Research Of Road Crack Identification And Classification Based On Image

Crack is a common damage of pavement. Traditional manual methods gradually can't adapt to the demands of highway development. Intelligent crack detection system has become a critical research. However, undeveloped hardware and software technology of image acquisition and complex road conditions, lead to low quality of image. This aspect brings great difficulties to process the image, which is the main reason of low rate of crack detection. Therefore, research and design on a robust identification system which can adapt to different image background, has an important significance.The paper is a study based on image processing method. According to the crack shape in the image of different backgrounds, there are four types of crack in this paper, including linear crack, block cracks, map cracks and repaired cracks. The first three types of cracks are deeply studied for crack detection. And according to the results of the detection algorithm, classification algorithm is also studied, forming a set of solutions from the crack detection to the classification.On the study of crack detection algorithm, in the three modules of image preprocessing, image segmentation and object extraction of cracks, this paper put forward the template operation, reducing noise based on linear and region information and some other methods. These methods are innovative, and achieve a high rate of crack detection. In addition, based on the result of test, this paper specially designed an algorithm to achieve adaptive threshold, to solve the problem of stability, caused by fixed threshold.On the study of crack classification algorithm, according to the characteristics of test results, a method achieve a good classification of linear crack, block cracks, and map cracks, which using chain code or feature mapping to decide the feature of direction, and triangulation to decide the feature of distribution.