case study

Titanium-coated surfaces a prone to tiny defects such as very small cracks that are not easily observable to the naked eye or optical microscopy. We want to determine if images from a High Resolution Scanner (up to 1000 dpi) are good enough to be used for analysis to determine quality and distinguish defects. 

Two new binarization algorithms that were based on Otsu’s method were developed and tested on samples of titanium-coated aluminium that were scanned at different resolutions. The two methods were the only ones able to detect the cracks correctly at lower resolution (700 dpi), and performed similarly to Minimum Error Thresholding and generally better than the Otsu’s Method at higher resolutions. A full disclosure of the algorithm, and the findings of the project were published in the IEEE Transaction on Industrial Informatics in 2015.

Reference:

M. Win, A. R. Bushroa, M. A. Hassan, N. M. Hilman and A. Ide-Ektessabi, “A Contrast Adjustment Thresholding Method for Surface Defect Detection Based on Mesoscopy,” in IEEE Transactions on Industrial Informatics, vol. 11, no. 3, pp. 642-649, June 2015, doi: 10.1109/TII.2015.2417676.

I worked as the Programmer to code the two new algorithms and four existing algorithms in MATLAB. I have also used GUIDE in MATLAB to design a simple user interface to make it easier for anyone to run tests on samples using the 6 algorithms simultaneously. I was also involved in preparing the manuscript and replying to reviewers for successful publication.