Automatic road crack identification and its measurement

The methods of automatic road crack identification and its measurement were studied for improving the efficiency of road crack detection, ensuring the security of the detection work, and strengthening the standardization of detection data. Through analyzing the defects of existing methods, the image smoothing and segmentation method with mean shift technique were proposed to make a preliminary segmentation of candidate cracks; a directional tracking method was used to extract the crack skeleton; using the crack skeleton to interpolate the full cracks, automatic crack extraction and encoding were achieved; finally, the crack skeleton segments were used to accurately calculate the parameters of crack characterization. All those could guarantee a complete description of crack characterization. The results show that the proposed method, which possesses a higher precision and reliability, is insensitive to the roughness of pavement and the directions of cracks, and is able to detect small cracks, whose accuracy reaches less than 0.5 pixels with relative error of length less than 2%.