RESUMO
In this work, NIH ImageJ plugins for extended depth-from-focus reconstructions (EDFR) based on spatial domain operations were compared and tested for usage optimization. Also, some preprocessing solutions for light microscopy image stacks were evaluated, suggesting a general routine for the ImageJ user to get reliable elevation maps from grayscale image stacks. Two reflected light microscope image stacks were used to test the EDFR plugins: one bright-field image stack for the fracture of carbon-epoxy composite and its darkfield corresponding stack at same (x,y,z) spatial coordinates. Image quality analysis consisted of the comparison of signal-to-noise ratio and resolution parameters with the consistence of elevation maps, based on roughness and fractal measurements. Darkfield illumination contributed to enhance the homogeneity of images in stack and resulting height maps, reducing the influence of digital image processing choices on the dispersion of topographic measurements. The subtract background filter, as a preprocessing tool, contributed to produce sharper focused images. In general, the increasing of kernel size for EDFR spatial domain-based solutions will produce smooth height maps. Finally, this work has the main objective to establish suitable guidelines to generate elevation maps by light microscopy.
RESUMO
Uncoated fracture surfaces of carbon-epoxy composites are investigated using a variable-pressure environmental scanning electron microscope (VP-ESEM), under optimized conditions for topographic description, image quality and sample preservation. Always using freeware or open source programs, parameters for low-voltage and low vacuum are stipulated with the support of Monte Carlo simulations combined to topographic measurements, tailoring the VP-ESEM setup for visualization of fine relief details. Based on topographic information from atomic force microscope (AFM) images, finest fracture steps were measured. These were the references to optimize and define boundaries for applied beam voltages and chamber pressures, restricted by the beam penetration depth and gas-electron interactions, guided by Monte Carlo simulations and signal-to-noise measurements. For VP mode, ideal chamber pressure was found around 30-40Pa at 3keV beam voltage and 6mm working distance. Lower pressures will cause noise due to electron charging and gas excess provokes resolution degradation and noise due to positive charging and electron beam scattering, raising the skirt radius. When a larger working distance is necessary, it can be compensated by adjusting the detector bias and the probe current, or even lowering chamber pressure, but the signal-to-noise ratio will certainly change. Monte Carlo simulations provided a good approach to optimize imaging conditions under low vacuum and low voltage for fractographic analysis of carbon-epoxy composites.