A License Plate Locating Algorithm based on Multiple Gauss Filters and Morphology Mathematics

In this paper, we propose a new automatic located algorithm of license plate based on multiple Gauss filters and morphology mathematics, there are two steps in our algorithm, first, to locate the license plate coarsely from a car image based on multiple Gauss filters, it contains three steps, firstly, to compute the vertical difference of the standardized image and then to get the horizontal projective curve of the difference image, secondly, to filter the curve by a window gauss filter and then to transform the smooth horizontal projective curve by another Gauss function, thirdly, to locate the license plate coarsely by seeking the vales of the transformed curve. Second step, to extract the license plate accurately based on morphology mathematics from the coarse license plate, it contain three steps, firstly, to compute the vertical difference of the coarse license plate and then to process the difference image by morphology operation, secondly, to get the vertical projective curve of the difference image and then to smooth the curve by window gauss filter, thirdly, the accurate position of the license plate can be got by seeking the wave of maximal area of the curve. Finally, the accurate license plate can be got after these steps, the experimental results have proved that this is an effective, robust and fast method to locate the license plate from a car image captured from different backgrounds.