Abstract

In order to overcome the high computational complexity of beamlet tree-structure algorithm, the paper proposes an improved algorithm and applies it to pavement crack detection, thereby; solving the problem of pavement crack detection which has the disadvantages of poor noise immunity and inaccurate test results. First, the pavement crack image is rectified by multiplicative factors to eliminate the influence of non-uniform background illumination. Then, the image is transferred to the binary image by Otsu’s threshold segmentation algorithm. At last, based on discrete Beamlet transform and integrating multi-scale tree structure of beamlet itself, extract pavement crack from the binary image by using the beamlet tree-structured algorithm, which change a “bottom to top” strategy to look for the optimal value of objective function into a “top to bottom” searching process. Therefore, the proposed method reduces calculation complexity and time. Experimental results show that the proposed method can quickly extract crack from the complex pavement background and noises. Moreover, it keeps the continuity of the crack as well.

Key words: Beamlet transform, tree-structured algorithm, Otsu algorithm, crack detection.