Theory And Practice Of Bridge Crack Detection Based On Image

Bridges serve as hubs for transportation systems such as roads,highways and railways,and their structural health is closely related to the lives and property safety of the people,regional economic development,and social stability.The current number of bridges in China is huge,as of the end of 2019,there were 878300 road bridges in the country,totaling 60.6346 million meters.However,bridge safety accidents due to various reasons such as design defects,substandard construction quality,natural disasters and overloaded operations are occurring frequently.Cracks are not only a symptom of bridge safety problems,but also seriously affect the structural strength and durability of the bridge.Therefore,crack detection is one of the key contents of bridge safety inspection.Image-based bridge crack detection technology will gradually replace traditional manual detection methods and become the mainstream measurement method due to its advantages of fast detection speed,convenient method and high degree of automation.However,the current hardware systems and software algorithms based on image detection have limitations,which can not meet the needs of actual detection.Based on this,in view of the many shortcomings of traditional manual detection methods,combined with the current research status of existing machine vision measurement systems,digital image processing algorithms and deep learning technologies,this dissertation has completed the following research content:(1)A new set of image measurement methods is proposed.This method designs laser virtual scale equipment and proposes a new measurement theory,which can accurately obtain the scale of the measured image under non-contact conditions,and accurately measure the width,length and area of the crack;(2)The processing of crack images in the above method is further optimized.Design a process image processing algorithm,which can effectively extract the edge contour of bridge cracks in the image under complex background and noise interference;(3)The intelligent theory of the above image method is further developed.Using image slicing and convolutional neural network recognition methods,in high-resolution bridge photos collected in a large field of view,automatically detect the cracked area of the structure.