Content-based Image Retrieval of Environmental Microorganisms Using Double-stage Optimisation-based Fusion

Abstract

Environmental Microorganisms (EMs) are very tiny living beings which impact the entire biosphere by their environmental functions. Traditionally, a lot of manual efforts through morphological analysis using microscopes have been put on looking for EMs. However, these methods are time-consuming and laborious. To overcome this, we develop a Contentbased Image Retrieval (CBIR) system for the EM image retrieval task within a Doublestage Optimisation-based Fusion framework. In the first stage, in order to effectively use the colour information of EM images, a Multiple Colour Channel Fusion (MCCR) method based on a Particle Swarm Optimisation (PSO) is developed to search for similar database images to a query image using local features. In the second stage, in order to enhance the retrieval performance of the first stage, a retrieval method based on Immune Evolutionary Particle Swarm Optimisation -Shuffled Frog Leaping Algorithm (IEPSO-SFLA) is devised to further combine global features. Finally, the experimental result shows that our doublestage fusion method obtains a mean average precision of 35.87% for 21 classes of EMs, which is superior to the existing methods.