[in Japanese]

View full abstract

A pointing input method that uses gaze detection, along with a mouse, is proposed. This method enhances the efficiency of the pointing task in a man-machine interface. In this method, a cursor is displayed at the detected gaze point on the CRT and if the cursor is on target, the operator clicks the mouse button. If it is not on target, it is moved onto the target using the mouse and then the button is clicked. This method is compared to the conventional method, which uses only a mouse, and pointing time, error rate, and gaze point movement are analyzed. The results indicate that if the gaze detection device has an error angle of 3 degrees or less (at a sight distance of 60 cm), this method is better than the conventional method in terms of pointing time and cursor controllability.

View full abstract

This paper determines the optimum color design for color image scanners that use multiple light sources, when the number of light sources is greater than three. Since it is difficult to control light source spectrum characteristics, additional, commercially available light sources are introduced. To obtain ideal color acquisition, a method is developed to minimize the sum of the square errors between the overall characteristics of the scanner and the color matching functions. Using ten different light sources, the average colorimetric error, ΔE, is 1.48. Eliminating the non-significant light sources raises ΔE to 1.89, which is still within acceptable limits for a Class A colorimeter, which has a specified tolerance of ΔE≤2.5. A total of 47 different color chips are tested to evaluate the optimally designed color scanner. The average color differences, ΔE^*_ab, is approximately unity.

View full abstract

This paper describes a new method for converting color notation between the Munsell and CIE color systems by means of neural networks. A multilayer feedforward network is regarded as a nonlinear transformer and is trained to learn the mapping between the two color spaces using the Back-Propagation learning rule. The mapping is then represented in a simple network architecture in which nonlinear units are linked in parallel and in layers. Using Nickerson's tables for computer simulation of the training and testing phases, an effective network structure and training strategy is determined based on experiments under different conditions. The conversion accuracy is demonstrated in both directions, Munsell-to-CIE and CIE-to-Munsell. The Munsell colors are displayed on a calibrated color monitor and their accuracy is confirmed visually.

View full abstract

A complementary pyramid scheme is proposed for the multiresolutional representation of images, in which various images with different resolutions are integrated. While typical pyramid coding can reconstruct multiresolutional images without phase distortion, the sum of the samples in the pyramid is larger than that in the original image. In the complementary pyramid, however, comlementary subsampling is employed and the number of samples is equal to that of the original image. Distortion in the reconstructed image is reduced by assigning eath quantization level an offset determined by the upper level quantization noise. The coding gain of a complementary Laplacian pyramid is calculated and is found to be nearly equal to that of the wavelet transform.

View full abstract

A digital audio signal transmission system that uses quadrature modulation for the video carrier is evaluated for use in 2nd generation EDTV (Enhanced TV) systems.
Based on studies of channel coding methods, and considering compatibility with the NTSC system, the dicode method is considered to be the most suitable. For the receiving bandwidth of the multiplexed signal in quadrature modulation receivers, the partial response class 4 is considered to be the most advantageous.
A new receiver system with excellent performance is developed that employs the dicode method for transmission channel coding and converts the received signal into partial response class 4 by using 1+D processing.

View full abstract

A moving picture storage and display system is developed to subjectively evaluate super-high-definition (SHD) images, which provide spatial resolution equivalent to 35-mm film and temporal resolution without flicker. This system is designed to display full-color motion images using 2048 by 2048 pixels at 60 frames per second progressive for 4 seconds. It utilizes about 8000 4-M DRAM chips for image memory, a high-speed data transfer bus, about 1 Gbytes per second, and a D/A converter unit, about 360-M samples per second. Sufficient video quality is confirmed by evaluating the output signals.

View full abstract

[in Japanese]

View full abstract