Design And Implementation Of License Plate Recognition System Based On Skeletonization Of The HLS Window

License plate recognition system is an important research in the field of the Imge recognition. It is mainly divided into three parts including license plate location, character segmentation and character recognition. The research about the license plate recognition system is necessary for the reasons that the portion of the license area is low in the complete Imge which results in the uncertainty about color, size and position of the license, that it is affected by some factors such as the light, weather, environment and complex background and that it needs to be accurate and real-time. At present there are several types of technology about the extraction of the license generally. For example, under the premise of the Imge pretreatment including gray proportion and binarization we firstly make the erosion and dilation operations and segmenting the character by the means of projection,then the groups of character which fit the correspond formation can be found,and that is the license. The mathematical morphology rule based on gray-scale Imge stands for the features of license shape and the arrangement formation of character. And after the arithmetic of mathematical morphology for gray-scale Imge that was already treated, we can get a character Imge which is formed by combining the character in the rectangular area. Based on the Segmentation of enhancing fractal character string we can use a linear filter to emphasize the texture of license, and segments the license area by splitting the threshO scheme, then computes the difference number of slots that contain both feature-based Imge and the Imge in the sample storehouse to recognize the character Imge.The above methOs are not made in the color space. So some color information is not displayed. For example, the number of iterations is large, which affects the accuracy and efficiency. This paper that is based on the pretreatment of de-noising about the monitor Imge and the conversion of color space of HLS uses the K-means clustering of multi-vector mOel to locate the license area, and constructs the variable window to connect domain of linear dividend barrier tracking to implement character segmention, then refines target area by using the skeleton. Finally the license plate recognition is implanted.