Método de processamento digital de imagens para inferência da qualidade de materiais preparados com fibras vegetais

Abstract

This Dissertation presents a method development based on image processing techniques, which enables the verification and quality analysis of organic fibers mixture in thermoplastic starch (TPS). In order to do this analysis, we implemented a Python programming language module, which is able to compute invariant moments of Hu and other statistical calculations to quantify homogeneity, we applied this calculations to a set of high resolution tomography images of materials resulting from the mixture of thermoplastic starch and fibers. TPS based compositions, have recently received much attention, from industry and academia, for being one of the most economical options to produce biodegradable plastics. Although promising, there are two major limitations to the use of TPS: their mechanical fragility and sensibility to water. A solution to this problem is to mix TPS with vegetable fibers materials, maintaining its biodegradability, thus offering greater mechanic and humidity resistance. Nowadays there are methods techniques and equipment used to check mechanical resistance, water absorption and crystallinity of biodegradable samples. However, there is a lack of methods used specifically for the measurement of organic fibers homogeneity distribution in TPS, therefore assisting the production quality verification of those samples. As a result of this work, we obtained a method which infers the manufacturing process quality of new biodegradable materials, produced from the mixture of vegetable fibers and TPS, based on the homogeneity and the roughness degree of the samples produced.