The Pavement Crack Recognition Algorithm Research Based On Matlab

The pavement crack detection is important part for the maintenance and management ofhighway road disease.In recent years, automatic detection of pavement cracks has beenwidely used. And because of the complexity of the pavement crack image, the detectionalgorithm affects the accuracy of the test results directly. Therefore, this paper will focus onthe detection of pavement crack disease. In order to improve the accuracy of detection, thispaper dose the further research from the crack image denoising, image enhancement, imagesegmentation and image feature extraction of pavement cracks after detection respectively.In the pavement crack image, the cracks information and the background low contrast,thus it’s very difficult to detect cracks directly. For image preprocessing, this paper first doesthe image gray-level correction, and then does the image filtering. After that, proposes animproved median filtering method for image denoising, then enhances the image further basedon the principle of fuzzy theory and improves the pavement crack image contrast effectively.For pavement crack image segmentation, this paper uses the threshold-basedsegmentation and morphological multi-scale thoughts, for regular shape of crack images usinga thresholding to deal with, for irregularly shaped of crack images, puts forward a kind ofmulti-structural elements anti-noise edge detection operator, and according to that differenttypes of crack edge have different filling chances, it sets algorithm of self-adaptive weights,which enables the operator test all types of edge crack, improve the detection accuracyeffectively.For the problem of fracture cracks and noise that may be present in the pavement cracksafter image segmentation, in this paper, the narrower fracture cracks in the image with theclosing operation and open operation in morphology to deal with, for the wider fracture cracks,puts forward based on the growth of the active method of connecting the fracture cracks piece,so as to improve the efficiency and accuracy of the connection, make whole segmentationresults are clear and complete. Ultimately, taking use of the results of previous processing，wecan extract the useful information of pavement cracks. By supposing a set of conditions，theconnected domains of cracks can be extracted, then judgment the type of cracks, and finallycalculate the area of the map cracking and calculate the width and length of the linear cracks.