[in Japanese]

View full abstract

In recent years there are studies to input the data of a hand motion in itself as a new Man-Machine Interface to realize a virtual world. Especially, the motion detection method without wearing input sensors, using only an image processing engine, is recognized as effective one, so a large number of studies have been proposed on this method. We are concerned with a finger motion. The study proposed in this paper to detect a finger motion in real time and interactively. The subjects of this paper are as follows. At first, we mark a base and tip of finger with a color tape to make easy a process to detect the coordinates of a finger in a space of 3-dimension in real time. Next, we make a finger flame model and a decision rule of a finger shape in a computer to make a finger model on display with minimum data. Lastly, we discuss the model maching method to move a model made in a computer in time to a finger motion in physical space.

View full abstract

A technique for detecting a continuous position of an object in a naturally moving image is proposed. This algorithm is composed of these steps as follows. (1) Simple object pointing by a user (2) Histogram using membership function is robust against signal noise (3) Object segmentation against a background image by "local histogram growing method" (4) Histograms in perceptual color system (Hue, Saturation, Brightness) insensitive to video-camera movement and rotation as well as deformation, shrinkage and enlargement of an object (5) Detecting of a position of an object by "optimal histogram matching method" Real-time experiments have been carried out in order to verify the effectiveness of the technique.

View full abstract

When modifying surfaces by offsetting, the deformation should be intuitive and their resultant shapes should reflect global features of their initial shapes. Physically-based approaches can generate intuitive deformation, but do not necessarily have a comprehensible geometric interpretation. This paper therefore considers this deformation as a constrained optimization problem, and proposes a Laplacian-based energy functional that has both geometric and physical meaning. Analytical solutions to this problem are presented under G^1-connections constraints.

View full abstract

In this note, we introduce and survey our CAD project being carried on in our laboratory. Main stress is placed on clarifying the difference between the drawing understanding and the drawing recognition. The former, in which we feel major interest, is intrinsically a difficult problem because it inherently contains combinatorial search to require more than polynomial time. Actually, understanding drawings is regarded as a process to recover the information lost in projecting 3-D objects to 2-D drawings. But, solid modeling by automatic understanding of the given drawings is one of the promising approach, which is described precisely in the text. Reviewing the studies performed so far, we summarize the future direction of the project and inevitable open problems left.

View full abstract

Force feedback in virtual environment is the major research objective of Iwata Lab. in the University of Tsukuba. Virtual reality is based on active human behavior, which is distinct from other information media. Operation by human hand in virtual environment requires force feedback from virtual objects. Force feedback is only realized by physical interaction with the outer world. Construction of force feedback device contains hard problems and research of this area is at an early stage. Our laboratory has been dealing with development of mechanical configuration of force feedback devices and its applications. This paper introduces following three current results: (1) Pen -based force display, (2) volume haptization, and (3) artificial life in haptic virtual environment.

View full abstract