文摘
An artificial neural network is used for the classification of minerals. Optical data using thin sections is acquired using the rotating polarizing microscope stage, which extracts a basic set of seven primary images during each sampling. A selected set of parameters based on hue, saturation, intensity and texture measurements are extracted from the segmented minerals within each data set. Parameters such as pleochroism, plane light hue, and gradient homogeneity were a few that proved to yield class-discriminating properties. Texture parameters are shown to have the ability to classify colourless minerals. The neural network is trained on manually classified mineral samples. The most successful artificial network to date is a three-layer feed forward network using generalized delta error correction. The network uses 27 distinct input parameters to classify 10 different minerals. Testing the network on previously unseen mineral samples yielded successful results as high as 93 % .