Research On Image Calibration And Crack Identification Methods Of Vehicle-based Detection System Of Pavement Cracks

With the rapid development of Chinese economy, the highway traffic plays an increasingly important role in the national economy. On the occasion of the rapid development of highway construction in China, the maintenance and management of highway technology should also be developed. Pavement crack is one of the important factors influencing the quality of highway pavement, it’s the initial manifestation of highway pavement most diseases, which directly affects the traffic safety and service life of highway. Therefore, early detection of highway pavement cracks on the pavement crack and timely maintenance can effectively prevent the further development of the crack, and keep driving comfortably of the vehicle on the highway pavement. Research on vehicle detection system of highway pavement crack bearing capacity, improve vehicle safety and reduce road traffic accidents is of great significance to guarantee the highway pavement.According to the crack detection research status of highway pavement at present, the calibration of vehicle pavement crack detection system and methods of image identification of crack. In order to obtain the pavement crack information quickly and clearly, the design of vehicle pavement crack image acquisition system. Since the acquisition system for vehicle pavement crack image in pixels, and not directly to the real dimensions of the pavement crack, so the reconstruction of pavement crack image calibration and real crack size were studied. First, two-dimensional target board is put on a level road, through vehicle pavement crack acquisition system for obtaining the target in the image, and a two-dimensional target feature points ranking method is presented, to improve the calibration process of target feature points recognition efficiency, the corresponding relation of target plate character image coordinates and the real world coordinate to obtain the transformation matrix. Finally, the world coordinates of the pavement crack are obtained according to the image coordinates of the crack region in pavement crack image and obtain the transformation matrix, the size of real crack image reconstruction. In the crack image acquisition system of vehicle pavement, due to the interference of the light, mechanical vibration and other noise, the pavement crack images should also be processed. mechanical vibration and other noise. First of all, the linear filter and median filter of images are used tofilter the image, and the effects of the two filtering methods are compared, and then using histogram equalization and gray level transformation for crack image gray enhancement, to gain the better visual effects, and the high contrast of image background and crack area for pavement crack image. In order to make further identification and statistics of the crack area in the image. First of all, using the OSTU method to binarize segmentation of the pavement crack images, and then, the banalization image corrosion expansion processing, and the value of the two images in each connected region area and the connected regions of the external rectangle ratio and the threshold comparison method to determine the crack area,identify the two value of crack area in the image; finally, using the Hilditch thinning algorithm and fast parallel thinning algorithm for crack regions in the image thinning, to obtain the crack skeleton image. According to the calibration projection in the image coordinate system and the results of the crack skeleton, rebuilt in the world coordinate real scale cracks. Put forward the evaluation formula of reticular crack density, the calculation of classification and statistical information of the pavement cracks.