We have developed an intelligent robot camera system to support the production of broadcast television programs. The objective was to develop a system that automatically shoots scenes that are completely natural as if taken by an experienced camera operator, and that enables a single operator to control multiple cameras at the same time. We have developed an automatic program production system that is composed of several network-connected robot cameras programmed to mimic the shooting and camerawork techniques of professional camera operators. In this paper we give an overview of a prototype system using a new subject detection system and three robot cameras that are linked over a network and present a detailed summary of the trial results.

An ID recognition system is described that uses optical beacons and a high-speed image sensor. The ID sensor captures a scene like an ordinary camera and recognizes the ID of a beacon emitted over a long distance. The ID recognition system has three features. The system is robust to changes in the optical environment, e.g. complete darkness, spotlights, and sunlight. It can recognize up to 255 multiple optical beacons simultaneously. Furthermore, it can recognize beacons even over a long distance, e.g. 40 m indoors and 20 m outdoors. Implementation and evaluation of this ID recognition system showed that a mobile augmented reality system can be achieved by combining this ID recognition system with a PDA and a wireless network.

In order to improve the quality of microprocessor tests, the use of instruction sets for testing is indispensable. This paper discusses how fault coverage can be improved with a short test pattern that repeatedly samples an R number of instructions as L sets from an S number of instructions and selects from amongst these L sets those for which the fault coverage can be improved. Continuing, it argues that the processing speed can be increased by selecting a certain number of sets containing an R number of instructions from an S number of instructions. This approach proved effective in tests using software created on these principles.

In order to improve the quality of microprocessor tests, the use of instruction sets for testing is indispensable. In this paper, we will present a new method consisting of the automatic generation of a functional test pattern, formed by a combination of instructions sets and enabling the efficient improvement of the fault coverage. With this method, a test pattern is first generated to test all of an S number of instruction mnemonics. Then, for the faults that were undetected by that test pattern, an L number of sets of K number of instructions are drawn from the S number of instructions, and the set enabling the efficient improvement of the fault coverage is selected. By repeating this procedure, a high fault coverage can be obtained with a short test pattern. The effectiveness of our method was proved by the results of experiments obtained with the software that was created based on this method.

Presents a fast computation method of the normalized correlation for multiple rotated templates by using multiresolution eigenimages. This method allows the authors to accurately detect both location and orientation of the object in a scene at faster rate than applying conventional template matching to the rotated object. Since the correlation among slightly rotated templates is high, the authors first apply the Karhunen-Loeve expansion to a set of rotated templates and extract "eigenimages" from them. Each template in this set can be approximated by a linear combination of these eigenimages and it substitute for the template in computing the normalized correlation. The number of eigenimages is smaller than that of original templates and computation cost becomes small. Second, the authors employ a multiresolution image structure to reduce the number of rotated templates and location search area. For the lower resolution image, the position and angle are coarsely obtained in a wide region. Then not only searching area for the position but also the range of rotation angle of templates at the next layer can be limited to the neighbor of the prior results. The authors implemented the proposed algorithm on a vision system and realized computation time around 600 msec and achieved sub pixel resolution for translation and 0.3 degree maximum error for 360 degree rotation on the 512 by 480 gray scale image. Experimental results are shown to demonstrate the accuracy, efficiency and feasibility of the proposed method.>