The Adaptive Threshold Segmentation Method For Pavement Crack Disease Research And Implementation

The construction of high-grade highway network in our country has been in the leading position of the world.With the development of roads becoming more mature and perfect,crack detection has become the focus of the road maintenance management department.Traditional manual testing has been unable to meet the fast-paced development needs of modern highways,making automated pavement crack detection technology particularly important.In recent years,the rapid development of digital image processing technology has provided a new space for automatic crack detection.This paper focuses on the most critical segmentation algorithm for pavement crack detection.For the actual road surface crack image,on the one hand,there are external disturbances such as uneven illumination and environmental noise.On the other hand,the direction of the crack itself is complex and variable.Therefore,achieving ideal segmentation is a difficult point in pavement crack detection.Aiming at the problems of tedious and inconspicuous processing of some current research algorithms,this paper chooses the adaptive threshold-based segmentation algorithm to achieve the balance between complexity and segmentation quality,and introduces the basic principle of threshold-based correlation algorithms.On this basis,this paper firstly combines radar signal detection and digital image processing related techniques to propose an adaptive threshold crack segmentation algorithm based on one-dimensional constant false alarm model.The method establishes the signal model by extracting the one-dimensional pixel gray value of the image,uses the constant false alarm algorithm model in the radar signal processing to detect the threshold,determines whether the crack signal of the road surface exists,obtains the gray level-position information of the crack,and then divides the crack from the background.come out.Then a D-neighbor sliding window adaptive threshold crack segmentation algorithm based on two-dimensional statistical features is proposed.The method uses the statistical features of the gray distribution of the image combined with the D neighborhood sliding window to discriminate the pixel points in order,and finally morphologically filters the results to complete the accurate crack segmentation.Finally,using MATLAB simulation combined with FPGA hardware implementation algorithm theory verification.Taking three different types of cracks in horizontal,vertical and block shapes as examples,the experimental results show that the two proposed algorithms can not only accurately segment different types of cracks in non-uniform background,but also improve the quality of segmentation compared with traditional edge detection algorithms.It can fully meet the requirements of modern highway crack detection.