Accurate estimation of noise and signal power is of fundamental interest in a wide variety of vision applications as it is critical to thresholding and decision processes. This paper proposes two methods for the estimation of nonstationary noise based upon models of image structure which locally separate signal from noise. The resulting algorithms are noniterative and thereby fast. The accuracy of the proposed and existing methods is compared, first separately and then in application to two common image processing tasks: edge and corner detection. It is demonstrated that the proposed model can be used to improve the stability of both, in the presence of contrast change and nonstationary noise.

We have developed a visual sensing system that monitors respiration for 24 hours non-restrictively. The principle of the system is inter-image subtraction. Compared with conventional methods, the system has advantages which make it suitable for a long-term monitoring: (1) patients experience no physiological burden during sleep, (2) no manual operation is necessary since it has self-optimizing functions for image processing and (3) it judges the patient's sleep condition adequately using statistical analysis. The system thus can, be applied to predict abnormality of health conditions, and thus for example, alert care providers so that action can be taken with a view to preventing the sudden death of elderly people or, in case of children, to prevent occurrence of sudden infant death syndrome. It can be also applied to the detect clinically dangerous sleep apnea syndrome. Experimental results in a nursing home for the aged have shown the efficiency of the system.

We developed a robot system which can play one-on-one volleyball games with a human player. It can not only hit a flying ball back to the player correctly but also have various autonomous functionalities to interact with humans necessary for playing games, i.e., it can pick up a ball of a requested color by recognizing voice instructions, can locate the players by recognizing their faces, and can shake hands. This robot performed the entire game sequence over 100 times at a company-sponsored exhibition, where the ball rally continued over 25 times with an arbitrary player. In this paper, we mainly describe the real-time visual feedback system of the robot which is essential for playing ball games.

Reports a system for detecting human like moving objects in time-varying images. The authors show how it is possible to detect the image trajectories of people moving in ordinary indoor scenes. The system consists of three subprocesses: changing region detection, moving object tracking and movement interpretation. The processes are executed in parallel so hat each one can recover from the others' errors. This ensures the reliable detection of the trajectories in difficult cases such as movement across complicated backgrounds. The authors have built a trial detection system using a parallel image processing system. The details of the trial system and experimental results of walking person detection are described.>