Application Of Image Segmentation Method Based On Kinetic-Molecular Theory Optimization Algorithm In Vehicle Image Segmentation

Kinetic molecular theory optimization algorithm is a new heuristic global optimization algorithm in recent years.Because of its attraction,rejection and fluctuation mechanism,the algorithm is easy to implement and fast convergence.It also provides the possibility to taking into account the diversity of the population.Therefore,the algorithm has received widely attention,and has been successfully applied to the fields of power system economic scheduling,power system reactive power optimization,intelligent building optimization scheduling and image segmentation algorithm,etc.In the application fields,such as image segmentation,which need a lot of computation,many problems in image domain can be transformed into optimization problems due to the development of optimization algorithm.,image threshold segmentation can be converted into parameter optimal calculation in optimization problems.Optimization algorithm is used as a tool to calculate the optimal value of objective function under a certain criterion iteratively,that is,the optimal threshold value of image segmentation.To solve the multidimensional data in images,the optimization algorithm,especially the intelligent optimization algorithm,greatly reduces the computation and shortens the computation time.Many experts and scholars have done a lot of research on the advantages of the optimization algorithm.Based on the previous work,some improvement measures is proposed in this paper for the shortcomings of kinetic molecular theory optimization algorithm.First,the dual population idea is adopted to realize the optimization process through division of work and cooperation.Second,in order to make the common population search in a large area,the diversity fluctuation operator is proposed,the global search ability of the algorithm is enhanced based on the feedback diversity information regulation mutation rate and the global search algorithm is improved.Third,in order to fine search elite population,a cooperative learning operator is designed to avoid the problem of excessive randomness,the experimental results show that the improved algorithm has different degrees of improvement in the convergence precision of the algorithm.In the multi threshold image segmentation application based on Otsu criterion,the improved algorithm also has the advantages of fast convergence,high stability and good comprehensive effect.When the image contains noise,the common segmentation model based on Otsu algorithm can not effectively segment the image,a Otsu method based on the line intercept histogram is found to denoise effectively,but when the image is mixed with salt and pepper noise,the image can not be segmented correctly.Then the line intercept histogram Otsu based on the post processing strategy is proposed,the post processing strategy can effectively remove the Gaussian and salt-pepper noises in the image,and apply the dual population kinetic molecular theory optimization algorithm.to the threshold search.The efficiency and quality of image segmentation are ensured by improved algorithm and optimized segmentation model.The complete vehicle image is detected by setting the virtual detection area in the video captured by the video capture card.Through the improved image segmentation method in this paper,the contour and other information of the vehicle can be effectively segmented to avoid the interference of other factors to the segmentation results,and the performance of the improved segmentation method is effective.The effectiveness of the improved image segmentation method is verified.