In this paper, an intelligent intrusion detection system, based on UPnP technology for smart home is proposed. In the developed system, home users equipped with PDAs may locally and remotely receive the alert of home intrusion with captured subject image. There are two key components developed in this system. One is the intrusion image processing (IIP) to recognize the intrusion and extract the intrusion subject. The other is the universal-plug-and-play (UPnP-based) instant messaging (UIM) The proposed IIP employs a modified three-step search (TSS) to improve the accuracy of the subject identification. Through the integrating of the two software component, intelligent intrusion detection is capable of offering a smart home security system. The implementation results show the accuracy of subject identification is about 90%~100%, depending on the granulity of the subject searching. The presented system demonstrates a smart and automatic solution in comparison with legacy approaches.