A Method for Rapid Localisation of Hydrocarbon Compounds on Surfaces Using Chemical Imaging and Back-Projection

摘要: Discharges of unwanted chemicals into nature is a large and complex problem. Easy, fast reliable methods for detecting these contaminants can be viewed as first step towards solution. Using hyperspectral camera operating in the range 923-1665~nm, chemometric such discriminant partial least squares regression (DPLSR) k-nearest neighbours (k-NN) well various forms preprocessing, models hydrocarbons on surfaces soil, sand, stone, humus vegetation were developed. These chosen examples naturally occurring where locating interesting useful. Paraffin used hydrocarbon. For some DPLSR models, it proved to great importance error whether or not spectra normalised. This probably due uneven that penetration would differ from place place. As usually will smooth, natural assume this have an even greater impact backgrounds than laboratory environment. Smoothing data also improved many models. A simpler model, only 16 148 available wavelengths used, turned out give surprisingly good results most surfaces. model was less selective detect other paraffin. recognising C-H-bond. The percentage varied widely between different strengths weaknesses. Water caused false positives time series images soil with without paraffin taken over year, showed based week had low all images. However, indicate could soil. able distinguish n-hexane n-heptane substantiates possible create distinguishes several hydrocarbons.