An Improved Crack Detection Algorithm Based on Characteristics of Texture Analysis and Gravity Model

Aiming at the defect that conventional crack detection methods cannot accurately extract crack information from complicated background noises, a crack detection algorithm is proposed on the basis of characteristics of texture analysis and gravitational model to overcome the shortcomings of the existing global methods. Shape and texture features of the crack sub-patches are extracted and multiple regularized terms are established. To the local feature, the direction factors and directional derivative discrimination criterions are designed, which reflect the linear features of cracks. The experimental results indicated the validity of the proposed method.