Accident analysis of a surveillance camera system through a frequency-based processing

The surveillance camera systems are commonly used in detecting the traffic violation on the roads. A key requirement in most detection systems is the ability for rapid automated analysis to identify the accidents. This paper describes an accident analysis based on frequency information for detecting the accident, identifying the cause of the accident and storing the accident scene. When an accident happens, the surveillance system analyzes the accident sound and stores the accident scene in real-time. The proposed system utilizes the sound magnitude (dB) and frequency analysis (FFT) in order to increase the accuracy of accident assessments. In addition, this system accomplishes a continuous buffering without the presence of an accident. It stores the prior accident scene, accident scene and post accident scene. When this system is applied to the actual crossroads, it can detect the accident presence and save the inspection expense. Results of applying the proposed technique to data obtained from the detection of accidents are presented and compared.