一种新的车牌字符分割方法

A New Method for Character Segmentation and Noise Elimination of License Plate

在车牌识别系统中,字符分割的效果起着至关重要的作用。传统的方法对车牌图像的要求太高,这在一定程度上影响了车牌的识别率。针对字符断裂和背景噪声非常复杂情况下的车牌,利用小波变换是信号局部特征奇异性检测的有效方法,提出了三级分割算法,该方法能相当准确地提取出车牌中的字符,同时,在分割过程中,各字符间及字符内的细微噪声也得到了很好的去除。实验结果表明该方法能行之有效地提取车牌字符,准确率高,抗干扰能力强,在实际应用中取得了很好的效果。

Character segmentation plays a great role in the License Plate Recognition system. Traditional methods always demand much from the image quality and thus will inevitably affect the ultimate recognition rate. On the condition of the license plate with the parting of characters and very complex background yawp, the wavelet transform is used as a valid method for detecting the local character singularity of signal. At the same time, the presented method extracts characters from the image with the three - level - segmentation method. The noise between and inside characters will be removed simultaneously. The experiment shows that this method can segment characters precisely and it performs excellently in the application.