Research And Application Of Improved Continuous Ant Colony Algorithm Based On Multi-strategy

The ant colony optimizer(ACOR)is a swarm intelligent algorithm based on ant foraging behavior.It can solve problems in the continuous domain and has been widely used in many fields.Due to the complexity and diversity of practical problems,ACOR still has a slow convergence rate and is easy to fall into local optimization.Considering the influence of population diversity and balance on the convergence rate and the ability to escape from local optimal,this paper introduces a variety of optimization strategies into ACOR and proposes three improved ACOR.In order to verify the performance of the improved algorithm,IEEE CEC2017 and IEEE CEC2014 function sets were used in the experiment for comparison experiments.Through the experimental results and analysis of function set,the performance of the three proposed algorithms is verified.Furthermore,based on three improved ant colony algorithms,three multi-level image segmentation methods were constructed by fusion of two-dimensional(2D)histogram,two-dimensional Kapur entropy and non-local mean image segmentation techniques,and were successfully applied to melanoma image segmentation,COVID-19 X-ray image segmentation and breast cancer image segmentation.The main contributions of this paper are as follows:1.To improve the performance of the algorithm by improving the quality of the algorithm population,a hybrid mechanism based ant colony optimizer named LACOR was proposed.In this algorithm,the sine cosine strategy(SC),disperse foraging strategy(DFS),and specular reflection learning strategy(SRL)are introduced into the ant colony optimizer,which enhances the ability of the algorithm to search for the optimal global value and avoid falling into the local optimal.To verify LACOR’s performance,a series of comparative experiments were designed based on the IEEE CEC 2014 function set.Experimental results show that the algorithm has excellent convergence speed and search ability.Furthermore,a multi-level image segmentation model based on LACOR was proposed and applied to pathological images of melanoma.The experimental results show that LACOR has a good segmentation effect in melanoma pathological image segmentation.2.To improve the performance of the algorithm by improving the balance of ACOR’s exploration and exploitation capabilities,this paper introduced the directional crossover strategy(DX)and the directional mutation strategy(DM)into the ant colony optimizer,named XMACO.Among them,the DX strategy improves the ability of algorithm exploitation and the speed of algorithm convergence.DM strategy improves algorithm exploration ability.To verify the algorithm performance of XMACO,a series of function experiments are designed based on IEEE CEC2014 and IEEE CEC2017 function sets.The experimental results show that XMACO has a faster convergence speed and better searching ability.Furthermore,this paper designs a multi-level image segmentation model based on XMACO and applies it to COVID-19 X-ray image segmentation.Through the analysis of experimental results,the image segmentation model constructed in this paper shows more stable and excellent results for COVID-19 X-ray image.3.To improve the algorithm performance by increasing the diversity of the population,this paper proposes a novel ant colony optimizer(EACO)based on the Lévy mutation mechanism.Lévy mutation mechanism has the characteristics of random step search,which can greatly improve the distance between population individuals,enrich population diversity,and enhance the ability of local search.In the verification experiment,a series of experiments were designed based on the IEEE CEC2017 function set to verify the performance of EACO.The experimental results show that EACO has better convergence,accuracy,and the ability to avoid falling into local optimum than its counterparts.Furthermore,EACO combined 2D Kapur entropy and other strategies to construct a multi-level image segmentation model,which was applied to the breast cancer pathological image segmentation experiment.The experimental results show that the segmentation model based on EACO has better capability of breast cancer image segmentation.