Human and carried baggage detection & classification based on RSD-HOG in video frame

In the twenty-first century, crimes are destructively increased in public spaces. Besides, oblivious human movement sometimes dug a death trap for them. These issues are gradually dragging the attention of the computer vision researchers' to automatic video surveillance and warning system (AVSWS). As a fresh branch of AVSWS, human and carried baggage detection and classification has a wide area of applications that are, monitoring suspicious movement of human in public places, warn human in human restricted areas like in atomic power station, detect carriage of illegal materials like weapons, gold into baggage in airport and detect baggage in baggage restricted super shop. However, in this paper an accurate detection & classification framework of human and carried baggage is proposed. Initially, background subtraction is performed to speed up the system and use HSI model to cope up with different illumination condition. Then, rotational signal descriptor (RSD-HOG) is extracted which make the detection accurate. Finally, dynamic human body parameter setting enables the system to detect & classify single or multiple baggage even if some portions of human are occluded or disappear from window. The experimental results discover the system has satisfactory accuracy and faster comparative to others.