Posted: 2017-10-15

Measurement of bitumen adhesion using CV

Keywords: computer vision, modified bitumen, Wolfram Mathematica, aggregate adhesion

DISCO (Decent Inexpensive Script Computing Opacity) is a script created for the purpose of measuring adhesion between bitumen and aggregate in hot mix asphalt.


A raw image of coated aggregates is loaded by the script. The software transforms the image (Figure 1.) by means of removing the background, scaling, detecting edges of size markers etc. After the procedure. A .png image is produced with the detected bitumen coating and information regarding the percentage of coated area. Additionaly in case of batch processing a tab separated file with [number, sample name, coating percentage] is generated data.txt. Assessment of Polyethylen-Modified Bitumen Adhesion Using Computer Image Analysis article was created based on the results obtained using DISCO.

Figure 1. Image transformation pipeline

Code grimoire

The script was written in Wolfram Mathematica 10. The whole DISCO script may be downloaded from this github repo.


First run the init.nb. Modify the absolutePath variable so that it points to the location of the cloned repo. Then select Evaluation >> Evaluate notebook.

Single file processing

In order to use the single-file.nb you need to first evaluate init.nb which contains all the procedures, sample variable holds which test image to in string format. The pipeline variables were calibrated for this particular sample set which is in sample/ directory. In order to evaluate a single file select Evaluation >> Evaluate notebook.


In order to use the batch.nb you need to first evaluate init.nb. Set the start and stop parameters pointing to the range of files designated for processing. In order to start the batch process select Evaluation >> Evaluate notebook.


In order to use the debug.nb you need to first evaluate init.nb. The debug modul allows for calibration of the CV parameters namely, edge detection, background chroma etc. Before adjusting the sliders run the notebook by selecting Evaluation >> Evaluate notebook.


Figure 2. Results produced by the script

The method proves to be highly effective in comparison to the traditional approach of measuring bitumen-aggregate adhesion (Figure 2).