中文摘要：

运动车辆图像中车牌具有所占比例小、位置不固定和大小不一的特点,因此,对车辆图像分割时车牌区域容易产生过分割与欠分割问题.脉冲耦合神经网络（Pulse Coupled Neural Network,PCNN）被誉为＂第三代神经网络＂并广泛应用于图像分割.在利用PCNN模拟人类视觉的图像分割过程中,由于传统PCNN模型中的连接矩阵使用固定值表示,使得PCNN模型不能满足图像分割时尺度变化的需求.为了解决这个问题,本文提出了基于多尺度空间PCNN模型的车辆图像分割算法,将尺度空间引入PCNN模型,使PCNN模型具有了尺度特性,提高了系统自适应分割车牌图像的能力.

英文摘要：

The license plate of the moving vehicle image has some characteristics such as small proportion,random positions and different sizes.The plate regions are easily in a state of under-segmentation and over-segmentation when we segment the vehicle images.Pulse coupled neural network（PCNN） is known as the third generation neural network and is widely used in image segmentation.In the process of image segmentation which uses pulse coupled neural networks to simulate human vision,traditional PCNN model can＇t meet the scale change needs for image segmentation because of the fixed values in the connection weight matrix.In order to solve this problem,the method of image segmentation based on pulse coupled neural networks of multi-scale space was proposed.The scale space was introduced into traditional PCNN model to make the model possess the scale characteristics and improve the system＇s ability to segment license plate image adaptively.