Software development of a computer aided method to capture and determine crack lengths of fracture mechanics test specimens

Bc. Michal Dostálek

Master's thesis

Abstract:

Cracks are important indicators reflecting the material properties and their effects on the resulting constructions. This thesis presents an automatic crack capturing, detection and classification methodology for a crack test stand. The algorithm is written in C/C++. The complete graphical user interface was created as well. The test samples are captured and stored with use of complementary metaloxide-semiconductor industrial cameras. The captured samples are converted into a grey-scale images and areas with possible cracks are found by utilizing morphological image processing techniques and thresholding operations. Cracks are classified and measured with the help of a scaling factor and mathematical elementary functions. The classification methods were tested on real samples with success over 95%. The experimental results revealed the influence of parameter settings and filter selection and also proved that the proposed approach is effective for automatic crack detection and classification. The testing algorithm is universal and usable for all sorts of materials from plastics, metals to wood and composites. The program was developed and tested using also fibre materials with epoxide core. Other samples were e.g materials which were tested for their suitability for use in industry. Part of this thesis also covers the basic principles used in image processing.

Abstract:

Cracks are important indicators reflecting the material properties and their effects on the resulting constructions. This thesis presents an automatic crack capturing, detection and classification methodology for a crack test stand. The algorithm is written in C/C++. The complete graphical user interface was created as well. The test samples are captured and stored with use of complementary metaloxide-semiconductor industrial cameras. The captured samples are converted into a grey-scale images and areas with possible cracks are found by utilizing morphological image processing techniques and thresholding operations. Cracks are classified and measured with the help of a scaling factor and mathematical elementary functions. The classification methods were tested on real samples with success over 95%. The experimental results revealed the influence of parameter settings and filter selection and also proved that the proposed approach is effective for automatic crack detection and classification. The testing algorithm is universal and usable for all sorts of materials from plastics, metals to wood and composites. The program was developed and tested using also fibre materials with epoxide core. Other samples were e.g materials which were tested for their suitability for use in industry. Part of this thesis also covers the basic principles used in image processing.

Keywords

Computer vision, crack classification, fracture mechanics, crack detection

Language used: English

Date on which the thesis was submitted / produced: 3. 1. 2017

Thesis defence

• Date of defence: 2. 2. 2017
• Supervisor: prof. RNDr. Stefan Bischoff

Technical University of Liberec

Faculty of Mechatronics, Informatics and Interdisciplinary Studies

Master programme / field:
Electrical Engineering and Informatics / Mechatronics