Tracking and localisation of moving vehicle license plate via signature analysis

In this paper, a new algorithm for dynamic vehicle license plate localisation is proposed on the basis of signature analysis and connected component analysis. Most existing techniques use low level features such as detection of colour and edges, these methods may have issues with reliability. The proposed algorithm however, aims to manipulate more accurate perceptual motion information. The camera is set to record the moving vehicle, while the angle of view and the distance from the moving vehicle is changed according to the observational setup. Image differencing is used to detect the motion, and noise is filtered out to obtain a binary image that can stand out from the motion of the moving vehicle. The algorithm was tested using moving vehicles with different backgrounds, under illumination change and varying poses and angle.