Mapping asbestos-cement roofing with the use of APEX hyperspectral airborne imagery: Karpacz area, Poland – a case study

Krówczyńska, Małgorzata; Wilk, Ewa; Pabjanek, Piotr; Zagajewski, Bogdan; Meuleman, Koen

doi:10.1515/mgrsd-2016-0007

Cited by 13 publications

(13 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…With the spatial resolution of MIVIS imagery at 3-4 m, distinguishing buildings with asbestos-cement roofing in the imagery is subject to great uncertainty, so it is difficult to refer to the results obtained by the authors. Fiumi et al [19] and Krówczyńska et al [18] present very high classification accuracy but include a small number of buildings with asbestos roofing. The average roof area was quite large and amounts to about 1200 m 2 [42].…”

Section: Discussionmentioning

confidence: 99%

Asbestos—Cement Roofing Identification Using Remote Sensing and Convolutional Neural Networks (CNNs)

et al. 2020

Self Cite

View full text Add to dashboard Cite

Due to the pathogenic nature of asbestos, a statutory ban on asbestos-containing products has been in place in Poland since 1997. In order to protect human health and the environment, it is crucial to estimate the quantity of asbestos–cement products in use. It has been evaluated that about 90% of them are roof coverings. Different methods are used to estimate the amount of asbestos–cement products, such as the use of indicators, field inventory, remote sensing data, and multi- and hyperspectral images; the latter are used for relatively small areas. Other methods are sought for the reliable estimation of the quantity of asbestos-containing products, as well as their spatial distribution. The objective of this paper is to present the use of convolutional neural networks for the identification of asbestos–cement roofing on aerial photographs in natural color (RGB) and color infrared (CIR) compositions. The study was conducted for the Chęciny commune. Aerial photographs, each with the spatial resolution of 25 cm in RGB and CIR compositions, were used, and field studies were conducted to verify data and to develop a database for Convolutional Neural Networks (CNNs) training. Network training was carried out using the TensorFlow and R-Keras libraries in the R programming environment. The classification was carried out using a convolutional neural network consisting of two convolutional blocks, a spatial dropout layer, and two blocks of fully connected perceptrons. Asbestos–cement roofing products were classified with the producer’s accuracy of 89% and overall accuracy of 87% and 89%, depending on the image composition used. Attempts have been made at the identification of asbestos–cement roofing. They focus primarily on the use of hyperspectral data and multispectral imagery. The following classification algorithms were usually employed: Spectral Angle Mapper, Support Vector Machine, object classification, Spectral Feature Fitting, and decision trees. Previous studies undertaken by other researchers showed that low spectral resolution only allowed for a rough classification of roofing materials. The use of one coherent method would allow data comparison between regions. Determining the amount of asbestos–cement products in use is important for assessing environmental exposure to asbestos fibres, determining patterns of disease, and ultimately modelling potential solutions to counteract threats.

show abstract

Section: Discussionmentioning

confidence: 99%

Asbestos—Cement Roofing Identification Using Remote Sensing and Convolutional Neural Networks (CNNs)

et al. 2020

Self Cite

View full text Add to dashboard Cite

show abstract

“…Barakat, D. et al (2017) developed a rule-based method and achieved 93 per cent OA in asbestos identification with WV-2 data. Krówczyńska, M. et al (2016) achieved 95 per cent OA with hyperspectral data, but the asbestos identification was biased by 38 per cent omission and 27 per cent commission errors.…”

Section: The Linear Type Of Discriminant Analysismentioning

confidence: 98%

“…The topic has its relevance from various perspectives: materials can be flammable (wooden, hay) or can be risk factors of 'carcinogenicity' (asbestos). In this study we focus on traditional roofing materials and asbestos (Cilia, C. et al 2015;Wilk, E. et al 2015;Gibril, M.B.A. et al 2016;Krówczyńska, M. et al 2016). Asbestos mapping has a wide literature, and several authors have attempted to identify this dangerous type of roofing.…”

Section: Introductionmentioning

confidence: 99%

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

et al. 2018

View full text Add to dashboard Cite

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.

show abstract

“…The wealth of spectral information available from advanced hyperspectral imaging instruments has opened new perspectives in many application domains. For example, monitoring of natural grasslands and the synanthropic patterns of invasive species [1], mapping of tree species in a complex mixed forest ecosystem [2] and analyzing the dominant minerals and rocks in a mountain region [3], mapping and estimation of the amounts of hazardous materials to human health [4] and urban land-cover mapping with high resolution images [5] are among others. Besides, retrieving the atmospheric condition parameters aid to the atmospheric correction and the atmosphere monitoring processes.…”

Section: Introductionmentioning

confidence: 99%

Estimation of AOD Under Uncertainty: An Approach for Hyperspectral Airborne Data

et al. 2018

View full text Add to dashboard Cite

Abstract:A key parameter for atmospheric correction (AC) is Aerosol Optical Depth (AOD), which is often estimated from sensor radiance (L rs,t (λ)). Noise, the dependency on surface type, viewing and illumination geometry cause uncertainty in AOD inference. We propose a method that determines pre-estimates of surface reflectance (ρ t,pre ) where effects associated with L rs,t (λ) are less influential. The method identifies pixels comprising pure materials from ρ t,pre . AOD values at the pure pixels are iteratively estimated using l 2 -norm optimization. Using the adjacency range function, the AOD is estimated at each pixel. We applied the method on Hyperspectral Mapper and Airborne Prism Experiment instruments for experiments on synthetic data and on real data. To simulate real imaging conditions, noise was added to the data. The estimation error of the AOD is minimized to 0.06-0.08 with a signal-to-reconstruction-error equal to 35 dB. We compared the proposed method with a dense dark vegetation (DDV)-based state-of-the-art method. This reference method, resulted in a larger variability in AOD estimates resulting in low signal-to-reconstruction-error between 5-10 dB. For per-pixel estimation of AOD, the performance of the reference method further degraded. We conclude that the proposed method is more precise than the DDV methods and can be extended to other AC parameters.

show abstract

Mapping asbestos-cement roofing with the use of APEX hyperspectral airborne imagery: Karpacz area, Poland – a case study

Cited by 13 publications

References 14 publications

Asbestos—Cement Roofing Identification Using Remote Sensing and Convolutional Neural Networks (CNNs)

Asbestos—Cement Roofing Identification Using Remote Sensing and Convolutional Neural Networks (CNNs)

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

Estimation of AOD Under Uncertainty: An Approach for Hyperspectral Airborne Data

Contact Info

Product

Resources

About