摘要
准确提取各种运动目标的特征并将它们加以分类识别,是近年来图像处理和人工智能研究中的热点之一。针对识别运动车辆车型需求,提出了在利用脉冲神经网络模型对运动车辆进行边缘提取的基础上提取运动目标的不变线矩特征,再用这些特征训练神经网络对车型进行识别的方法。试验结果表明该模型能准确的提取运动目标的特征,从而提高分类的效果。在今后的智能监控系统中有广阔的应用前景。
Accurate feature extraction and recognition of moving objects are the hot spots in image processing and artificial intelligence research domains.In order to identify the types of moving vehicles,this paper proposed an identification approach in which edges of the moving vehicle are extracted by a spiking neural network model.The line moment of moving vehicle is used as the features to train a neural network,and then the neural network is used to identify type of the moving vehicle.The results of the simulation show that the approach can accurately extract the features of the moving vehicles so that the accuracy of identification has been improved.This approach shows a promising prospect in the application of intelligent surveillance systems in the future.
出处
《计算机系统应用》
2011年第4期182-185,共4页
Computer Systems & Applications
基金
福建省自然科学基金(2009J05141)
福建省教育厅科技计划(JA09040)
关键词
脉冲神经网络
线矩
特征提取
车型分类
spiking neural networks
line moment
feature extraction
vehicle-type classification
