Crack detection and spatial marking of ancient buildings by near-view and far-view combination

In the protection of ancient buildings, not only crack parameters need to be detected but also crack positions need to be known in the spatial structure. As the existing image detection methods cannot consider both detection accuracy and field of view scope, a detection scheme that combines near-view images and far-view images is proposed. Near-view image realizes crack identification and parameter extraction. A region growth algorithm based on local Otsu and local mean is proposed to improve the efficiency of crack extraction. A mapping method based on ASIFT and homography matrix is proposed to realize the crack mapping from near-view images to far-view images and complete the crack marking of far-view images. The 3D reconstruction of the processed far-view images is carried out by the multi-view geometry method to realize the marking of cracks in the spatial structure. Through the actual measurement of two ancient buildings in the Da Ci’en Temple in Xi’an, the effectiveness of crack identification is verified. The detection accuracy of the average crack width is less than 0.1 mm, and the spatial marking of cracks is realized. The crack detection of ancient buildings on stereo vision is realized, which can effectively guide the restoration of ancient buildings and has important reference value for the protection of ancient buildings.