[in Japanese]

View full abstract

For the color design of outdoor industrial products, such as cars, models rendered by computer must harmonize with background photographs of each frame of the video sequence in color and shape. This is possible only using high dynamic range luminance corresponding to not only the achromatic luminance signal but also individual video signals (E_R, E_G, E_B). Portability and simplicity are also desired to obtain the data for the luminance distribution corresponding to weather changes. We developed a method for taking digital photos with a fish-eye lens and an ND filters that acquires a complete skylight luminance distribution based on a video signal (E_R, E_G, E_B).

View full abstract

In recent years, the voxel model has attracted attention as a data structure for geometric modelers. Designers can define shape models more effectively if they convert a voxel model into a boundary representation model. In general, the Marching Cubes method is used for this. However, the number of meshes generated by the Marching Cubes method is enormous, and this mesh data is unsuitable for a shape model. In this paper, we propose a technique to extract features from voxel data with a morphology operator. Our method reconstructs the meshes that were generated by the Marching Cubes method and uses their features while reducing the number of meshes. Furthermore, we propose an extended QEM to be used in the shape simplification field. Our method defines the error allowance range in the original voxel data. This gives an effective guarantee of precision for using the original voxel data. Our experimental results show that the proposed method is valid for reducing meshes.

View full abstract

We propose motion generation computed by redundant parameters of inverse kinematics (IK) from an optimized secondary skeleton constructed based on a bodily skeleton. The secondary skeleton numerically embeds arbitrary characteristics of motions into the mechanism of IK and thus preserves their plausibility for similar movements of end-effectors. The feasibility of this extended IK is demonstrated with the motions generated by deforming the trajectory of hands or feet.

View full abstract

A method for accurately controlling the flow of walking crowds by introducing a psychological model is proposed. In this model, we consider the effects of human memory and personal space fof sensing the density of a crowd and predicting collisions. The speed of each walker is determined from the crowd density, and collision-avoiding behaviors are controlled by assigning rules to each condition. Various walking motions are automatically generated through playback and transition of motion capture data. Our simulation controls the flow of crowds to imitate actual crowd phenomenon. This property increases the prediction accuracy of group behaviors in very crowded spaces.

View full abstract

We describe a novel system that achieves intuitive retrieval of 3D meshes from a database of 3D models based on the shape geometry and its functionality. By using a point-sampled representation of a 3D shape both for a query and models, our system allows user to physically and interactively express a geometrical and content-based query by simply assembling a set of ActiveCubes. The user can retrieve 3D models from the database based on their shape and can physically integrate functional annotated blocks in the query. The user intuitively defines an interactive spatial query that has geometrical and functional aspects.

View full abstract

A nonphotorealistic rendering (NPR) technique for generating canvas textures is presented. Pixels in an input image are randomly sampled and interpolated between sampled points. Canvas textures are generated on the basis of errors between the interpolated images and the original one. This NPR technique is then applied to moving images. This technique enables us to easily generate an NPR video coherent in time without flicker.

View full abstract

We developed a novel image-based rendering method that uses inputs from unstructured cameras and synthesizes free-viewpoint images of high quality. Our method uses a set of depth layers to deal with scenes with large depth ranges. To each pixel on the synthesized image, the optimal depth layer is assigned automatically based on our on-the-fly focus measurement algorithm. This method is a modified version of our prior work with another definition of the algorithm in the spatial domain, which enables us to extend the application scope for images captured by an unstructured array of cameras. We implemented the proposed method and achieved nearly interactive frame-rates on a commodity PC.

View full abstract

We have proposed a real-time IBR system that captures multiviewpoint images of a 3D dynamic scene with a single camera through thousands of lenslets, and synthesizes arbitrary views of the scene in real time. We have also proposed a method for all-in-focus rendering of arbitrary views of a 3D scene captured by an array of cameras. This paper combines these two works to achieve a new all-in-focus real-time IBR system. Our experimental results indicate that the proposed method can synthesize clearer views than the previous method.

