| The difficult point of license plate research is the method of license plate localizaition and the charater recognition. For the moment, there are many ways to realize the license plate recognization, but there is no way to apply to all the application environment.In this artical, we have used the technology of digital image processing and pattern recognition. The license plate recognition system is composed of four parts:Image pretreatment, license plate localization, character segmentation,license plate recognition.Image pretreatment part mainly used gray level histogram equalization smoothing fliter, binarization image processing. In this part, we have a research on the method of improving image quality of license plate, then we used MATLAB software to make a comparative analysis of the actual effect of license plate image after image pretreatment.License plate localization part mainly used the integrated method of edge detection and mathematical morphology, we studied the method of specifying and segmenting license plate area, then we used MATLAB software to realize the localization of license plate.Character segmentation part mainly used Radon transform to realize the incline rectification of license plate and character segmentation method based on horizontal vertical projection, then we used MATLAB software to realize the incline rectification and character segment.License plate recognition part, we have a research on the method of using BP artificial neural network to realize character recognition, then we used MATLAB software to build a neural network networks, and realized the license plate character recognition.Based on Zigbee and GPRS wireless communication technology, we achieved the results of the license plate recognition of remote communication in the parking lot. In this artical, we used MATLAB software to do the experiment. Proved by experiment, the effect of license plate recognition is better, and the information of recognition realized a reliable transmission. |
PDF Full Text Request