Estabelecimento de vínculo entre um sistema óptico e a presença ou ausência de degradação

Abstract

In this work, we present a study to identify degradation in an optical system that has been exposed to adverse environmental conditions, made from photographs taken outdoors of the vertical and horizontal bar target developed by the American Air Force in 1951 (USAF-1951). The photographs were taken without application of jamming and with the application of jamming to the USAF-1951 target, simulating degradation. The analysis and processing of photographs were made following adapted suggestions from the ISO-12233 standard and using the Modulation Transfer Function (MTF), which is a measure of capacity of an optical system of transferring the contrast of an object to an image in a certain resolution. From the results obtained from the MTF analysis of the photographs taken with and without applying interference in the target, were ploted the graphs with MTF curves. These MTF curves allowed us, through comparison, classify the photographs into the two classes of interest, with presence of degradation and absence of degradation. However, to improve the classification efficiency and increase its robustness, the MTF analysis results were also applied to machine learning algorithms, mean-based, K-nearest neighbors (KNN), centroid-based and neural networks. From the results obtained from all four algorithms was possible to classify the photographs as having presence of degradation and with no degradation, and it is important to highlight that the KNN algorithm had the best performance, obtaining accuracy results above 92%.