摘要： 提取车牌字符的Haar特征作为输入，通过级联的神经网络分类器对其进行识别，以定位车牌字符的位置，并根据所得车牌字符位置的相对关系确定车牌位置。该方法不需要进行倾斜矫正，减少了车牌定位后进行字符分割时的工作量。实验结果表明，该方法能准确快速地定位并分割车牌。

关键词: 车牌定位, 字符识别, 特征提取, 倾斜校正

Abstract: This paper extracts the Haar features of license plate characters as input and identifies the input by a cascaded neural network classifier to locate the position of license plate characters. It locates the license plate position according to the relative relationship of the obtained license plate characters. This method does not require tilt correction and reduces the work of character segmentation after location of license plate. Experimental results show that the method can locate and divide the plate quickly and accurately.

Key words: location of license plate, character recognition, feature extraction, tilt correction

中图分类号: 

• TP391.41