2026-04-07T13:33:24Zhttps://ebiltegia.mondragon.edu/oai/request

oai:ebiltegia.mondragon.edu:20.500.11984/69382025-04-03T06:15:29Zcom_20.500.11984_473col_20.500.11984_478

Maestro-Watson, Daniel Balzategui, Julen Eciolaza, Luka Arana-Arexolaleiba, Nestor 2025-04-02T07:14:32Z 2025-04-02T07:14:32Z 2020 1560-229X https://katalogoa.mondragon.edu/janium-bin/janium_login_opac.pl?find&ficha_no=161733 https://hdl.handle.net/20.500.11984/6938 The purpose of this paper is to explore the use of fully convolutional neural networks (FCN) to perform a semantic segmentation of deflectometric recordings for quality control of reflective surfaces. The proposed method relies on a U-net network to identify the location and boundaries of the object and the possible defective areas present on it by performing a pixel-wise classification based on local curvatures and data modulation. Experiments were performed on a real industrial problem using four variations of the architecture. The results demonstrate that the method combining geometric and photometric information enables the identification of a wider variety of shape and texture imperfections, with the resulting segmentations closely correlated with the visual impact of the defects. In addition, several suggestions are presented for near-term industrial utilization. eng © 2020 Society of Photo-Optical Instrumentation Engineers Specular surfaces Defect detection Deflectometry Artificial Neural Networks Deflectometric data segmentation for surface inspection: a fully convolutional neural network approach http://purl.org/coar/resource_type/c_6501