Identification of roofing materials with Discriminant Function Analysis and Random Forest classifiers on pan-sharpened WorldView-2 imagery – a comparison

Abstract

Identification of roofing material is an important issue in the urban environment due to hazardous and risky materials. We conducted an analysis with Discriminant Function Analysis (DFA) and Random Forest (RF) on WorldView-2 imagery. We applied a three-and a six-class approach (red tile, brown tile and asbestos; then dividing the data into shadowed and sunny roof parts). Furthermore, we applied pan-sharpening to the image. Our aim was to reveal the efficiency of the classifiers with a different number of classes and the efficiency of pan-sharpening. We found that all classifiers were efficient in roofing material identification with the classes involved, and the overall accuracy was above 85 per cent. The best results were gained by RF, both with three and with six classes; however, quadratic DFA was also successful in the classification of three classes. Usually, linear DFA performed the worst, but only relatively so, given that the result was 85 per cent. Asbestos was identified successfully with all classifiers. The results can be used by local authorities for roof mapping to build registers of buildings at risk.

A QGIS Tool for Automatically Identifying Asbestos Roofing

“…Roofing classification was performed through the digital processing of images obtained by the WorldView-3 sensor. Such images were pre-processed, so as to make them suitable for subsequent classification [20]. The outcomes, as achieved in an open context, underwent a validation process, which consisted of analyzing the confusion matrix, and subsequently assessing the classification errors.…”

“…The outcomes, as achieved in an open context, underwent a validation process, which consisted of analyzing the confusion matrix, and subsequently assessing the classification errors. Given that the most difficult task was testing the tool's impact in discerning asbestos and cement, the decision was made to only use structures with well-known roofing materials as the test buildings, according to recent surveys [18][19][20]. A second validation was carried out on the basis of a verification process, with data being collected at the Health Information System for Collective Prevention (SISPC) [32].…”

“…Some researchers used an object-based approach to extract information from the images and have proven that their processing methods have the potential to detect roof materials from the WorldView-2 images [18,19]. More recently, other authors have illustrated research carried out with discriminant function analysis (DFA) and random forest (RF) on WorldView-2 imagery [20]. Their aim was to reveal the efficiency of these classifiers, and the efficiency of pansharpening.…”

“…This approach makes it possible to filter the images without using, or training, object-oriented classification algorithms or neural networks [20]. Moreover, it makes it possible to reduce errors due to false positives that are represented by objects like asphalt on roads.…”

A review of climatic and vegetation surveys in urban environment with laser scanning: A literature-based analysis

“…Buildings can be discriminated from roads and pavements if we can assign a minimum height: a 3 m relative height combined with a vegetation index (e.g. normalized difference vegetation index, NDVI) gained from optical imaging is a clear difference (Abriha et al, 2018;Szabó et al, 2014). This method also helps to distinguish the low vegetation (i.e.…”