Een zelfcorrigerend algoritme voor nauwkeurige rekmetingen met behulp van globale digitale beeldcorrelatie

摘要

Digital image correlation is a optical-numeric measuring technique which makes it possible to measure movements and complex deformations over the surface of an object. The major advantages of this technique are that the measurement happens without contact, yields on the entire surface are determined (so-called 'full-field') and it is applicable to any material under arbitrary loading conditions. With these properties DIC differs fundamentally from more traditional strain measurements such as strain gauges.This method (using the subset method) has already proven its usefulness in many applications, but has still some important drawbacks. The goal of the proposed research is to counter these disadvantages by implementing and validating a new DIC procedure that uses concepts derived from the adaptive finite element method and the traditional image processing. The new algorithm no longer makes use of an independent block method (subsets) to trace corresponding pixels between the reference and distorted images, but uses an adaptive finite element mesh which is automatically generated on the images. The relationship between adjacent elements guarantees the continuity of the displacements on the surface and with higher order elements also highly heterogeneous deformations at small strains may be registered. This research will eventually lead to a fully automatic, self-correcting procedure to perform displacement and strain measurements. Finally, by the use of a mesh in the algorithm, the correspondence with a numerical model suddenly becomes much simpler.