A Novel Median and Switching Based Method Used to Remove the Impulse Noise from the Traffic Video

Along with the traffic video widely used by the intelligent traffic system to monitor the vehicles, their information (such as the license plate), the vehicle type, and so on, is successfully acquired. However, if the traffic video has impulse noise, none of the above systems can get the satisfied results due to the noisy sensor or channel transmission errors. Therefore, it is necessary to suppress the impulse noise. Although the classical median based method has been widely used to suppress the impulse noise in recent years, the noise suppression and the detail preservation result cannot be satisfied due to the noise filtering on each pixel in the whole image, no matter whether or not the image pixel is corrupted by the impulse noise. This paper proposes a novel median and switching based method to detect the impulse noise, with the detected impulse noise replaced as the evaluated value and the detail pixel protected as the original value. Experimental results on the testing image and the traffic video image show that the proposed method can work better than other methods in both the quantitative and qualitative aspects.