A novel laser stripe center extraction method for pavement rut detection

The detection of road rutting is of great significance for reducing traffic accidents, verifying the degree of road damage, and improving the comfort of driving. The fast, real-time and accurate stripe center extraction algorithm is the key to ensure the real-time and stable operation of the system, which will directly affect the ultimate rut depth acquisition accuracy. For the existing laser stripe center extraction methods cannot meet the characteristics of good robustness, high precision, real-time, strong anti-noise ability at the same time. A signal correlation method, which always used in radar data processing, is proposed in this paper to extract the center of the laser stripes. The above algorithm can solve the problem of disconnection, but due to the use of a fixed reference column, the selection of the reference column is accidental, the relevant results will be affected by different reference columns. In order to solve this problem, this paper uses the method of multi-column data correlation to take the maximum value. The algorithm is simple and practical, which can achieve the sub-pixel level extraction accuracy, meet the real-time requirements, has strong anti-noise ability and repair ability for broken lines. The experimental results show that the algorithm is insensitive to high light intensity background, and the laser stripes center extraction accuracy can reach 0.15mm, which is far less than the 1mm inspection requirement for rutting road test specifications.