Prediction of Aggregate Gradation of Bituminous Mixtures Using Image Analysis

The gradation in the bituminous mixtures plays a significant role in predicting its performance. Laboratory experiments such as centrifugal extraction method and ignition method are used to determine the gradation. However, the solvents used in centrifugal extraction method are considered to be carcinogenic. Since then, several studies have focused on using the principles of image analysis to determine the gradation from the images. However, limited studies are conducted on evaluating the suitability of image acquisition methods and comparing different particle size estimation algorithms. In this study, the gradation of dense bituminous concrete (BC-19) and open-graded friction course (OGFC) were predicted using the principles of image processing and analysis. The experimental variables included compaction method (Marshall and SuperPave gyratory method), sectioning method (vertical and horizontal sectioning’s), and particle size estimation algorithm (circle fitting and contour fitting methods). The analysis results indicated that gradation predicted from the images was found to be finer above the cut-off sieve size and coarser below the cut-off sieve size. The gradation predicted from images obtained using Marshall method of compaction agreed well with gradation that was used in specimen preparation compared to SuperPave gyratory compaction method. The combination of the horizontal sectioning method with circle fitting algorithm has resulted in highest R2.