Abstract:
To solve the problems of traditional tunnel crack detection, such as low efficiency, subjectivity and poor safety, among others, a method of tunnel crack extraction based on adaptive threshold and connected domain marking is proposed using orientation and continuity characteristics. First, according to the orientation characteristics of cracks, an Arabia digital algorithm (Algorithmic) is designed to roughly extract cracks. Adaptive threshold is chosen for threshold selection in the formula, and the improved threshold iteration method is used to automatically obtain the best threshold without manual intervention. Then, according to the continuity characteristics of cracks, the connected region labelling method in mathematical morphology is used to extract the cracks. By controlling the area of the connected domain, the denoising processing after the rough extraction of the crack is realized, and the repair treatment after the rough extraction of the crack is realized by expansion and corrosion operations. Finally, a total of 165 different types of crack images were selected as experimental samples, and simulation experiments were carried out on MATLAB. From the experimental data, we observe that the extraction precision of the adaptive threshold and the connected domain label extraction method can be as high as 94.2%, the average running time is only 35.4 s, the error recognition rate and the rejection rate have been controlled at 2.7% and 1.1%, respectively. Compared with traditional image processing methods, our proposed method demonstrates remarkable improvement and shows promise for future applications.