摘要： This paper presents an improvement method for enhancing performance of a system for measuring VDT working hours using a webcam. This system is utilized in companies to avoid long VDT work. Assuming large-volume introduction of the system into companies, the system monitors the PC user using images obtained by a common webcam considering cost performance. However, the system can hardly measure the time, i.e., cannot recognize existence of the PC user in the cases when a user wears glasses or masks. This monitoring differs from regular face recognition and it is not easy to recognize existence of a user who works in front of a monitor because users do not always face to the camera. In this research, a method for recognizing glasses and a mask in YCbCr color space has been introduced into the existing system. From experimental results that accuracy ratios for recognizing existence of a user who wears one of the items were more than 80% to 9 of 10 participants and very similar to accuracy in the existing system for users who do not wear them, we confirmed the improvement is effective enough in the system.