Improved krill group-based region growing algorithm for image segmentation

ABSTRACT Traditional image segment algorithms have some demerits: edge blur, discontinuous edge is susceptible to peripheral irrelevant information, Poor noise resistance. Therefore, we propose an improved krill group-based region growing algorithm for image segmentation in this paper. First, texture feature of the image is extracted by using Gabor filter. Then, we modify the krill group by the way that the population particle dynamics is divided into two categories: worse fitness value of particle and better fitness value of particle. The inertial weight of worse fitness value of particle is reset to zero, so as to eliminate the adverse effect of inertial weight on the current iteration of the algorithm. The inertial weight of better fitness value particle remains unchanged. The step size scaling factor in the algorithm is treated with non-linear declination to further improve the global exploring ability of the algorithm. Finally, the improved kill group optimisation is used to search for the optimal solution, and the segmentation results are finally obtained. The experimental results show that the proposed image segmentation model can not only improve the segmentation accuracy and reduce the number of iterations but also suppress the background. The contour positioning of the object area is effective.