Digital description and recognition of pollen granules with invariant moments

Abstract

Pollen granules are one of the most stable microstructures of herbal flowers, and their ektexines possess strong anti-acid, anti-base and anti-biolysis properties. Therefore, the microstructures of pollen granules are not destroyed during storage, manufacturing and the production of different preparations. The shapes, sizes and colors of pollen granules are different in different families, genera and species of herbal flowers. These characteristics make the pollen granules of herbal flowers ideal for microscopic identification. To facilitate automatic recognition and improve the identification of Chinese herbal flowers, 1320 microscopic images of pollen granules from 13 different species were described and identified by pseudo-Jacobi Fourier moments in this paper. The experimental results showed that the moments can digitally describe and successfully identify the pollen granules, and the average identification rate was 90.2%. These results pave the way for the digitalization and automatic recognition of the microstructures of Chinese herbs.