| Summary: Asphalt quality is directly related to the proportions of the components contained in its composition. Manual measurement of this ratio is a time-consuming and laborious process. In this study, an automatic segmentation method based on images obtained from the cut surfaces of asphalt samples was proposed first. For this purpose, an imaging system was set to obtain images under controlled lighting conditions from asphalt samples. A database of 46 asphalt samples was created with the images obtained from this setup. Although these images were taken under homogenous lighting conditions, they were also filtered through various filters to remove possible gray level irregularities. Then, using the histogram of gray level regulated images, the asphalt samples were segmented into three classes as aggregate, bitumen, and voids. The results were compared with manually segmented images by the expert and the performance analysis was given. According to the results of the segmentation performance analysis, it was observed that the proposed method obtained approximately 85% general classification accuracy on the images in the database. |
