Abstract
Color image segmentation of fishes with complex background in water considered a big challenge. In this paper five segmentation methods for fishes are discussed; they are Grabcut algorithm, Otsu thresholding method, Edge detection technique, Mean-shift method and Region-Growing algorithm. However, most of them are manually segmentation methods or require a white or uniform background. In order to evaluate the segmentation methods, they were tested using a new dataset which contains about 270 fish species from natural scenes. The results revealed that the Grabcut Algorithm has achieved a very good results comparing with other methods.