A camera-based strain measurement technique for elastomer tensile testing

The present work investigates the precision characteristics of two optical strain measurement techniques applied to elastomers subjected to large deformations. The measurement approach is based on generating intensity profiles by using several horizontal image lines in the region of an optical marker to be detected. For detectability and accuracy reasons, these lines are combined using a rank (maximum) value filter and a moving average filter, while the calculation of the marker centers is carried out either geo- or gravimetrically. Based on simulated profiles, the first part of this work investigates the influence of method related parameters on the measurement precision obtained, expressed in terms of the root mean square error (rms). Further, it establishes model relations between most important image/profile parameters and rms. In the second part, experimental image data obtained during tensile testing of an elastomer sample is analyzed by i) applying the strain measurement technique, ii) determining experimental rms-values and iii) discussing them in comparison to values predicted by the rms-model of part one. It was found that, for the gravimetrical center calculation, rms strongly depends on the number of profile pixels which is caused by image noise. In the present approach, image noise was reduced by multiple image line fusion, which can be assumed to be in terms of computation effort more effective than averaging multiple images. The developed rms-models were found to represent the strain dependent decrease of accuracy efficiently up to high strains (e.g. 900%) under practical conditions. To obtain optimal measurement precision with the presented methods in practice, appropriate low marker detection threshold intensities of about 0.3g(s) (g(s) -signal intensity) and application of a single application cycle of the moving average filter were proved to yield optimal results. At high strains, the application of the rank filter in combination with a geometric center calculation results in best measurement precision, while the differences to the gravimetric method are less but its trend is comparable to the simulation.