Identificação de regiões com desplacamento de cerâmica em paredes utilizando análise de dados funcionais

Abstract

Ceramic wall claddings may conceal serious structural defects that are often detectable only through destructive methods or percussion testing, which are impractical for large façades. Infrared thermography has emerged as a non-invasive alternative, capable of generating thermal images that reveal temperature variations associated with mortar detachment. In this context, this study proposes an automatic classifier that identifies regions with cladding defects based on daily temperature curves extracted from thermographic images segmented into superpixels and converted into RGB values. To achieve this, the color intensities observed throughout each day for each superpixel were smoothed using Fourier basis functions, and the resulting coefficients were employed as explanatory variables in classification models. The performance of the best fitted model proved to be satisfactory.