Análise das imagens tomográficas de corpos de prova compostos por agregados de BGS por meio do software ImageJ

Abstract

This study aims the comparation of indexes tested in laboratory and indexes obtained from the image processing of the software ImageJ. Compacted samples using basalt and riodacite were produced, representing the Rio Grande do Sul, where most part of the materials used in the pavement industry are composed by this materials. The images were generated by computed tomography and were treated and processed using the software mentioned, focusing on void indexes. Different calculation routines were made trying to clarify the structure of the samples in the images. Two algorithms were tested to denoise the images before processing into the ImageJ and four different thresholding algorithms were tested, improving the range of the study. The study shows that when comparing the indexes obtained in laboratory and by computed tomography, the best results were obtained when the routine was using the Non-Local Means denoising algorithms and then the Default thresholding algorithms. The variation of the results for both methods after the image processing was 25% at most, composing a difference, in values, from 0,11 to 0,14 in the sample of basalt. Finally, these results demonstrate the applicability of computed tomography tests, especially in cases where it is desired to maintain the internal structure of the sample.