Pollen classification using brightness-based and shape-based descriptors

Pollen grain classification has recently received more attention from computer vision researchers. To distinguish among taxa, palynologist makes direct use of keys such as the size, exine structure and sculpture of the pollen grains. We propose a framework in which the pollen grains of each taxa are characterized using brightness and shape descriptors derived from their intensity images. These descriptors are associated to the ornamentation and morphology of the pollen grain. The method is statistically evaluated on preparations containing species of the Urticaceae family.