A Munsell colour-based approach for soil classification using Fuzzy Logic and Artificial Neural Networks

Abstract

Munsell soil-colour charts are widely used for soil classification. These charts contain 238 standardised colours in small rectangular chips arranged in seven charts and encoded in the Munsell system. Each chart uses three coordinates well correlated with the visual colour attributes: hue, value and chroma. The colour of a soil sample is commonly estimated by visual comparison between the actual soil colour and the Munsell chips, looking for the closest one and taking its Munsell notation. Consequently, the visual determination of soil colour with Munsell charts is a difficult task due to the subjectivity of the observer to match the colour of a soil sample with a single standard Munsell chip. For this reason, to avoid misclassification caused by subjective coincidence, we propose an intelligent method to provide the closest values of the Munsell chips to an unknown colour of a soil sample by using artificial neural networks and fuzzy logic.