| Vehicle license plate recognition is an important part of the intelligent transport system, and it plays an important role in traffic monitoring and control. Its main techniques include: license plate location, character segmentation and character recognition. Among them, fast and accurate segmentation of the license plate area is the prerequisite for license plate recognition system and the key to improve the recognition rate.Plate region has an unique texture and color characteristics, analyze it using mathematical methods is the basis for a variety of license plate location algorithm. Wavelet analysis has good features and multi-resolution time-frequency analysis of the characteristics, therefore, using wavelet transform analysis of texture features for image segmentation has become a hotspot of current research and important research direction.In this paper, we discussed the basic theory of image segmentation and registration methods of the classic positioning, given the three traditional methods of experimental results and analysis. In the study of basic theory of wavelet analysis, and texture features extracting method, we proposed an algorithm of wavelet texture features and color combination determined by neural network in recognition of license plate location. The algorithm first make two-dimensional wavelet decomposition of the image, set an sliding window scanning high-frequency sub-band, calculated region wavelet texture feature, while in the corresponding low-frequency graph of the sliding window, convert RGB color component for the HSV color values, extract color features, then using BP neural network algorithm for determining identification and location of the license plate. The method is an integrated feature location, overcomes the shortcomings of the single feature location and limitations, from the results it can be seen that the algorithm can be more accurately extracted from the background image. |
PDF Full Text Request