【摘要】 在车牌定位上,针对当前实际中存在的一些困难,结合先验知识,提出了一种Hough变换多线检测的由粗到精的定位算法。在字符分割中,通过合理改进,使投影法的准确性进一步得到提升。并在大量实验的基础上,将BP神经网络反复修正,使得系统理论识别率达到了91.00%,结果表明此方法是行之有效的。更多还原

【Abstract】 In the locating process,a multi-line defection based on Hough transformation,from sketchy to precision algorithm combined with prior knowledge is recommended.In the segmentation process,the accuracy of projection can be raised up further by rational improvement.Lastly,by taking a large amount of experiments to improve the BP neural network,recognition accuracy of the system theoretically achieves 91.00%,the results show that it is an effective method.更多还原