การรู้จำการทำกิจกรรมผู้สูงอายุจากการวิเคราะห์ภาพ

Abstract

The trend number of elderly in Thailand is increasing and most elderly are desire to live on their own homes. The development in modern technologies assists older people to live alone and make their life more comfortable. In this work, the focus is to analyze and monitor abnormal activities in four basic rooms by observing 12 different activities of 10 different persons. The comparison of characteristics is found different from the normal pattern. The changing pattern of activities in daily life is importance to refer the unhealthy or abnormal behaviour. This feature is used to analyze both normal and abnormal activity. A method has been proposed in this study which uses Transition Matrix Model (TMM) and Root mean square error (RMSE) to observe the 12 basic activities in four rooms for discovering normal pattern and use for monitoring abnormal activities. Ten different sequence activities in daily life has been analyzed by using this method. A sequence of activity in daily life has been defined as the normal activity. The data were compared by proposed method with the same sequence of activity by adding different levels of noise such as 10, 25, 35 and 50 %. Less noise in the room shows 100% accuracy while adding more noise will result in decrement of accuracy.