A method of multi-view vehicle license plates location based on rectangle features

In this paper, a method of image processing for Multi-View Vehicle License Plates Location, which is able to process images rapidly and achieve high recognition rates even when images are fairly complex, in poor illumination, and noise, is discussed. This method includes two steps. The first step is to locate the regions of license plates; the second is a transform for each valid area to form an upright image. This paper is distinguished by one key contribution, which is used to find the contour of each valid area including four vertexes, four lines and several characters by using simple rectangle features selected by the AdaBoost algorithm. The other contribution of this work is that a transform based on nonlinear distortion model that is applied to each located license plate to form an upright image convenient for recognition.