Damage Detection System For Highway W-beam Guardrail Based On Machine Vision

Wave beam barrier is one of the important components of highway traffic engineering safety facilities.It is mainly composed of W-Beam steel barrier board and barrier column.In case of traffic accident,the W-Beam barrier will be seriously bent and damaged after being hit.If it cannot be found and repaired in time,it will leave serious hidden dangers to traffic safety.At present,the domestic multi-functional road detection vehicle does not have the automatic detection function of W-Beam barrier damage,but the traditional manual detection method will waste a large labor cost and low detection efficiency.Combined with machine vision detection technology,this paper designs a vision based damage detection system for Highway W-Beam barrier.It can further improve the function of multi-functional road detection vehicle.The main research contents of this paper include:(1)Hardware system design: the hardware system is composed of image acquisition mechanism and image acquisition control mechanism.The design principle of image acquisition mechanism is to play a fixed role and have damping effect at the same time.Prevent the collected image from being unclear due to the road fluctuation and vibration,so as to reduce the complexity of the detection and recognition method of W-Beam barrier.The image acquisition mechanism is installed on the top of the detection vehicle,and the collected image shall include the characteristics of the W-Beam barrier area to the greatest extent.The external trigger module of the camera is composed of Arduino development board and opposed laser sensor.It is installed on both sides of the detection vehicle wheel.Through the relationship between the pulse signal output by the laser sensor and the driving distance of the detection vehicle,the camera is controlled to continuously collect images.All frame images cover the whole W-Beam barrier to avoid missed detection.The industrial camera,trigger device,control system and other related hardware are selected.(2)Software system design: select the color of barrier through human-computer interaction interface,convert the color space in the obtained color image,and design the target area extraction algorithm for barriers with different colors.In order to reduce the operation time of barrier detection algorithm,a target tracking detection algorithm in the barrier area of W-Beam is proposed in this paper.In the detected barrier area,carry out the next step of barrier damage detection.Calculate the slope of the fitted straight line through edge detection,straight line fitting and other operations.If it meets the requirements,it is judged to be normal,otherwise it is damaged.If no barrier area is detected,judge whether to enter the bridge shadow area,movable barrier area and concrete barrier area based on the gray difference of window sampling and image texture features.In the detection and recognition of movable barrier and road natural background,hog algorithm is used to extract the eigenvalues of the detection object,which has strong adaptability to illumination and scale change.Then,support vector machine(SVM)is used to train the extracted multi-dimensional feature vector to realize the classification of moving barrier and natural background.Finally,according to the GPS positioning module of the detected vehicle,the image position information is recorded and the detection report is output.(3)Experiment and analysis: the functional modules of the W-Beam barrier damage detection system are tested and analyzed.It includes bridge shadow area detection experiment,W-Beam barrier damage detection experiment,road fluctuation influence experiment,and movable barrier detection and identification experiment.