Research On Data Collecting Techniques For Pavement Performance Parameters Based On Video Images

In order to detect pavement performance parameters effectively, accurately and automatically, a method based on video image or image pairs is adopted to detect pavement performance parameters. Based on principle of photogrammetry, a stereo image pairs is used to calculate the three dimension coordinators of the object points on pavement. Computer image processing technique is used to pre-process the pavement distress image in order to get binary pavement distress image.Based on the fundamental theory of photogrammetry, considering the property of non-measuring camera, the direct linear transformation (DLT) method is adopted to calculate the DLT indexes of common digital camera. It is confirmed that it is not feasible to calculate the orientation elements of common digital camera according to the DLT indexes using image inner direction method. The calculating equations for orientation elements of common digital camera are deducted using DLT method based on digital array coordinator assumption. According to the definition of epipolar line in photogrammetry, using the orientation elements of non-measuring camera, the epipolar line passing given image point in left (right) image of the stereo image pairs and its cognominal epipolar line in the other image are deducted.The established binocular recognition theory for non-measuring camera is applied to detect the pavement performance parameters. For deformation distress in pavement such as rut, active and passive pavement rut depth detecting methods are put forward. The lane width measuring methods under real lines and all dashed lines conditions are established.On the basis of image noise detecting, the adaptive median filtering and adaptive weighted mean filtering algorithms are adopted to filter the pulse and Gaussian noise in pavement distress image respectively. According to the distribution character of the gray level of each pixels in pavement distress image, the weighted threshold algorithm jointed with global threshold and regional threshold of the pavement image is used to segment the pavement distress image. Another segmenting algorithm for pavement distress image is established and the gray level histogram of the sub-region image and interpolating algorithm are used to determine the threshold of the sub-region image.