Skip to Main Content
This paper is a progress report on a continuing project aimed at the eventual realization of low-cost, automatic systems for the recognition and counting of both ancient and live pollen. A previous paper has reported on the classification of optical microscope images of pollen grains using Gabor transforms and shape described by moment invariants. This work has since been extended by the introduction of a range of additional texture measures including grey-level cooccurrence matrices (GLCM), laws features and a wavelet decomposition. The additional texture measures have improved the pollen recognition rate of the system. This paper primarily reports on the development and evaluation of image capture and segmentation schemes. Further system developments in progress are briefly reported.
Date of Conference: 15-17 Sept. 2005