Design and Implementation of a Human Tracking CCTV System Using IP-Cameras

Closed Circuit Television (CCTV) technology made a big impact on how crimes were solved. CCTV footages were used as a material to review crime scenes and were used to identify culprits who were then placed as one of those "wanted" persons.CCTV systems only provide footages and lack the ability to analyze these footages. In this study, an IP camera based CCTV system with the ability to detect, recognize and track a person of interest was proposed. Initial set-up used five IP cameras to capture the front and side angles of the person of interest. These were also used to identify the direction of heading. The Haar feature-based cascade classifier was used for face detection. The Karhunen-Loeve transform was used for face recognition. And optical flow was used for tracking which was implemented in Processing.Gender, eye and face recognition was performed and results show that the system can detect and recognize the person of interest with more than 81.0 % accuracy. On the average, gender can be classified with 83% accuracy particularly when the person of interest is a female. Higher accuracies were obtained when the person of interest is wearing eye glasses. Consistently, face recognition performs better when the person of interest is a female wearing eye glasses.