Research On Improved Intelligent Optimization Algorithm Of Birds’ Species In Image Segmentation

Image segmentation technology is an important means of image processing,analysis,understanding and computer vision,and the threshold image segmentation scheme has attracted extensive attention of researchers because of its high efficiency.But the computational complexity required for multi-threshold image segmentation increases sharply with the increase of the number of thresholds,which greatly reduces the efficiency of image segmentation.The application of intelligent optimization algorithm has significantly improved the efficiency and accuracy of multi-threshold image segmentation.Based on the comprehensive analysis and summary of existing intelligent optimization algorithms,intelligent optimization algorithms for birds are an important branch.The multi-threshold image segmentation method of the optimization algorithm further optimizes the image segmentation effect through the improvement of the satin blue bower bird optimization algorithm,the vulture algorithm and the northern goshawk optimization algorithm.The specific improvement plan is as follows:(1)Aiming at the problems of low optimization accuracy and slow convergence speed of Satin Bowerbird Optimization Algorithm(SBO),a Satin Bowerbird Optimization Algorithm based on chaotic initialization and Cauchy mutation strategy(Improved Satin Bowerbird Optimization Algorithm,ISBO).In the process of improving the algorithm,in order to improve the algorithm’s optimization accuracy,convergence speed and pertinence of the initial population,the Logistic chaotic map is introduced to initialize the population.In order to prevent the algorithm from falling into local optimum(premature)prematurely,the search performance of the algorithm is improved through the Cauchy mutation strategy.In order to reflect the value of the improved algorithm in engineering and practical applications,this paper applies ISBO to the segmentation of natural images,medical images and plant images.Based on a large amount of visual and quantitative data analysis,this paper compared the improved satin blue bower bird optimization algorithm(ISBO)with the original satin blue bower bird optimization algorithm(SBO),fuzzy Modified Discrete Gray Wolf Optimizer and aggregation algorithm(FMDGWOA)and Fuzzy Coyote The Optimization Algorithm(FCOA)has been fully compared and analyzed,and the final experimental results show that the ISBO algorithm has achieved better segmentation results for natural,medical and plant images.(2)On the basis of improving the Bald Eagle Search Algorithm(BESA)in the three stages of initialization,selecting the search space and searching for the target prey,this paper takes fuzzy Kapur and Tsallis entropy as the objective function,and uses the improved BESA to complete The basic framework of thresholding segmentation,and through the median aggregation method to assist in the realization of high-precision multi-threshold image segmentation,finally formed the Multi-stage Improved Bald Eagle Search and Aggregation Algorithm(MIBESA).According to the order of BESA stage division,the elite reverse learning strategy is introduced in the initialization stage to obtain a better initial population.In the stage of selecting the search space,a new adaptive factor is defined by the number of iterations,which is used to replace the original random position change factor.In the stage of searching for the target prey,the inertia weight is introduced to effectively balance the local and global search capabilities of the algorithm.Based on these three stages of optimization improvement,with Bald Eagle Search and Aggregation Algorithm(BESA),FMDGWOA,Beta Differential Evolution Algorithm(BDE),Artificial Bee Colony Using Sine-Cosine Algorithm(ABCSCA),Modified Whale Optimization Algorithm(MWOA)and Improved Cuckoo Search Algorithm(ICS)comparison of six similar algorithms.The final experimental results show that MIBESA has ideal results in natural,medical and plant image segmentation.(3)Taking fuzzy Kapur as the objective function,a multi-threshold image segmentation method based on the improved Levy and Simulated Annealing to Northern Goshawk Optimization(LSANGO)algorithm is proposed.In the process of improving the algorithm,the Levy flight strategy is firstly added in the exploration stage,and the random jumping of the step size of this strategy is used to enhance the global search ability of the algorithm,thus obtaining a faster convergence speed.Then in the position update stage,the simulated annealing mechanism(Simulated Annealing,SA)is introduced,which effectively solves the problem that the algorithm is easy to fall into local optimum during the optimization process.In order to better evaluate the performance of the algorithm,the LSANGO algorithm was compared with Northern Goshawk Optimization(NGO),FCOA,Whale Optimization Algorithm(WOA)and Altruistic Harris Hawks’ Optimization Algorithm(HHO＿Altruism)for image segmentation.Finally,through the quantitative analysis of peak signal-to-noise ratio(PSNR)and feature similarity(FSIM)performance indicators,the superiority of LSANGO algorithm in natural,medical and plant image segmentation is verified.The three biologically inspired algorithms cited in this article start from different angles and use different strategies to solve the problem,all of which improve the image segmentation accuracy to varying degrees.At the same time,the improvement of the three intelligent optimization algorithms for birds has the same effect.Also,different improvements have been made in the algorithm initialization phase,population update phase,population mutation phase and population elimination phase,and excellent results have been achieved.