科研之友
微信
新浪微博
Facebook
分享链接
摘要
Fault inspection plays an important role in ensuring the safe operation of freight cars. With the development of computer vision technology, vision-based fault inspection has become one of the principal means of fault inspection. A coupler yoke is an important component of the train's connection system, and if the bolt goes missing, it would cause the separation of the train from the coupler, resulting in a serious accident. In this paper, we propose an automatic image inspection system to inspect the faults in coupler yokes during the operation of a freight train. Images of the coupler are acquired and the inspection process is divided into two parts: the localization part and the recognition part. In the localization part, we combine the normalized gradient magnitude with the histogram of gradients on six orientations to form the Multiple Dimension Features, design a fast algorithm to compute the multi-resolution image features, and use a linear support vector machine to locate the position of the coupler yoke in an image. In the recognition part, we use Haar features to study the appearances of coupler yokes, propose a fast decision trees training method by pre-pruning non-effective features, and use AdaBoost decision trees as the final classifier to determine whether there is a fault. The two main parts above combine to create the whole inspection system, and experimental results show that this proposed method can achieve a fault inspection rate of 98.6% while the average processing time of an image is about 98 ms, which shows that our system has high inspection accuracy and a good real-time performance.
- 出版日期2018-2
- 单位北京航空航天大学
