Detection Method of Wet Shotcrete Images in Complex Tunnel Construction Environment

The intelligent operation of wet shotcrete trolleys in complex tunnel construction environment has attracted widespread attention. The problems of uneven illumination, low contrast, and serious noise effect are common in wet shotcrete images collected in large space tunnels, which greatly affect the accuracy of image detection. Therefore, this paper proposes a pretreatment method suitable for tunnel wet shotcrete images combining the Mask dodging algorithm and the contrast limited adaptive histogram equalization algorithm. A threshold segmentation algorithm based on Sobel edge detection and Otsu method is proposed to realize the preliminary extraction of steel arches. Then, use the morphologic closing operation and the characteristics of an area, rectangularity, and length of steel arches to filter the noises to obtain a purer map of steel arch distribution. Finally, according to the outermost corner coordinates of each steel arch, the maximum external polygon is fitted to cover all the area which need to be wet sprayed next. The experimental results show that the proposed method can accurately extract the distribution of steel arches and mark the wet shotcrete area, which provides the perceptual information for wet shotcrete trolleys.