-
Yusuke YONEDA, Mamoru KUBO, Masahiro FURUKAWA, Ken-ichiro MURAMOTO
Session ID: 05-02-01
Published: October 07, 2005
Released on J-STAGE: October 21, 2005
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
The horizontal and vertical falling motions were photographed using five CCD cameras from horizontal and vertical directions. We calculated some shape features and trajectory of snowflakes using video data. The spline interpolation was applied to the trajectory for calculating the angle of horizontal movement. We analyzed relationship between shape features and horizontal velocity or moving angle.
View full abstract
-
Kumiko TAKIZAWA, Mamoru KUBO, Ken-ichiro MURAMOTO, Nobuhiro NAGUMO, Ya ...
Session ID: 05-02-02
Published: October 07, 2005
Released on J-STAGE: October 21, 2005
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
In order to clarify the process of formation and its differences by weather conditions of snowflakes, it is necessary to measure and analyze the shapes of snowflakes. This paper describes a measuring system for getting two-dimensional image data using three CCD cameras mounted in different locations. In addition, 2D-Video-Distrometer (" 2DVD" ), the raindrop analyzer, was used for automatic data acquisition. By line scanning, it can get data such as drop size distribution, terminal velocity, oblateness, and silhouette image of snowflakes. We present the method for producing the three-dimensional image of snowflakes by image processing.
View full abstract
-
Masaru HOSHINO, Mamoru KUBO, Masahiro FURUKAWA, Ken-ichiro MURAMOTO
Session ID: 05-02-03
Published: October 07, 2005
Released on J-STAGE: October 21, 2005
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
In this paper, we propose image archiving system of falling snow particle for long-term observation. Natural falling snow particles were photographed using video camera. Snow particle images and that parameters were stored into the system. The functions of this system are retrieval by parameters of snow particle, and display the graph of size distribution, velocity distribution and relationship between diameter and velocity. This archiving system is useful for snow research.
View full abstract
-
Makoto KITAMURA, Fumitaka ONO
Session ID: 05-02-04
Published: October 07, 2005
Released on J-STAGE: October 21, 2005
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
In order to develop the efficient Markov model coding methods for error diffused images, it is important to reveal the relation of the number of reference pixels and the conditional entropy, by knowing the combination of reference pixels which provides the minimum entropy when the number of reference pixels are given. In this paper, a practical searching method of the effective reference pixels for error diffused images is newly proposed, and the simulation result using the proposed method is reported.
View full abstract
-
Junji KAWASAKI, Shigeto KAWAMOTO, Masashi OKAJIMA, Syuitsu MATSUMURA, ...
Session ID: 05-02-05
Published: October 07, 2005
Released on J-STAGE: October 21, 2005
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
Up to now, the image quality evaluation method by the visual model whom we are proposing has evaluated it in some binary images. Eight has been assumed to be the best in case of ..preparing.. between in a ordered dither method in the past and binary matrix size n. But the objective evaluation method is not found. The purpose of this paper is objectively to evaluate the best dither matrix size of ordered dither method. Number L of steps compared it according to dither matrix size n to a ordered dither method of 4, 8, and 16. Then, the objective evaluation was done to preparing between in a ordered dither method for number L of steps and matrix size n by the proposed approximation level and the best matrix size was examined. Girl, Moon-surface, and Milk-drop were used for the image.
View full abstract
-
Kazuaki SUZUKI, Kunio KONDO
Session ID: 05-02-06
Published: October 07, 2005
Released on J-STAGE: October 21, 2005
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
Pixel art is effective means for row resolution display. It is need long time to draw pixel art manually. Video game publishers often come to use polygons, because they want to reduce making game time and computers become high performance. But pixel art is sometimes better express than polygons. If they make pixel art easily, scopes of pixel art field expand for one method of expression. Our research purpose is to draw outline, especially size of pixel art by photograph. First, we investigate procedure of drawing and connected direction of outline to find pixel art's feature. Pixel art outline width is often one pixel. Existing methods can't draw one pixel outline. By the way, we reduce binarized picture by existing methods, then outline become unsharp grayscale image or losing outline information. To solve this problem, we propose a method making outline pictures by photograph and reduction outline. Then we can draw a binarized outline & edge image for pixel art.
View full abstract
-
Eiji YAMAMOTO, Mamoru KUBO, Ken-ichiro MURAMOTO
Session ID: 05-02-07
Published: October 07, 2005
Released on J-STAGE: October 21, 2005
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
In order to make tree crown map from forest multi-date IKONOS satellite images, we propose the method of segmentation using watershed algorithm and classification of segmented regions using maximum likelihood decision rules. We extracted spectral features for classification from multi-date satellite data
View full abstract
-
Satoshi INOUE, Kousuke IMAMURA, Hideo HASHIMOTO
Session ID: 05-02-08
Published: October 07, 2005
Released on J-STAGE: October 21, 2005
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
In recent years, moving-objects segmentation which devides video sequences into a meaningful domains is employed as pre-processing in the object base coding system such as MPEG-4 or MPEG-7. In this paper, motion detection accuracy is improved in considering the robustness of affine parameters, in order to improve in moving-objects segmentation.
View full abstract
-
Yasunori TSUCHIDA, Kousuke IMAMURA, Hideo HASHIMOTO
Session ID: 05-02-09
Published: October 07, 2005
Released on J-STAGE: October 21, 2005
CONFERENCE PROCEEDINGS
RESTRICTED ACCESS
In recent years, research and development for object-based coding which performs encoding each object individually in video sequence is furthered. In this paper, we propose moving objects extraction method which is needed for the above new coding. This method, at first, extracts rough outlines of moving objects by active contour models (snakes) using frame difference energy. Next, refinement of the rough outlines is performed by applying the watershed algorithm.
View full abstract
