| 基于神经网络和不变矩的路面破损图像识别 |
| |
| 引用本文： | 庞玉红,李孟歆,孙长龙.基于神经网络和不变矩的路面破损图像识别[J].机电产品开发与创新,2007,20(6):134-136. |
| |
| 作者姓名： | 庞玉红 李孟歆 孙长龙 |
| |
| 作者单位： | 沈阳建筑大学,信息与控制工程学院,辽宁,沈阳,110168;沈阳市政二公司,辽宁,沈阳,110014 |
| |
| 基金项目： | 建设部资助项目
,
辽宁省教育厅资助项目 |
| |
| 摘 要： | 提出了一种应用BP神经网络识别沥青路面破损图像的图像分割方法.将路面图像等分,用灰度方差值描述子块图像特征,利用BP神经网络对子块图像进行模式分类,并将图像子块模式矩阵的不变矩作为图像的整体特征,在此基础上设计了基于全局优化算法的前馈神经网络分类器,并进行了图像识别试验,对二值图像进行特征提取,提出了学习算法,以加快收敛速度,从而实现图像识别.
|
| 关 键 词： | 神经网络 沥青路面 不变矩 图像识别 分类器 |
| 文章编号： | 1002-6673（2007）06-134-03 |
| 收稿时间： | 2007-09-17 |
| 修稿时间： | 2007年9月17日 |
Asphalt Pavement Surface Distress Image Recognition Based on Neural Network and Moment Invariant |
| |
| PANG Yu-Hong,LI Meng-Xin,SUN Chang-Long.Asphalt Pavement Surface Distress Image Recognition Based on Neural Network and Moment Invariant[J].Development & Innovation of Machinery & Electrical Products,2007,20(6):134-136. |
| |
| Authors: | PANG Yu-Hong LI Meng-Xin SUN Chang-Long |
| |
| Abstract: | A method of image feature representation is put forward in the paper, which is based on BP neural network. Pavement surface image is divided into equal subimages, and greyness variances are used to represent subimages feature. The subimages are classified with BP neural network, and moment invariant of matrix is used to represent image feature. A pavement distress forward feed neural network classifier is designed based on global optimization algorithm, and image recognition experiment is done and study algorithm is put forward to accelerate convergence rate. |
| |
| Keywords: | neural network asphalt pavement surface moment invariant image recognition classifier |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
