A traffic incident detection method based on the steady-state differences of adjacent videos

摘要

At present the method of detecting traffic affairs through video sequences is epidemic at home and abroad in intelligent transport systems, though improving the accuracy and reducing false alarm rate are the focuses of today's research. This paper proposes an incident detection method based on the steady-state differences of adjacent videos. The method first acquires data on basic features of the video image such as gray value, texture value and variance ration, which are gained through analysis and calculation of the grayscale video sequences by using the digital video and image processing technology. And then by analyzing the corresponding stationary features extracted from the obtained data, it realizes the detection of traffic incidents. Simulation results show that this method is very sensitive with low false alarm rate and a broad application prospection.