Statistical Pattern Based Real-Time Smoke Detection Using DWT Energy

Abstract

This paper proposes a novel method to detect smoke using statistical patterns which are DWT energy. In general, shape of smoke is not clear and color and diffusion direction of smoke depends on the environment. Therefore, if small pieces of smoke's information are used, false detection rate is increased. In this paper, the foreground is detected by robust method to environment changes. After its detection, DWT energy, shape, and color information of objects in the foreground are used to determine the smoke. The proposed method is suitable for the early detection. The proposed method takes the average processing time of 30 fps and approximately 7 seconds at the detection smoke from the moment the initial fire. False detection rate for the proposed method is lower than that for the previous method.