Image Analysis Protocol for Pollen Viability Identification on Selected Genotypes of Rice (Oryza sativa L.)

Abstract

Pollen viability plays a key role in the efficient sexual reproduction of plants. However, estimating pollen viability is a labor-intensive and time-consuming task for researchers. Moreover, it is difficult to evaluate a large dataset of microscopic images as most of it needs to be meticulously examined. Existing methods have been developed to efficiently estimate total pollen count but their success in computing viability percentage is rather limited since they do not integrate viability testing techniques such as pollen staining. Some of these existing methods greatly depend on features such as the diameter of the pollen grain which requires a manual standardization to come up with two exclusive distributions for viable and nonviable pollen grains. An image analysis protocol was presented to compute the viability percentage on selected genotypes of rice (Oryza sativa L.). The effectiveness of the segmentation was evaluated in terms of accuracy in classification of the pollen grains. Results suggest the proposed method can be an alternative to the traditional labor-intensive manual counting.