|