| Summary: | 碩士 === 國立中央大學 === 資訊工程研究所 === 98 === Due to the fast development of computer and video technologies and the cost-down of capturing devices, surveillance systems are widely applied in our daily life. People use the home security system to protect their family and property. However, most security systems will send alarm messages to users when the sensors were triggered, but cannot identify what the intruder is. If the security systems often make the false alarm caused by animals, people may relax their vigilance as time pass. In order to solve those common problems of traditional home security systems, we combine the image recognition, motion detection, image processing, and Neural Network to build a system that can identify whether the illegal intruder is human or other animal. First, we construct the background map to obtain the foreground by using Gaussian mixture model (GMM). Then, we combine shape-based detection and Back-propagation Neural Network to determine whether the foreground is human or not. If the foreground is human, the system will issue an alarm. Some experimental results were demonstrated to verify the performance of the proposed method.
|
|---|
