Evaluation and comparison of two fuzzy classifiers for multi-spectral imagery analysis

Two fuzzy classifiers were evaluated and compared in this study. A parameter, fuzzy classification accuracy, was proposed in order to evaluate the two classifiers. The experimental results indicate that the two fuzzy classifiers are potential algorithms for classifying mixed pixels in multispectral images, and the neural network classifier trained with the data containing mixed pixels performed better in classification quality for mixed pixels, compared to the posterior probability classifier and the neural network classifier trained with the data containing converted pure pixels from mixed pixels.