Recently, due to the rapid progress of computer hardware, the 3 dimensional computer graphics have become quite familiar. In order to represent a complicated 3 dimensional object using 3DCG, its surface is usually described by a number of triangular patches. To depict an object accurately, smaller triangular patches are preferable. In regard to the drawing speed, however, larger and less triangular patches are required. To compromise the conflicting requirements, it is proposed that the complex portion is represented by small patches and the smooth portion by large ones. This paper describes a new method in which at first the characteristic cross sections of a 3 dimensional object such as a human head and a spherical surface are extracted and then characteristic points are selected by thinning. Because the characteristic points are located on either one of the characteristic cross sections, triangular patches are formed systematically. The effectiveness of the proposed method is demonstrated by several examples.

View full abstract

Volume data is widely used for the expression of 3D objects. It has various advantages, but its handling is still inconvenient because of huge size. This paper discusses a 3D extension of DCT (Discrete Cosine Transform) for the compression of volume data. 2D DCT is used in JPEG for the image data compression. In this paper, we used two sets of volume data for the evaluation of the 3D DCT : one is obtained through a CT scanner and the other is artificial one made by compater graphics. It turned out that 3D DCT can compress image data to. 1/10 sizes without dropping of image quality.

View full abstract

We propose a restoration method of 3D-models from three-view drawings with abbreviation for sheet metals. Former methods are premised on drawings which are drafted completely without abbreviation. So current methods are not practical because actual drawings usually contain some conventional abbreviation of drafting. Sheet metals with cutting, holing, and bending are chosen as the object of the study. We develop an interence method to find out those abbreviated lines, which should stand for invisible edges such as cross sections caused by cutting, silhouettes of holes, and bending lines parallel to some coordinate axis. Experiments to apply our method to practical drawings are also performed to prove the efficiency.

View full abstract

We have developed a schedule evaluation system for plant construction in order to improve the accuracy and the efficiency of plant construction schedule. This system evaluates schedules on the basis of the difficulty of construction process by using the congestion rate of working space defined for each process. This function is implemented using such steps as shown below. (1) Define working space for each detail process, and calculate volume rate of installed parts in that space as a interference measure. (2) Order the detail processes automatically to minimize the interference measure. (3) Improve the efficiency of the construction schedule, based on (a) construction simulation which shows the situation of workers and working spaces, (b) automatic work difficulty evaluation with the volume rate mentioned above, and (c) interactive adjustment of construction schedule according to simulated working situation.

View full abstract

This paper describes a newly developed real time locus indication system for a golf ball in flight. The system, named "Shot View", detects the three dimensional position of the ball according to triangulation principles using the motion angle of two cameras to track the ball, and indicates the locus on a course layout background. The experiments conducted to ascertain its applicability to golf, and the system has been applied to long drives shots at a distant hole during several golf broadcasts.

View full abstract

The fractal coding offers high compression ratios. However, because this coding method is based on the assumption that images possess some form of self-similarity, a disadvantage lies in long search time for domain blocks. In this paper, a fractal coding using quadtrce for wavelet transform coeffcients of images is examined and some properties of recoustructed images are investigated., and it is shown that thw proposed method is very effective.

View full abstract