| Concrete has the advantages of wide source of raw materials,low cost and good durability,and has been widely used in the field of highway and construction.With the increasement of the service life of concrete buildings,due to the erosion of climate and other external environment,the strength of concrete decreases,and various kinds of surface diseases will appear,which seriously affect the safety and durability of concrete buildings.Among them,crack is the most common one of concrete surface diseases.Timely crack detection is significant for judging the safety state of concrete buildings and subsequent repair and reinforcement.Traditional artificial crack detection methods are time-consuming and inefficient,the existing image processing algorithms are not applicable to concrete images with complex background,much noise and weak crack information.In view of the above problems,this paper designs a concrete crack identification and extraction method based on image processing.The main research contents are as follows:(1)Research on the preprocessing algorithm of concrete crack image.The weighted average method is used to reduce the amount of image processing.Mean filtering is used to smooth gray image and reduce salt and pepper noise.In view of the poor contrast of crack image and the poor adaptability of existing enhancement algorithms,an image enhancement algorithm based on pixel gradient was proposed,which effectively enhanced the edge of crack and improved the contrast of crack image.Probabilistic Hough transform algorithm is used to identify the grooves in concrete pavement,which can effectively eliminate the interference of grooves to crack extraction.(2)Research on concrete crack segmentation and extraction algorithm.According to the local gray scale difference of the image,the image is segmented by Sauvola algorithm,which automatically obtains the optimal segmentation threshold.Moreover,the time complexity of the segmentation algorithm is simplified by using the idea of image integration,and the segmentation of the crack image is realized.Aiming at the problem that noise is widely distributed in binary image and difficult to be removed,a linear noise removal method based on crack contour offset Angle is proposed,which can be used to remove noise in binary image and extract concrete cracks accurately by combining crack boundary length and roundness features.(3)Research on calculation method of concrete crack characteristic parameters.Improved Zhang parallel thinning algorithm is used to analyse the fracture refinement,a template based on the burr removal methods to eliminate false skeleton branches,through statistical skeleton pixel number calculation length of crack and central point search method is used to calculate the average width of crack,statistics of pixel quantity calculation cracks in the area,through an external rectangular box to choose automatic tagging of complete fracture crack position and provide reference for safety evaluation of concrete. |
