The Research Of Pavement Crack Detection Algorithm Based On Natural Light Condition

Highway plays a more and more important role in the national life.maintenance of roads are indispensable.The formulation maintenance plans depends on the feature data of pavement diseases.How to quickly and accurately detect pavement diseases has become a crucial task.In order to overcome the time-consuming and labor-consuming of traditional manual detection,as well as the high cost and poor applicability of crack detection vehicles,the crack detection technology based on motion cameras has emerged.There is a shading problem for the pavement image collected by the motion camera.This paper presents a pavement crack detection algorithm based on natural lighting conditions.Firstly,morphological operations,Gaussian filtering,threshold segmentation and edge detection operators are used to detect the umbral region and penumbra region of the image.The objective of shadow removal in the umbra region is achieved by using the illumination compensation model.Since the pixel values of the penumbra region are mutated,the cubic spline interpolation algorithm was used to remove the penumbra shadow and the light compensation model was used to recover the texture information of penumbra region,thus a complete non-shadow image is obtained.Secondly,the non-shadow image is processed by K-means clustering algorithm and Otsu threshold segmentation algorithm to obtain the preliminary binary image of crack respectively.Through comparison and analysis,K-means clustering is superior to Otsu threshold segmentation algorithm,so K-means clustering is selected to finish crack detection.Finally,because of the interference of oil stains,texture information,and road joints on crack detection,the crack binary image filtering is completed based on the area,aspect ratio characteristics of the connected area,and the Hough Transform algorithm to obtain complete crack detection.results.As a result,crack feature extraction and type determination was performed based on the crack detection results.A large number of experimental studies have been carried out on the shadow removal algorithm and crack detection techniques proposed in this paper,and compared with thecurrent typical shadow removal algorithms and crack detection algorithms.The experimental data show that the pavement detection algorithm based on natural lighting conditions proposed in this paper is effective by combining the shadow removal algorithm and crack detection algorithm,It can efficiently and accurately identify cracks for different types of crack images,and thus provides a good strategy for the formulation of pavement maintenance.