Crack Shape Detection On The Structural Surface Based On Image Analysis Technology

Crack shape is one of the important parameters in structural state evaluation, so it is very important to accurately detect the length and width of cracks on the structure surface. Currently, the detection of the length and width of the crack is basically based on the method of artificial close contact measurement.This method is extremely dependent on the experience of the testing personnel, which leads to poor repeatability of detection results, large environmental limitation and low detection efficiency. In view of the deficiencies of manual detection and evaluation of cracks on the structure surface, it is necessary to introduce an objective and quantitative measurement method which can be used for remote detection without contact.Based on the image analysis technology, this paper proposes a novel structural surface crack detection approach that can detect the crack characteristic parameters in the image. The corresponding program is written in MATLAB and the validity of the proposed method is verified by experiments, the experimental objects include different width, length, angle, and type of lines which are precise printed on photo paper to simulate cracks, 10 linear cracks and 1 net cracks on the surface of concrete structure. The main research results are as follows:(1)This paper proposes a novel structural surface crack detection and assessment approach. A standard square attached on the structural surface was used as a reference mark to calibrate the original deformed images through the projection transformation,then extracting the crack skeleton and removing the branches and burr, etc. Finally,the crack characteristic parameters in the image can be calculated.(2) Using 12 standard straight lines and 3 curves on the photo paper to simulate the crack, the validity of the image correction and calibration algorithm is verified.Basing on different shooting angles, the length identification accuracy of vertical straight lines, inclined straight lines and curves can reach 99.73%, 99.64% and99.56% respectively, and when the width is not less than 0.2mm, the recognition accuracy of width is 95.84%, 95.60% and 95.52% respectively.(3)The light intensity has a large influence on the recognition accuracy of crack width, but the impact on the accuracy of crack length is small. When the light intensity is 1400lux~3100lux, the recognition accuracy of the crack width is relatively high, and the average relative error is less than 5%.(4)The algorithm proposed in this paper has a high accuracy in the identification of the net crack characteristic parameters. Based on the vertical and oblique image,the length recognition accuracy of the net crack can reach 97.36% and 96.95%respectively, and the width identification accuracy of the crack is 94.79% and 94.46%respectively.(5)The detection results of 10 linear cracks on the surface of concrete structure show that the method proposed in this paper has a high detection accuracy in the length and width of cracks. Based on the vertical and oblique image, the length identification accuracy of concrete surface cracks can reach 97.91%% and 97.65%respectively, and the identification accuracy of width is 94.80% and 94.14%respectively, it shows that this new method can reach the requirements of practical engineering application.