Clustering and discernment of Algerian bee pollen using an image analysis system

Abstract

In this paper, we suggest a framework for multi-focal image classification and identification, the methodology being demonstrated on microscope pollen images (image processing and classification techniques). The framework is intended to be generic and based on a brute force-like approach aimed to be efficient not only on any kind, and any number, of pollen images (regardless of the pollen type), but also on any kind of multi-focal images. Microscope images information obtained from bee pollen samples (72 samples) of different floral origin from various Algerian counties were used to formulate a method for rapid classification using Hierarchical Cluster Analysis (HCA). Both stages of the framework’s pipeline are planned to be used in an automated fashion. First, the optimum focus is chosen using the absolute gradient method. Then, pollen grains are collected using a coarse-to-fine method involving both clustering and morphological techniques. Finally, features are extracted and selected using a generalized method, and their classification is checked with using HCA. Our findings indicate that HCA meets the demands for automatic pollen detection making it an alternative method for research concerning pollen.