A Two-stage 3D Pavement Crack Detection Procedure Based on Primary Surface Profile Optimization

Pavement cracks are considered to be the main indicator of pavement performance. Because traditional artificial detection is dangerous, time-consuming, people have developed high-speed road image acquisition technology, and a large number of algorithms have been proposed to detect pavement cracks. However, vehicle jitter and road conditions are not taken into consideration, making the quality of the images collected by the data acquisition system not high, and the detection performance of various algorithms will inevitably be affected. Therefore, this research optimizes the Primary Surface Profile (PSP) filtering algorithm, and proposes a two-stage 3D pavement crack detection hybrid program. First, the 3D road surface data is divided into two stages for processing. Then, an improved adaptive PSP generation algorithm is introduced, and in each detection stage, the obtained PSP curve is embedded in the two-stage calculation. Finally, the matched filtering algorithm and image morphology processing are used as the refinement of the crack, and then the crack area is obtained accurately. The experimental results show that our method is superior to the current more advanced 3D pavement crack detection algorithms in terms of Precisions and Recalls.