Dither methods are generally used for representing multi-level images to a bi-level display, and various techniques of dither methods have been proposed. Especially, the minimized average error method is known as one of the excellent methods in the quality of dithered image. However, in this method, it is difficult to get a high compression performance. In this paper, we propose a data compression method using a concept of the Schalkwijk code for the image generated by the minimized average error method. First of all, the dithered image is divided into some blocks in this method. Secondly, the Schalkwijk code is applied to each divided block. We report simulation results of compression ratio.

View full abstract

In this paper, we proposed the method which can estimate the relative position between leading vehicle and following vehicle based on the movie acquired by following vehicle's Drive Recorder. This method is designed to analyze several traffic accidents - especially fender bender - in automatically. At the first step, our method estimates the depression angle of drive recorder's camera based on feature points tracking and Eight-Points Algorithm. At the second step, the region of the leading vehicle is given by operator at selected frame. Our method then tracks given leading vehicle image in whole of the video image sequence based on particle filter algorithm. Finally, the distance between leading vehicle and following vehicle is estimated based on depression angle and image tracking result.

View full abstract

This paper presents the system providing experience of hands of different sizes composed of optical systems and analogous objects. The hand size is recognized to be different from one's own when presented a hand of a different size in an environment that seems changeless in size by adjusting the optical zoom and actual object size, but not the hand size, proportionately. For successful manipulation of objects by hand, we need to feel “a sense of agency” and “sense of ownership” adequately. As these senses are affected by visual delay, the effect of visual delay on the experience was examined by subjective evaluations. The answers to a questionnaire about hand size, sense of agency and sense of ownership were measured after a button-pushing task. The experimental results showed that the allowable visual delay is 100ms for experiencing hands of different sizes.

View full abstract

This paper proposes a method to detect candidate pixels that compose the contour of an uneven area defect on LCD display. In the current LCD manufacturing line, the visual inspection by skillful inspectors is required to detect an uneven area defect because its location, size and status are unpredictable in advance. In the proposed method, the influence by moiré pattern is reduced by a simple method using focused image and unfocused image at first. Next, the special filter that can change its direction and size according to a surrounding image pattern is applied. After that, candidate pixels for contour of uneven area defect are detected through the dynamic threshold technique by local blocks. The experimental results showed that the major parts of contours of uneven area defects were detected by the proposed method.

View full abstract

In this paper, we propose a new method for automatic classification of spinal deformity from moiré topographic images by using four asymmetric features on right- and left-hand side of human back. Spinal deformity is one of serious disease and it is mainly suffer by teenagers during their growth stage particularly fifth year in the elementary school to second year in the middle school. In the mass school screening, a moiré method has been proposed which takes moiré topographic images of human subject backs and checks its symmetry/asymmetry of the moiré patterns in a two-dimensional way. We perform the proposed method to 1200 real moiré images in the classification employing Artificial Neural Networks based on back propagation and Support Vector Machines. As a result, on the total average, classification rate of 90.3%, and 85.3% were achieved in the ANN, SVM, respectively.

View full abstract

In this paper, a novel image coding scheme which uses digital watermarking techniques are proposed. In the proposed scheme a part of image coding data is embedded into the other parts of own image coding data instead of the digital watermarking information like copyrights. By way of example, a fractal image coding method adopting the proposed scheme is presented. The results of the fundamental experiments show that the fractal image coding using the proposed scheme has a good chance of realization. Also it is presented the proposed scheme may be available for other coding methods.

View full abstract

As a process of projecting 3D holographic images in the space, a spatially projecting process employed misty screen is well-known. In this process, it seems to be very important to suppress fluctuations of represented images as much as possible. In this report, taking much attention to this process, we show that in order to make a misty flow to be stable and to make the size of the system so small, it is very effective to adopt an attachment structure in which nozzles are easily put on and off the system, and that a nice image representation can be really carried out by such a system. In addition, studying geometrical structures of nozzles, we found that it is effective to adopt a two dimensional array of small-sized nozzles.

View full abstract

[in Japanese]

View full abstract

[in Japanese]

View full abstract