Image Segmentation of Drosophila's Compound Eyes via Two-Dimensional OTSU Thresholding on the Basis of AGA

Abstract

Image segmentation is a key step in the process of realizing automatic diagnosis of the drosophila’s pathologic compound eyes. This article uses the chromatic aberration variable Cr ,extracted from the color images of drosophila’s compound eyes, to establish a two-dimensional histogram, with which and according to the principle of OTSU thresholding segmentation, image segmentation of the drosophila’s compound eyes has been realized. Based on massive computing of the two-dimensional OTSU thresholding, the Adaptive Genetic Algorithm (AGA), including robustness and parallelity, is then used to search for the best non-linear vector of two-dimensional threshold. A number of experiments are made and the application result proves that the two-dimensional OTSU thresholding method proposed in this paper takes a good effect.