Image Threshold Segmentation Based on BEMD and Genetic Algorithm

Abstract

For image threshold segmentation based on genetic algorithm, it is easy for the population to become precocious and occasionally fall into the local optimal solution, the segmenting targets thus are lost and the effect is not well when the complex image is segmented. In this paper, we propose a new threshold segmentation method based on combining bidimensional empirical mode decomposition (BEMD) and genetic algorithm (GA). First, the morphological difference method(MDM) is used to remove the background of an image, and then the BEMD algorithm is used to decompose the processed image to obtain sub-images, the high-frequency sub-images of the targets are retained, the gray level transformation(GLT) is performed, and the enhanced image is obtained by adding transformed high-frequency sub-images for the fusion. Finally, we use GA to segment the enhanced image. Experiment results show that the proposed algorithm can effectively segment images.