MOVING OBJECT DETECTION BASED ON PASSIVE RADAR FOR SMART CITY by Chen Xin Thesis Supervisor: Prof. Kam-Weng Tam Electrical and Computer Engineering In this thesis, we propose a new methodology to realize a moving object detection system from TV signals based on passive radar which is applied to RFID Intelligent Cane. This intelligent cane is designed for Visually Impaired People (VIP) which can be promoted as a tool in Smart City Transportation. The innovation of our detection system is about the system architecture integration of the 920MHz UHF RFID, 700MHz band analogue TV based passive radar, image comparison and machine learning algorithms together to realize the practical moving object detection system with high accuracy. The UHF RFID attached intelligent cane works as a controller for the Visually Impaired People (VIP). Received Signal Strength Indication (RSSI) from the attached RFID tag is used for control signal generation. While the gesture of the intelligent cane changes from horizontal to vertical, the cane is detected from ‘ON’ to ‘OFF’. This thesis utilizes TV signal propagation based on passive radar to capture the detection data. Due to the TV waveform of the transmitted signal is recovered in an ideal receiver with correct frequency response, the fluctuation of propagation paths introduced by moving objects can always be perceived on a domestic television. Hence, the effect of moving objects can always be seen on TV. Such TV Receiver can be easily integrated into the cane in future. Image comparison and detection methods are also applied to complete the detection system. By using the recorded videos based on passive radar, the image comparison is applied to calculate the difference between the reference and test group. Then, the moving object detection methods are applied to further complete the detection system. At last, machine learning algorithms, such as Logistic Regression, Support Vector Machine, are used to train the moving object detection system and gain the accuracy of the overall moving object detection system.