Image Processing–Based Classification of Asphalt Pavement Cracks Using Support Vector Machine Optimized by Artificial Bee Colony

Timely and accurate detection of asphalt pavement crack is very crucial in pavement maintenance. This study establishes an intelligent approach for automatic recognition of pavement crack patterns. Image processing techniques including nonlocal means, steerable filter, projective integral, and image thresholding are used synergistically to extract useful features from digital images. A machine learning model that comprises the multiclass support vector machine and artificial bee colony optimization algorithm is constructed to perform pavement crack classification. Based on feature analysis, a set of features derived from the image projective integral is found to significantly enhance the prediction performance. Experimental results supported by statistical test demonstrate that the proposed integration of image processing and machine learning model achieves an outstanding classification accuracy rate that is more than 96%. Hence, the proposed approach is a promising alternative to assist transportation agencies in pavement inspection and maintenance planning.