Due to the different of image processing method between pavement disease detection and estimation system with handheld device and road detection system with vehicle, this paper proposes a concrete pavement crack detection method based on image processing, which can correct image distortion. Firstly, camera calibration is accomplished at first to obtain pterior parameters and exterior ones. Secondly, preprocessing is implemented, which includes gray image, grayscale stretching and median filter. Thirdly, choose threshold segmentation algorithm(OTSU) to segment crack and background. After that, based on the different count of background connection domain between linear crack image and reticular one, classify cracks automatically. Linear crack can be subdivided to vertical, horizontal and slant crack thought projection method. The length of Linear crack can be obtain by thinning to single pixel line, then the mean width of linear crack is calculated by its percent of plate pixel count. Finally, the damaged area is calculated with reticular crack minimum enclosing rectangle.