Accurate assessment of the wall motion in left ventricular is a prerequisite for disgnosing the coronary artery disease. By means of the combination of computerized image processing technique and stress inducing skill, the quantitative analysis of that motion can be performed, and the disease can also be predicted at its initial stage. In this study, we take and store 16 frames of image at one time. They can be replayed at any speed selected freely by the user, so that the doctors can observe the details of the heart contraction. With the temporal smoothing and spatial filting algorithms, the quality of image can be improved. An edge tracking algorithm has been used to calculate the regional area. Finally, a diagnosis report is submitted automatically. The results show that this method is a sensitive and accurate technique for diagnosing the coronary artery disease early.

View full abstract

An image database is projected collecting textile pictures to aid designers from a graphical aspect. Classification of images uses eight scales ranging each from -3 to +3 based on eight pairs of adjectives. Our reserch about the classification was carried out by questionnaires of some twenty-six people against about fifty pictures. Our reserch is to classify thousands of textile pictures, automatically. We propose graphical features which can represent first impressions of given pictures. The graphical features give the scales corresponding a pair of adjectives, warm and cool.

View full abstract

In a highly developed information society, personal identification for security measures has been highly demanded in a wide variety of usage. Authors propose to use auricles for the personal identification using human body characteristics in the next generation. In this paper we describe a method of feature extraction by the morphological operations relative to the square structuring elements for the purpose to roughly obtain shape feature of the auricle. This method leads to no errors of spectrum extraction and can extract feature points of the auricle based on the spectra. We also describe the examination of changes of the identical person's images with time under constant conditions of lighting and camera parameters.

View full abstract

The current research of image processing in Tokyo Inst. of Tech. is described. we introduce trend researches in Ogawa Lab., Sakai・Yoshida Lab. and Nakajima・kamei Lab. Our interests are image analysis, image recognition, image generation and soon with treating images mainly. In this paper, the researches on pattern recognition, neural networks, hyper-media network, computer graphics, virtual reality and so on are described.

View full abstract