提出了一种复杂背景下的多车牌图像分割和识别方法，首先采用统计和特征匹配相结合的方法进行背景提取，将可能存在车辆的区域提取出来；然后分别对可能的车辆区域进行局部边缘检测，并使用车牌的先验知识确定车牌的位置和单个字符分割，包括车牌倾斜时的字符分割；最后使用PCA和神经网络相结合的方法精确识别车牌。-Proposed a multi-plate image segmentation and recognition method under a complex background, the first use of statistics and feature matching method of combining background extraction, there may be extracted from the area of the vehicle　then were on the vehicle may be partial edge detection area and use prior knowledge to determine the location and the individual license plate character segmentation, including the license plate character segmentation when tilted　Finally PCA and neural network using a combination of methods to accurately identify the license plate.