International Research Journal of Education and Technology

A threshold optimization technique using the nature inspired Firefly algorithm for color image segmentation is proposed in this paper. Initial seeds are selected by segregating the image into several blocks. The proposed algorithm comprises two stages: During the first stage, a region growing algorithm is proposed in which regions are grown based on a particular threshold. Regions are grown based on the intensity. In the next phase, the threshold used for growing regions is optimized using firefly algorithm. Texture characterization is made using local entropy computation. . This obtained texture characteristic and the region growth map of the fully grown regions are considered for merging procedure. The performance of the proposed segmentation algorithm is evaluated using the images taken from the Berkeley dataset.

Keywords: Region growing, Metaheuristic, Optimization, Firefly, Texture Character