White light and x-ray digital speckle photography

Sammanfattning: The goals of this work were not only to develop new measuring techniques, but also to validate their performance, in terms of accuracy, spatial resolution, etc., by performing carefully chosen experiments. The measuring techniques were also to be applicable to industrial problems in production, control or product development. The common feature of all the techniques presented is their use of digital speckle photography (DSP). DSP is an image processing technique that calculates the displacement of a characteristic pattern, which somehow is attached or acts as if it was attached to an object’s surface, by using digital correlation. The digital correlation algorithm is robust and accurate. It allows measurements of small deformations with standard deviations as low as 1% of the pixel size of the detector used. A system built for the measurement of 3-D displacement fields and object shapes is presented. The so-called stereo-DSP system uses two CCD-cameras in a stereovision set-up that simultaneously captures images. The physical model that describes the stereovision set-up is comprehensive. It takes the effects of small misalignments and lens distortions into account. Therefore, the main sources of error in the measurements are random errors from the correlation algorithm. A new method to measure internal transient deformations in solids by using flashed x-ray radiation is presented. This technique measures the deformation of an internal plane ‘seeded’ with x-ray absorbent particles. Experiments, with one or two flash x-ray sources, were conducted to capture 2-D or 3-D displacements in specimens made out of polyester or cement. The x- ray DSP technique allowed the examination of what happens inside the material during impact and gives interesting data about the material flow field in a body during failure. The concept of temporal digital speckle photography is introduced for shape measuring purposes. A whole sequence of independent speckle patterns is projected onto the object, and with the aid of calibration, it turns out that the surface profile is found from a 1-D correlation in depth. The advantage of temporal DSP is that it allows higher spatial resolution in the measurements compared with earlier techniques that use projected random patterns. It is commonly known that a sheet of paper assumes a cylindrical shape when it is subjected to humidity changes, a phenomenon often referred to as curl. The parameters that are of interest are the radius of curvature of the paper and the orientation of the curl axis. We propose a method that combines optical shape measurement with least squares modelling of a half-cylinder to estimate curl in paper. The least-squares model is nonlinear, and an iterative technique based on the Gauss-Newton algorithm is used in the fitting process.