Feature Based Retrieval of Diatoms in an Image Database Using Decision Trees

A feature based retrieval scheme for microscopic images of diatoms in an image database is presented in this report. Diatoms are unicellular algae found in water and other places wherever there is humidity and enough light for photo synthesis. Several methods for feature extraction are described and experimental results on real diatom images are presented. The proposed feature based retrieval scheme is based on symmetry measures, geometric properties, moment invariants, Fourier descriptors and simple textural features. Based on this features the image database is divided into classes using a decision tree based classification approach. We have evaluated the discriminant power of the features and show experimental results on a diatom image database.