In the networked society, dealing with enormous amounts of video information is inevitable. Video handling is often referred to as one of the studies needed for making such information easier to handle. In this study, a new approach to video handling based on auditory information is proposed. A number of sound characteristics is extracted from the sound track of a video for sound indexing, that is, we add information to the video data showing where the particular sound is present. First, we choose a characteristic which appears on the musical sound as an index for music detection. The index is used with corresponding image data and a new video browsing interface is developed.

View full abstract

This report describes an object tracking method for video content description. Content description of video materials is very useful for video program production and browsing. But for the human, a task to extract the description from video materials is very time-consuming. So effective tools for the task are desired. Such tools must be able to extract and track the objects in the video materials that contain free camerawork, object occlusion, dynamic background. The reported method uses color information of the object and can track its behavior in the time-space of the video material. Experimetal results using real video data are also discussed.

View full abstract

We have proposed "Desk Top Program Production" (DTPP) as new program production environment. In DTPP, shapes and textures of objects are stored in data base to reconstruct arbitary pictures. We developed a new technique of shape restruction for the purpose. This technique can restruct a shape from images of an object on a turning table controlled by a computer. A merit of this technique is that it does not require special lights and information of camera angle, camera position and object position.

View full abstract

There are many methods of constructing the objects like a fog. But the objects like a fog have no shape and it have too complex density distribution to construct complete model. We introduce a method for constructing clouds using Blobby model. this method assume that scattered Blobby model which is the Blobby model added disturbingis the part of cloud. And they construct a cloud.

View full abstract

The weak string filter smoothes out noise involved in a given data while it detects step edges inherent to the original noise free data. However, a steep ramp in intensity often cannot be preserved by this filter. The weak rod filter which contains the second spatial difference in its regularizer, solves this problem, and can be realized by a double layer resistive network.

View full abstract

A slow motion replay using motion compensating interpolation is investigated. To produce satisfactory picture quality, especially the artifacts in covered and uncovered areas have to be reduced. Using motion vectors, we studied the motion estimation and detection of covered/uncovered areas at each pixel. In this paper, the study and the result of computer simulation for the generation of slow motion pictures using the proposed method have been described.

View full abstract

Two X-ray Images, one taken by higher photon energy X-ray and the other by lower one, were recorded on laser videodiscs. Then they were computer-processed in order to produce images of the vascular injected contrast material (iodine) and to improve the image quality, the resultant images were recorded on a laser video disc and displayed at real (original) frame rate. Fairly good images of the canine coronary arteries were obtained. The X-ray condition and the output image quality were analysed.

View full abstract

An X-ray computed tomographic image of an object including high absorption materials is degraded by streak artifacts, which are caused by the materials. In general the region of the high absorpsion materials could not known before image reconstruction. In a image reconstruction method proposed in this paper the portions or the projection data, which will cause the artifacts, are found and are rejected to reduce the artifacts. The method is based on ART (Algebraic Reconstruction Technique). In addition we propose the method to present good image with scarce number of iterations by improving ART.

View full abstract

We have developed a 2K×2K progressive scan color camera system. In the design process, two new key-technologies have been applied. One is the sensor operation technology at high rate. In the system, four chips of CMD (Charge Modulation Device) with 1920 (H)x1035 (V) pixel are driven at 167M pixel/sec in progressive scan mode respectively. The other technology is the four sensors pickup method in which two sensors are used for green and the rests are for red and blue. The spatial offset imaging method in the vertical direction has been applied to the two green sensors. Equivalently the number of vertical lines of green channel increases to 2070 lines which is equal to the twice number of lines in the CMD. By these technologies, the ultra high definition camera having the data rate of 334MHz pixel/sec has been realized and we are sure that the vertical limiting resolution is achieved to more than 1600TVL on a color monitor.

View full abstract

[in Japanese]

View full abstract

Metamorphosis between two images to create a visual special effect has been used in movies and TV commercials for several years. While the conventional method makes use of physical and optical techniques, morphing ; the novel electronic approach descrived here, makes use of computer processing. The morphing process first warps two images so that they have the similar shape, and then cross-dissolves the resulting images. This paper reviews the warping techniques developed for generating visual special effects appeared in recent SF movies.

View full abstract