Recognition of materials and damage on historical buildings using digital image classification

Abstract

Nowadays, techniques in digital image processing make it possible to detect damage, such as moisture or biological changes, on the surfaces of historical buildings. Digital classification techniques can be used to identify damages in construction materials in a non-destructive way. In this study, we evaluate the application of the object-oriented classification technique using photographs taken with a Fujifilm IS-Pro digital single lens reflex camera and the integration of the classified images in a three-dimensional model obtained through terrestrial laser scanning data in order to detect and locate damage affecting biocalcarenite stone employed in the construction of the Santa Marina Church (Córdoba, Spain). The Fujifilm IS-Pro camera captures spectral information in an extra-visible range, generating a wide spectral image with wavelengths ranging from ultraviolet to infrared. Techniques of object-oriented classification were applied, taking into account the shapes, textures, background information and spectral information in the image. This type of classification requires prior segmentation, defined as the search for homogeneous regions in an image. The second step is the classification process of these regions based on examples. The output data were classified according to the kind of damage that affects the biocalcarenite stone, reaching an overall classification accuracy of 92% and an excellent kappa statistic (85.7%). We have shown that multispectral classification with visible and near-infrared bands increased the degree of recognition among different damages. Post-analysis of these data integrated in a three-dimensional model allows us to obtain thematic maps with the size and position of the damage.

Non-Destructive Techniques Applied to Monumental Stone Conservation

“…They obtained a good automatic classification of granite ashlar with remaining lime/mortar and granite ashlars affected by high moisture content. Meroño et al [20] applied a combination of 3D model from a terrestrial laser scanner with multispectral images to the study of Santa Marina de Aguas church (Cordoba, Spain). They arrive to separate different areas with stone, plaster, wood and different kinds of degradations.…”

Sensors for Structural Health Monitoring of Agricultural Structures

“…However, the utility of sensors is impacted by the critical technical and economic constraints that impede the widespread adoption of sensors; this means that despite the clear evidence in support of the installation of sensors to mitigate material failure in agricultural systems, there are divergent views on the need to install sensors in buildings. For example, installing sensors can be contested in regions with a large stock of historical building structures [ 51 , 52 , 53 ]. In such cases, strict adherence to building standards does not offset the risk of structural failure.…”

Machine Learning Models for Cultural Heritage Image Classification: Comparison Based on Attribute Selection

“…In [7], the k-nearest neighbor algorithm was used for classification and detection of alterations on historical buildings. The method proved to be efficient with the obtained classification accuracy of 92%.…”