Development of Automatic Concrete Cracks Detection System Using Average Shifted Mesh

Abstract

Extending the service lives of bridges has become a topic of discussion. Inspections and diagnosis are performed through close visual inspections. In this study, we propose a diagnostic imaging technology with a super-high-resolution camera (SHRC) for efficient inspection. By using images captured with the SHRC, we will build an inspection environment similar to close visual inspection. We utilized image recognition technology to prepare AI that identifies cracks, and proposed a method that delineates locations with cracks based on the entire image of a pier. Finaly, We examined the improvement of crack detection accuracy by using Averaged Shifted Histograms (ASH) theory. It becomes crear that accurancy of AI has increased by average shift mesh.