MatrixCraCS: Automated tracking of matrix crack development in GFRP laminates undergoing large tensile strains

A novel image processing method for tracking development of matrix cracks in glass-fibre reinforced polymer (GFRP) laminates is presented. Images are acquired in a quasi-static tension test using transillumination white light imaging. These images show changes occurring due to developing damage. The method, called MatrixCraCS, tracks and quantifies matrix cracks while compensating for large specimen deformation and stitching threads, which obscure crack development and seriously impact crack detection. Matrix cracks are tracked via a sequential image difference and morphological filtering is used to detect cracks. MatrixCraCS is verified against synthetic images and shows an error of less than 7% in the quantified length of cracks and 0% detection failures. MatrixCraCS is validated against experimental results for a [±452°902°]S GFRP laminate, where microscopic analysis shows 5.92% error in crack detection. Error relative to manual crack tracking are quantified for detection failures and false positives as 3.3% and 1.57%, respectively. It is found that the method accurately quantifies cracks in GFRP laminates subject to large deformation and is robust against noise.