Research On Bridge Crack Defect Detection Technology Based On Machine Vision

With the continuous advancement of China’s national highway network construction and the continuous improvement of the highway transportation system,the inspection work of bridges has been paid more and more attention.The timely assessment of the health status of the bridge to take corresponding maintenance measures not only ensures the safe travel of the people,but also avoids economic losses.Cracks are the beginning of bridge lesions.Crack identification is also the core link in bridge inspection operations.It is of great significance to achieve non-destructive,rapid and accurate detection of bridge cracks.Therefore,based on machine vision technology,this paper has carried out research on bridge crack detection technology.The main contents include:(1)In view of the problems of complex background texture,diversity of crack shapes and inconsistent gray levels in the cracks of bridge images,this paper uses pre-processing technology to enhance the linear structure while suppressing noise,and secondly uses a sliding window grid clustering algorithm to detect local areas Then,based on the regional growth method,the coarse segmentation of cracks was achieved.Finally,the principle of structural similarity was used to eliminate noises such as pseudo cracks,and the crack structure was accurately extracted.(2)Aiming at the existing crack segmentation algorithm,the widening crack width causes measurement errors and easy to lose parts of low contrast and low shallow cracks.This paper proposes a minimum path pixel-level crack recognition method based on texture anisotropy.By analyzing the relationship between the pixel growth path texture and its own spatial position and gray scale,it proposes pixel-level intensity and gradient features,merges the saliency of multiple features,and adopts a hierarchical discriminant model to identify cracks in pixel detection.(3)In view of the problem of low detection efficiency of pixel-level crack identification methods,this paper proposes a crack area search algorithm to determine the crack area and narrow the detection range.At the same time,during the pixel detection process,conditional inter-cluster clustering is used for pixels on the growth path The method skips the detection of pixels with the same attribute,which greatly improves the detection efficiency of the crack detection algorithm.