A de-noising algorithm was designed on the basis of connectivity checking of pixel and crack block levels to remove isolated noises in pavement crack images and to retain the fine features of crack edges. This algorithm comprises three main parts, namely, (1) pixel connectivity checking to eliminate pixels with poor connexity, (2) a de-noising algorithm to remove insular target regions, and (3) an cracking block de-noising algorithm. The algorithm procedures and parameter selection issues were presented. Performance tests on the proposed algorithm and two commonly used methods (median filtering and the algorithm based on long line and black pixel percentage) were conducted using Visual Studio 2008 and OpenCV. Results indicated that the proposed algorithm could effectively remove isolated noises and maintain the continuity of cracking edges with higher precision (85.06%) and recall (85.80%) and better scores (0.74% to 19.19% and 0.20% to 12.06%) in comparison with the other two methods.
De-Noising Algorithm for Pavement Crack Images Based on Bi-Layer Connectivity Checking
2016-09-15
82016-01-01 pages
Article (Journal)
Electronic Resource
Unknown
Seed-Based Approach for Automated Crack Detection from Pavement Images
Transportation Research Record | 2016
|Deep Learning-Based Real-Time Crack Segmentation for Pavement Images
Springer Verlag | 2021
|