Research On Pavement Crack Recognition Based On Image Processing

With the rapid development of highway construction and gradual improvement of road network construction in China, road maintenance work has been paid more and more attention. Pavement crack is the main form of road diseases. It is also an important indicator of the road quality assessment. The traditional manual detection and recognition methods are not able to meet the requirement of rapid development of highways, so the research of pavement crack automatic detection and recognition is particularly urgent. Therefore, the research works on pavement crack recognition based on image processing have been done in this thesis.Firstly, the research of image pre-processing is made after the characteristics of the pavement crack image are analyzed. The pavement crack images which we collected inevitably contain much noise, which cause many difficulties in classification and recognition of pavement crack image. In order to facilitate subsequent operations, the image is enhanced based on gray transformation and weighted neighborhood average filter. Secondly, the research of image segmentation is made after the image preprocessing work is done. Although noise has been reduced after image preprocessing, the SNR still cannot meet the requirement of effective extracting the crack edge. For solving this question, the image is filtered by using multi-scale morphology operator which can hold more complete crack features and effectively reduced noise furthermore. And then, it is proved that using Sobel operator can get the best result in edge detection with the comparison of the several edge detection operators. Based on this, after the holes inside the edge are filled and the isolated and small regional noises are removed by using mathematical morphology operation. Furthermore, the binary crack image is extracted and the pavement crack image segmentation is completed.Finally, crack feature extraction and recognition are studied. On the basis of analysis of characteristics of various types of pavement crack characteristics was accomplished, three kinds of features are extracted from the pavement crack image. The first is to extract projection features of pavement crack image with the vertical projection and horizontal projection of pixel statistical chart. The second is to extract crack features based on proximity algorithm after getting the projection statistical chart. The third is to extract density factors of distress. Then, classification and recognition of the pavement crack image is completed based on SVM algorithm. The experimental results based on183images demonstrate the density factors of distress can get the best detection of about96.77%and the types of pavement crack are recognized effectively by the scheme in this thesis.