Research and implementation of an improved license plate recognition algorithm

For the purpose of recognizing the License Plate accurately and rapidly, the algorithm makes a full use of texture, shape and projection characteristics of the License Plate and expounds the License Plate Recognition in four parts under the condition of consideration and solution of adverse factors such as uneven illumination, license plate deflection and noise pollution during the image acquisition, including license plate location, deflection correction, character segmentation and character recognition. BP neural network recognition method and many image processing methods are involved, such as gray level transformation, Sobel horizontal edge detection, Otsu adaptive threshold segmentation, first-order horizontal difference and mathematical morphology. Total algorithm is realized in Visual C++ 2008 and OpenCV environment, and this recognition algorithm is experimentally satisfying in both accuracy rate and reliability.