Every year in summer and autumn, there are usually several typhoons pass through Taiwan. They often bring heavy rains and sometimes cause serious debris flows in the mountainous regions. It is very important to detect the debris flows in order to reduce their damages. To setup the fixed cameras to monitor the mountainous areas is an efficient and economical approach. An early warning can be announced if debris flow is detected. In this paper, we adopt Fisher's Linear Discriminant Analysis (FLDA) and the background subtraction to detect changes in video frames, and then calculate entropy and apply optimal theory to detect the debris flow. Finally, a comparison of detection result is conducted for performance analysis.