Digital image processing: application to automatic classification of tile panel pathology

Abstract

Digital image processing (DIP) is a technique to extract information from images with application on several areas of engineering and architecture, which can be used as a tool to map pathologies on tile panels. This paper presents the processing tasks that must be performed on a tile panel image, to perform corrections on the image, to delineate the pathologies and to classify it´s typology, in order to produce the graphical registration of anomalies or monitor multitemporal changes. Several types of anomalies could be identified like, tile rotation, glaze lack, glaze pores, craquelé, open joints, repaired joints, etc.. The images for these applications are digital photographs and the acquisition must follow some specifications to obtain accurate results. Depending of the characteristics of the tile panels (like repetitive patterned or historical) different strategies must be implemented for automatic pattern delineation and classification. Object oriented image analysis (OBIA) methodologies and a picture element (pixel) based system, named GeMAPA, developed at LNEC’ LabImagem, are presented. These studies are developed at LNEC´s Digital Image Processing Laboratory (LabImagem) under project «Numerical Images: applications to engineering» of LNEC’s Research and Innovation Plan 2012-2020.