View full abstract

Real-time rendering of realistic motion of fluids is a method that immerses a player into an interactive application such as computer games. Interaction of fluids with rigid bodies is important because fluids and rigid bodies move influencing each other. Fluid simulation based on Computational Fluid Dynamics (CFD) is useful for rendering a visually plausible behaviour of fluids. However, due to the high computational cost of CFD, real-time rendering of fluids needs a fast simulation. This paper describes the particle-based fluid simulation based on Smoothed Particle Hydrodynamics which includes interactions between fluids and rigid bodies, its fast implementation, and the manner of rendering a realistic water surface with optical phenomena such as reflection, refraction, and the Fresnel effect. The proposed method enables real-time animation of water with rigid body interaction.

View full abstract

Natural object modeling is in high demand in the application of computer graphics, especially in movies and games. Research on photoreal expression of food is, however, still rare. We focus on noodles because they are widespread while modeling realistic noodles is difficult. We propose an algorithm for automatic placement of noodes and provide a user interface for rendering an arbitrary form of noodles. A user can capture, move, and twist noodles by usinga mouse as chopsticks and/or a fork.

View full abstract

With increasing optical disk density, land/groove recording has become mainstream and accordinglytrack pitch has continued to become narrower. Therefore, to further increase recording density, it is essential to reduce cross talk and cross erase. In a system optimal recording power is required for recording. However, to determine the optimal recording power, it is necessary to evaluate cross erase characteristics in addition to recording power characteristics, which requires a complicated procedure. We have devised a recording power calibration method using a wobbled track. As a result, it was demonstrated that it is possible to determine the optimal recording power by recording and read-out a monosignal using a wobbled track to calibrate the recording power and observing a sideband wave that accompanies a mono signal. With this method, a considerable amount of work, which was needed to determine the optimal recording power, is no longer necessary. Instead it has become possible to determine optimal recording power with a simple procedure.

View full abstract

We propose a 3D display that allows multiple viewers to see 3D images from 360 degree of horizontal arc without wearing special glasses. It displays 3D images as reproductions of the light field around objects using a cylindrical parallax barrier and one-dimensional light source arrays. We have developed an experimental display system capable of displaying 3D color video images within a diameter and a height of 200 and 256 mm, respectively. Images are described by a resolution of 1254 pixels circularly, 256 pixels vertically, and refreshed at 30 Hz. The angular interval of the ray is less than 1 degree. In this case, viewers hardly perceive discreteness of parallax.

View full abstract

A high-speed and precise focusing/tracking servo system is indispensable for developing a HDTV optical disk camera-recorder. This system is needed for higher data transfer rate and recording density. In this work, we adopted a feedforward control employing the Zero Phase Error Tracking method for the focusing/tracking servo system, and this method reduced residual focusing/tracking errors compared with a feedback system. In addition, we confirmed that the method makes possible obtaining stable recordings and reproduces signals in optical disk systems with the high data transfer rate needed for broadcast use.

View full abstract

We propose a new 3D-input method using a relief-like depth map generation algorithm (REAL). REAL is based on the concept of reproducing a depth map, like the relief of a coin, from 3D input. Using REAL, a portable 3D-input system can be achieved using only a monocular camera and a photoflash. REAL is very simple and convenient compared with conventional technology that requires triangulation. In this paper, we describe an overview of the concept and the calculations of the depth map. We also report a subjective test of 3D images generated by REAL. The subjective test results indicated that observers perceive REAL images as natural 3D images.

View full abstract

Digital terrestrial television broadcasting began in Japan in December 2003. The phase noise of the local oscillator degrades the bit error rate of the Orthogonal Frequency Division Multiplexing (OFDM) transmission. We have conducted indoor experiments to find the relationship between the phase noise level and the bit error rate. We found that the degradation of the bit error rate could be altered by the offset frequency of the phase noise, and the OFDM receiver could suppress the phase noise in the low offset frequency of-20dB/dec. In this paper, we also propose new methods of regulating the phase noise level in the local oscillator.

View full abstract

We developed a prototype system to reduce asthenopia by using stereoscopic 3-D images that moved in depth direction, and evaluated the effects of the system. Accommodative step response time, profiles of mood states and refractions were measured as indexes for examining the subject's visual fatigue. In young subjects, the accommodative relaxation time tended to be shortened by watching the images. This tendency indicates that the ciliary muscle quickly relaxed. In contrast, we observed an increase of accommodation in some aged subjects while they viewed the images. These results suggest that the work of the ciliary muscle is activated by observing the images. Our results suggest the possibility that viewing stereoscopic 3-D images can reduce asthenopia.

View full abstract