Deformation image generation for testing a strain measurement algorithm

• Published in: Optical Engineering 2008-10, Vol.47 (10), p.107202-1072013
• Main Authors: Koljonen, Janne, Alander, Jarmo T
• Format: Article
• Language: English
• Subjects: algorithm testing; Algorithms; computer vision; Correlation analysis; Deformation; Extensometers; Heuristic; Noise; pattern recognition; Strain; strain analysis; test image generation
• ISSN: 0091-3286; 1560-2303
• DOI: 10.1117/1.2993319

An optical extensometer was tested using artificially deformed images with a known strain field. A real image series from a tensile test was used to obtain realistic deformation parameters, including spatial and temporal strain characteristics, changes in tonal pixel properties due to deformation, and the effect of nonuniform illumination. These parameters are used to artificially deform a real image taken from an object with a random speckle pattern. The signal-to-noise ratio of the resulting artificially deformed images is varied by applying a blurring pillbox filter and additive Gaussian noise to them. The optical extensometer uses digital image correlation to track homologous points of the object, and further to measure strains. The strain measurement algorithm includes a heuristic to dynamically control the template size in image correlation. Furthermore, several other methods to improve the accuracy-complexity ratio of the algorithm exist. The effects of different parameters and heuristics on the accuracy of the algorithm as well as its robustness against blur and noise are studied. Results show that the proposed test method is practical, and the heuristics improve the accuracy and robustness of the algorithm.