Output details

Application of texture analysis and Kohonen map for region segmentation of pavement images for crack detection

Distinguishing cracks from patch boundaries have significant consequence as 90% of the roads are highly patched and/or trenched. Since pavement management decisions are based on crack evaluation results, the overestimation of cracks caused by the detection of surface irregularities can cause erroneous prioritization and inefficient utilization of rehabilitation funds. An intelligence-based method, using textural and colour properties of different image regions with an un-supervised learning procedure, was developed to differentiate the edge of a patch repair and cracks from digital images. The simplicity and high accuracy (> 97%) is expected to attract interest from researchers and road authorities.