-
Article type: Cover
1992 Volume 16 Issue 78 Pages
Cover1-
Published: November 27, 1992
Released on J-STAGE: October 06, 2017
RESEARCH REPORT / TECHNICAL REPORT
FREE ACCESS
-
Article type: Index
1992 Volume 16 Issue 78 Pages
Toc1-
Published: November 27, 1992
Released on J-STAGE: October 06, 2017
RESEARCH REPORT / TECHNICAL REPORT
FREE ACCESS
-
Kuniaki YAJIMA, Yoshiaki TADOKORO
Article type: Article
1992 Volume 16 Issue 78 Pages
1-6
Published: November 27, 1992
Released on J-STAGE: October 06, 2017
RESEARCH REPORT / TECHNICAL REPORT
FREE ACCESS
This paper considers an image reconstruction algorithm using complex-Walsh transform. First, we develop a complex-Walsh transform corresponding to the discrete Fourier transform. Next, we present an image reconstruction algorithm using the complex-Walsh transform. Finally, we examine the characteristics of this algorithm by computer simulation.
View full abstract
-
Akira NAKAGAWA, Eishi MORIMATSU, Kiichi MATSUDA
Article type: Article
1992 Volume 16 Issue 78 Pages
7-12
Published: November 27, 1992
Released on J-STAGE: October 06, 2017
RESEARCH REPORT / TECHNICAL REPORT
FREE ACCESS
Recently, a new idea for visual communication called 'Knowledge Based Coding' was proposed, and has been studied by many researchers. We have been studying Media Conversion based on this idea to convert a text to talking images, utilizing knowledges of relation between uttering and shape around mouth. So far, we implemented a PC-Based experimental system for the Media Conversion from Japanese sentences to talking images with the synchronously synthesized voice. In this report, an implementation of a face synthesis system with a friendly man-machine interface is described, which makes up a whole system of the PC-Based Media Conversion.
View full abstract
-
Ikuo ISHII, Atsushi YOKOYAMA, Hideo MAKINO
Article type: Article
1992 Volume 16 Issue 78 Pages
13-18
Published: November 27, 1992
Released on J-STAGE: October 06, 2017
RESEARCH REPORT / TECHNICAL REPORT
FREE ACCESS
It is necessary for display of virtual environment to express the natural motion parallax. For that purpose, the positions of operator eyes are measured periodically, and the corresponded scene is displayed on the screen in the time less than 50msec. It is difficult for current graphics processors to satisfy this requirement for complicated scenes. In the method described here, the definition data of virtual objects are converted into the polygon based intermediate data suitable for the fast hardware for operation of motion parallax and stored in the buffer. As the other graphics operations except for motion parallax is executed before the conversion, it is not necessary to provide the specially fast hardware for these operations. By this way, a comparatively simple graphics processor for virtual operation can be
View full abstract
-
Nobuyuki SASAKI, Nobuyuki YAGI, Kazuo FUKUI, Kazumasa ENAMI
Article type: Article
1992 Volume 16 Issue 78 Pages
19-26
Published: November 27, 1992
Released on J-STAGE: October 06, 2017
RESEARCH REPORT / TECHNICAL REPORT
FREE ACCESS
The model of DSP which includes plural multiplyers(mpys), alus, input ports, output ports, variable delays and switching network is considered. All process can be applyed to HDTV, so both HDTV and NTSC signal can be processed with this same DSP model. Especially in case of NTSC, very efficient use of mpys and alus is obtained with some algorithm which use dynamic allocation of switching network and variable delays. In this paper, the algorithm and many case studys are described.
View full abstract
-
Shoji TOMINAGA
Article type: Article
1992 Volume 16 Issue 78 Pages
27-34
Published: November 27, 1992
Released on J-STAGE: October 06, 2017
RESEARCH REPORT / TECHNICAL REPORT
FREE ACCESS
This paper describes a new method for color notation conversion between the Munsell and CIE color systems by means of neural networks. A multilayer feedforward network is regarded as a nonlinear transformer which is trained to learn a mapping between the two color spaces. We adopt the Back-Propagation learning rule for the training. The mapping is then realized in a simple network architecture in which nonlinear units are linked in parallel and in layers. The tables of data by Nickerson, et al. are used in computer simulations of training and testing phases. We determine the most effective network structure and training strategy based on experiments under different conditions.
View full abstract
-
Yau-dong Chang, Mikio Takagi
Article type: Article
1992 Volume 16 Issue 78 Pages
35-41
Published: November 27, 1992
Released on J-STAGE: October 06, 2017
RESEARCH REPORT / TECHNICAL REPORT
FREE ACCESS
The size of meteorological satellite NOAA image is about 100 mega bytes. If the correction of geometric distortion is processed all pixels of NOAA image, it will waste a lot of processing time. Here, the NOAA satellite image is split into blocks, which are all the same size. The geometric correction is only processed on the 4 contours of each block, and we use the bi-linear interpolation for the remaining pixels inside blocks. Using this method, we try to reduce the processing time of the geometric correction. In this paper, the optimal block size of satellite imagery and map imagery is experimented.
View full abstract
-
Heeday Danjo, Teiichi Nishioka
Article type: Article
1992 Volume 16 Issue 78 Pages
43-50
Published: November 27, 1992
Released on J-STAGE: October 06, 2017
RESEARCH REPORT / TECHNICAL REPORT
FREE ACCESS
Hi-Vision broadcast images have appeared on nation-wide newspapers for the first time. We conducted an experiment with 3 newspaper publishers to convert Hi-Vision images of the Olympic Games of Barcelona for press use. Scenes selected from VTR recorded MUSE programs were captured as digital data and converted from RGB to CMYK color space. The resulting printing data was supplied to newspaper prepress systems in 6.5 minutes after scene selection, utilizing ISDN and JPEG compression. We conclude that broadcast images proved to be practical enough in quality for press use. Operations such as level correction and sharpening contributed to image quality.
View full abstract
-
Ikuo ISHII, Ichiro OZEKI, Hideo MAKINO
Article type: Article
1992 Volume 16 Issue 78 Pages
51-56
Published: November 27, 1992
Released on J-STAGE: October 06, 2017
RESEARCH REPORT / TECHNICAL REPORT
FREE ACCESS
For fast reconstruction, it is useful to construct outline of characters with combination of straight line and circular arc segments. It is able to use fast hardware based on Bresenham's algorithms for the digital reconstruction of these segments. But his act generation algorithm can arc only on the grid of integral coordinates, so approximation errors are caused at such points as the center of circular arc or the connecting points, and spread all over the arc. The new algorithm described here, the center and connecting points can be specified on the real number coordinates and the most suitable outline pixels are generated. Futhermore, the straight line segments can be also reconstructed by the arc generation hardware.
View full abstract
-
Article type: Appendix
1992 Volume 16 Issue 78 Pages
App1-
Published: November 27, 1992
Released on J-STAGE: October 06, 2017
RESEARCH REPORT / TECHNICAL REPORT
FREE ACCESS