Classification of Leaves Epidermis Microphotographies Using Texture Features
Elio Ramos, Denny S. Fernandez del Viso
Last modified: 2008-09-13
We present the results of a Gray Level Co-ocurrence Matrix (GLCM) analysis for a sample of leaf epidermis images. The leaves were collected from a dry forest in Mona Island, which is located between Dominican Republic and Puerto Rico. For each image seven GLCM textural features were calculated namely the energy, contrast, correlation, inverse differences, entropy, absolute value, and homogeneity. This parameters were complemented with several first order descriptive statistics of the grey levels like the mean, standard deviation, and the corresponding skewness and kurtosis. From the calculated statistics a features matrix for the eleven parameters was obtained. From the features matrix a training set and a test set was randomly generated and a minimum distance classifier was applied. Preliminary results indicate that the method is able to classify the test set with an accuracy of 92%.