一种基于模糊聚类的车牌图像分割方法

车牌图像分割是车牌特征提取和识别的前提,其分割效果直接影响到车牌识别的准确率。针对运动中的车辆牌照受天气等环境的影响,导致车牌图像存在不确定性和模糊性的特点,提出一种基于C均值模糊聚类的车牌图像分割方法,避免图像分割时的阈值设定问题。利用采集的原始车牌图像,在MATLAB7.0环境下进行分割仿真实验,实验结果表明,算法可以有效地去除干扰信息,突出车牌区域的图像特征,为后续的车牌定位、字符分割和字符识别提供重要保证。

An image segmentation method of license plate based on fuzzy clustering algorithm

The license plate image segmentation is the premise of feature extractions and recognitions, which directly influences the accuracy of the recognition. A license plate image has the feature of uncertainty and fuzziness because of the various weather surrounding, so a recognition method based on fuzzy C-means clustering is put forward, which avoids setting any thresholds. With MATLABT. 0, the experiment shows that this method can remove the noise messages and highlight region image features, which provide the guarantee for further procedure of plate location, character segmentation and character recognition.