Improved Swarm Intelligence Algorithm And Its Application In Segmentation Of Corn Disease Spot Images

In the real world,optimization problems in production and life practice are often complex.Although swarm intelligence algorithms can effectively solve such optimization problems,the traditional algorithms cannot meet practical needs when solving complex application problems with large amounts of data.Therefore,this paper studies three swarm intelligence algorithms,and proposes targeted optimization strategies to enhance the optimization performance of the algorithms,and uses them to deal with the problem of corn disease spot image segmentation.The main research content of this article is as follows:For the Harris hawk optimization algorithm,the Laplace crossover strategy is introduced to improve the local exploitation ability of the algorithm and to enhance the efficiency of the population search for the solution space.At the same time,a random replacement strategy is used to improve the quality of the optimal individual and guide the population to quickly search for the optimal solution.For the whale optimization algorithm,the individual disturbance strategy is proposed to realistically simulate the hunting behavior of whales and strengthen the population to jump out of the local optimal space.And the neighborhood mutation search mechanism is used to enhance the ability of whale individuals to search the surrounding space and improve the optimization accuracy.For the moth flame optimization algorithm,a combination of Cauchy and Gaussian mutation is applied to the population to reduce the probability of the algorithm falling into stagnation.To increase the chance of individuals moving closer to the optimal agent,a chemotaxis motion mechanism is introduced to further improve the algorithm.To verify the optimization performance of the improved algorithms,30 IEEE CEC2017 competition functions were selected in experiments to test the accuracy of the algorithms in searching for the optimal solution.The effectiveness of the optimization strategies,the stability of the algorithms and the performance advantages in swarm intelligence algorithms are explored through mechanism comparison,high-dimensional and comparison experiments with other algorithms,respectively.Experimental results show that the three enhanced algorithms proposed in this paper are effective global optimization tools.In solving the segmentation problem of corn disease spot images,corn with leaf spot,rust,and a mixture of two diseases is taken as the research object.The two-dimensional histograms of the images are constructed,thresholds in histograms are searched using improved algorithms,and the maximum entropy function is used as the objective function to judge the quality of the thresholds.In the experiments,a large number of algorithm models were used to compare segmentation effects.And peak signal-to-noise ratio,structural similarity and feature similarity are used as evaluation measurement.The results show that the improved moth flame optimization algorithm,Harris hawk optimization algorithm and whale optimization algorithm are suitable for low level,medium level and high level threshold segmentation problems respectively.