ABSTRACT
This work evaluates the performance of edge-detection algorithms to determine the sample geometry with high spatial accuracy from low-resolution MR images. In particular, we show that by applying such numerical methods it is possible to reconstruct the internal and external contours of the object with a spatial precision that surpasses the nominal spatial resolution of the image by more than one order of magnitude. Special attention is paid to find the spatial resolution and signal-to-noise ratio required by the described numerical methodology to achieve a desired spatial accuracy. Finally, we discuss the potential application of this image processing approach for in-line quality control of extruded rubber materials, where micrometer spatial precision has to be achieved from images measured in short experimental times. The results presented here prove that the sensitivity of mobile MRI sensors is enough to achieve the spatial accuracy required to proof check the production of extruded rubber fittings in acceptable experimental times.
