Spatial resolution, noise, and information delocalization of input data for inverse methods

Full-field measurements are most advantageous when information from the experiment is specific to the location on the sample. The methods inherently provide redundant data, with each measurement, so that the precision of information extracted from the data can be quite good, due to averaging of the errors on individual data points. The need to quantitatively combine redundant but heterogeneous data leads naturally to inverse problem approaches. The accuracy of the extracted information, indeed the feasibility of the method, are affected by systematic errors throughout the process. This paper considers the effect of inaccuracy in the spatial location associated with the data on the final result, with the specific example of deterioration of spatial information content due to data analysis in full-field strain measurement by digital image correlation. Data smoothing strategies that minimize information delocalization will be proposed. Conclusions will be more generally applicable to the application of other full-field measurements with inverse methods and the presentation of full-field measurement results in general.