摘要
介绍了一种基于图像识别技术的航空发动机内窥检测的新方法,采用最大类间方差法计算二值化分割的最佳阈值,并将提取的图像特征输入神经网络进行分层识别,最后由专家系统对损伤程度进行了诊断。通过现场测试,证明了该方法的有效性和实用性。
Introducing a new video-probe-inspection method based on image recognition inside engines of the mechanical damages. With the superior threshold of image binarization by OTSU(maximization of interclass variance) algorithm and a method to extract damage characteristics,the extracted image characteristics is inputted into neural network to identify step by step what kind of damage it is,and an expert system is involved to do a diagnosis and assess the damage degree An automatic classification of inside-vision damages and automatic diagnosis and verification of damages' extent is therefore brought about. Its effectiveness and reliability are testified through an on-the-spot experiment
出处
《中国民航学院学报》
2005年第4期15-19,共5页
Journal of Civil Aviation University of China
基金
中国民用航空总局科技基金项目(03126).
关键词
图像识别
内窥检测
特征提取
BP神经网络
专家系统
image recognition
video probe inspection
characteristic extraction
BP neural network
expert system