Skip to Main Content
A detecting approach has been developed in view of the properties of cracks in the pavement image. Because of the uneven illumination, threshold causes difficulties in applications of pavement image segmentation. By analyzing the signal model, we can use bilinear interpolation to obtain the correction image based on the background subset which is extracted from original pavement image. Then, the segmentation threshold could be calculated by the histogram of the correction image based on statistical criterion. After the operation of binary segmentation, final crack distress can be achieved by sequential operation of spreading cracks and eliminating isolated noises. Experimental results show that the approach is also applicable for detecting blocky distress.
Date of Conference: 24-25 April 2010