Research On Remote Sensing Image Segmentation Based On GAPSO-Kmeans Clustering

With the rapid development of remote sensing monitoring technology,the information covered by the images acquired by remote sensing platforms is also becoming increasingly abundant.The remote sensing image has laid a solid foundation for many researches because of its abundant data volume.However,there is a huge amount of data in remote sensing images,which is complicated,multi-spectral and multi-level.Special situations such as different target shapes often occur.Therefore,how to accurately segment remote sensing images has become an important research direction for remote sensing image processing.As a classic unsupervised learning method,clustering algorithm has been well applied in the field of image segmentation.Based on the results of Kmeans clustering algorithm in the field of computer vision,this paper focuses on improving the performance of Kmeans algorithm in remote sensing image segmentation.The main work of this thesis is as follows:(1)In order to solve the problem that the single calculation of the Kmeans clustering algorithm takes a long time,the distance between the data is calculated by combining the Manhattan distance and the Chebyshev distance into a linear distance model instead of the Euclidean distance,and then this linear distance is used.The model is clustered.Experiments show that the remote sensing image segmentation method based on the Kmeans clustering algorithm can effectively reduce the segmentation time of remote sensing images while ensuring the accuracy of segmentation.(2)Aiming at the problem that the Kmeans clustering results are too dependent on the initial clustering center selection,a remote sensing image segmentation algorithm based on particle swarm optimization and improved Kmeans clustering is proposed.First,the inertia weight in the particle swarm optimization algorithm is dynamically adjusted to obey a probability distribution.Then,the overall optimal solution output by the dynamic particle swarm optimization algorithm is used as the initial clustering center of the Kmeans clustering algorithm.The linear distance model completes the clustering.Experiments show that the improved remote sensing image segmentationmethod can effectively improve the remote sensing image segmentation visual effect and segmentation accuracy.(3)Aiming at the problem that the particle swarm optimization algorithm is prone to wander near the local optimum,a GAPSO hybrid model combining genetic algorithm and particle swarm algorithm is proposed for remote sensing image segmentation.In order to combine the global search ability of the genetic algorithm and the fast convergence ability of the particle swarm optimization algorithm,firstly,by comparing the optimal solutions of the corresponding populations of each generation of the genetic algorithm and the particle swarm optimization algorithm,and assigning operations between the two algorithms,the overall optimal solution is output as the result to the Kmeans algorithm as the initial clustering center,and clustering is completed using the proposed linear distance model.Experiments show that the improved remote sensing image segmentation method can effectively improve the remote sensing image segmentation performance.