Identification And Safety Warning Of Tunnel Crack Based On Image Processing

Tunnel cracks pose a serious threat to the safety and durability of the tunnel structure itself,which will directly lead to traffic paralysis,and even endanger safety,resulting in the loss of people’s lives and property.With the increasing of tunnel length,the detection of tunnel cracks takes a lot of time and manpower,and there will be problems of low efficiency and low accuracy in the process of detection.Therefore,improving the efficiency and accuracy of tunnel crack detection and increasing prevention measures are of great theoretical and practical significance for tunnel safety passage.A large number of tunnel wall crack images are obtained by using CMOS camera,and a large number of contrast tests are carried out by using image processing technology.The algorithm presented in this paper has good characteristics and performance,among which: in the crack identification,the accuracy is high;in the crack detection,the segmentation effect is good,the noise interference is less;in the crack width calculation and safety warning,the detected crack width value can be estimated and its safety condition can be warned.1.Aiming at the problem of low accuracy of tunnel crack recognition,a method of crack image pre-processing with uniform light compression and pre-segmentation is presented before extracting HOG features.the crack image is first uniformly processed,then the image wavelet transform is used to pre-segment the compressed crack image.after the process of pre-processing,the HOG features of the image are extracted SVM identify the crack image.This method reduces the time of feature calculation and recognition,and the accuracy of recognition is higher and the recognition time is shorter.2.Detect the crack area in the crack image.Aiming at the problem that crack image is noisy and prone to error detection,a new method of fracture segmentation based on dimension reduction fusion is proposed.PCA principal component analysis method is used to extract two fuzzy figures with differentprincipal components,and the high dimension image is preliminarily segmented and the Wiener filter is reduced.the initially segmented crack image is integrated into the low dimension image by Poisson fusion method,and the crack region is detected after secondary segmentation.Through contrast experiments,the segmentation area is accurate and the noise interference is less.3.The width calculation and safety warning are realized for the split crack area.A width calculation method based on slope division and Euclidean distance is proposed,which integrates the rating standard of tunnel crack hazard grade and realizes early warning of tunnel crack image.First,the image is divided into blocks to judge the crack type,that is,the slope of the connecting line,to calculate the Euclidean distance between the two intersection points in each block and its corresponding projection points,and to calculate the average crack width of all crack area blocks.The evaluation standard of the grade of the crack is integrated by dividing the width value range,and the grade division and safety alarm are realized for the damage degree of the crack.In this method,the error of the estimated value of crack width is controlled at about 5%,and the effective hazard rating and safety warning are carried out accurately.