Design And Implementation Of Highway Guardrail And Crack Detection System Under Complex Background

With the development and popularization of automobiles,the flow of road traffic in my country is increasing day by day.The road pressure caused by the increasing traffic flow will have a counterproductive effect on the road surface,guardrails,and other facilities along the route.It can damage the road infrastructure,and affect the safety of road traffic even.The emergence and development of the highway inspection industry provide important guarantees for urban traffic safety management,especially for highway sections with heavy traffic and frequent accidents.The traditional highway inspection takes the highway corrugated beam guardrail and pavement cracks as the main objects and adopts the form of manual inspection.Aiming at the problems of long inspection cycle,low efficiency,and untimely feedback of traditional methods,this paper is based on deep learning technology,starting from the two angles of highway guardrail detection and road crack detection,and completes the following research:First of all,to solve the problem of poor real-time performance and low accuracy in the detection of corrugated beam guardrails on expressways,a rapid detection algorithm for corrugated beam guardrails based on instance segmentation is proposed and implemented.The multi-scale adaptive feature fusion algorithm is used to solve the problem of information loss caused by the dimensionality reduction of the feature pyramid network,and the extreme value matrix iterative screening algorithm based on the Gaussian weakening mechanism is used to solve the excessive suppression problem of the Fast NMS and improve the accuracy of the algorithm.Experimental results show that this algorithm shows good performance in both detection accuracy and detection speed.Secondly,aiming at the problems of poor robustness,low detection precision,and serious information loss in the crack detection algorithm of highway asphalt pavement,a semantic segmentation-based highway asphalt pavement cracks detection network is constructed.The proposed lightweight residual learning algorithm based on Res Net152 is used to extract multi-scale features,and the network is simplified on the basis of ensuring accuracy;the proposed spatial domain-based self-attention mechanism algorithm is used to solve the problem of global information loss caused by convolution operations;The proposed asymmetric bilateral segmentation structure is used to build a network model for pavement crack detection,and features at different levels are combined.Experiments prove that this algorithm shows good performance for both the detection accuracy and detection speed in a complex environment.Finally,from the perspective of practical application,a highway guardrail and crack detection system is built.Rely on the guardrail detection algorithm based on instance segmentation and the crack detection algorithm based on semantic segmentation,the system designs and implements the corrugated beam guardrail detection module,the road crack detection module,the system management module,and the log collection module.On this basis,the system is tested in two aspects:function modules and system performance.The test results show that the various functional modules implemented by the system can be used normally,and the detection performance of the highway corrugated beam guardrail and the asphalt pavement crack can meet the actual application requirements,thereby realizing the intelligence of the highway inspection task and improving highway inspection efficiency.