Research On Crts? Track Slab Crack Based On Machine Vision Detection Technology

With the rapid development of Chinese economy,China has increased the construction of infrastructure.Among them,high-speed railways,as an important part of transportation,have ushered in opportunities for development.Track safety,as one of the necessary conditions for the normal operation of high-speed railways,involves the basic interests of the broad masses of the people,of which cracks in track slabs are a major hidden danger.At present,CRTS? type slab track is the longest ballastless track laid in Chinese high-speed railways.The problem of crack detection safety of this type of track slab has become a hot topic.Due to the problems of low image contrast and difficult image acquisition in the low-light environment at night,the detection of cracks on the track slab has become a difficult point,and when performing crack detection on the track slab,cross cracks and small cracks are also one of the difficult to overcome.In view of the above problems,based on a comprehensive analysis of domestic and foreign research and development,this thesis has fully studied the CRTS? type track slab crack detection technology based on machine vision.The main contents are as follows:(1)To solve the problem that the YOLO network has a poor detection effect on micro cracks.This thesis replaces the fully connected layer in the original network with a 1 * 1 convolution layer,and superimposes the low-level features extracted from the shallow convolutional network in the YOLO network and the abstract semantic features extracted from the deep convolutional network to improve Detection effect of tiny cracks.(2)To solve the problem that the YOLO network has a poor detection effect on cross cracks.This thesis adds a fur frame mechanism on the basis of the original network and optimizes it to form six kinds of fur frames with the same area difference,which improves the expressive power of the network and enhances the detection effect of cross cracks.(3)To verify the effectiveness of CRTS? type track slab cracks.This thesis develops a crack detection system based on CRTS? type track slab.This system can better combine the improved algorithm,user interface and image processing of this thesis,has good integration and scalability,and applies it to actual The track slab crack detection environment has achieved good results.