The license plate recognition (LPR), as an important measure of traffic controlling and information management, is being paid more and more attention. The location and detection of the inclined license plate is an important problem in the License Plate Recognition Sys-tem. But now most of the proposed location algorithms are based on the condition of the horizontal license. These methods don't work well on inclined license plate. At the same time, the method based on Hough Transform is pro-posed to detect the inclined angle of the license plate, but it has many limitations. In order to solve this problem,we propose a new method to locate the inclined plate region using Mathematical Morphology and Orientation Field. The algorithm consists of preprocessing, inclined an-gel detecting and license plate location modules. Firstly,we adopt the gray extension method to improve the ob-serving quality of the images and study the intrinsic char-acteristic of texture and shape of vehicle license, such as the size, space and height of characters, then we construct the morphologic operator to make a series of morphologicoperations and locate the plate region roughly in the pre-processed image; secondly we make the horizontal and ver-tical projection, get the coordinate of the down left corner and top right corner in the license plate, and detect the inclined angle of the license plate with Orientation Field;lastly, the results of projecting and Orientation Field cal-culating are used to locate the license plate precisely. We practise our method on images under different illumination condition, and the experimental results show that the accu-racy of inclined angel detection and license plate extraction based on Mathematical Morphology and Orientation Field is significant.
展开▼
