Abstract: Aiming at the threshold selectiong and slow speed in image segmentation, the paper suggests a effective method based on Fisher criterion and grey correlation analysis.In the method, a 2D entropy model is designed to locate the best threshold value via Fisher criterion.Additionally, Particle Swarm Optimization (PSO) is introduced to speed up the segmentation procedure.Finally, the segmented image was processed by grey correlation analysis according to image's character.Some exeperimental results indicate that the new algorithm ignores the disturbance of inherent speckle in image and get accurate image segmentation.