Comparison of Two Different Distance Functions of Image Retrieval for Detecting Species of Microscopic Fungi in Medical Mycology Laboratory

Abstract

In medicine, fungi are called as human pathogens. Physiological and immunological characteristics of fungi are not of great importance for detecting fungi species. Rather, the use of structural specifications of fungi in laboratory lam and fungal culture is the best way to detect them. In the method used in this paper, color histogram and horizontal, vertical and diagonal gradients of image edges are used to calculate attribute vector for images of different species of fungi. Then, we estimate the species of unknown fungus through calculating the distance of average attribute vector for different species and attribute vector of the query image and finding the shortest distance. Finaly , with comparison of the obtained results from two distance functions, Canberra & Bray Curtise, it was found that the Canberra dissimilarity is more efficient