Classification Of Mesoscale Remote Sensing Image Based On CSBP Model

Remote sensing image classification is an important technique for extracting the ground object information from remote sensing images,and it is also a hot research topic in the field of remote sensing.The mesoscale remote sensing image,such as LANDSAT image,has the characteristics of wide coverage and easy access,and is often used as the basic data of scientific research.We can get a larger range of land cover results by using the mesoscale remote sensing images as a data source for classification and recognition.BP neural network is a structure constructed by simple abstraction and Simulation of the brain nervous system,which has the characteristics of self-adaptive,self-organizing and self-learning,and can realize distributed storage and parallel processing of data.Because of the special data structure and strong ability for fitting,BP neural network is very suitable for dealing with the nonlinear problems such as image classification.With the continuous development of related theories in recent years,BP neural network has been widely used in the field of remote sensing image classification.It has been proved with practice that the use of BP neural network can significantly improve the accuracy of remote sensing image classification.However,there are still some problems in the practical application of BP neural network,such as sensitive to the initial weight threshold,easy to fall into the local optimal solution,etc.In view of the above problems,this paper uses the cuckoo search algorithm to optimize the BP neural network and build the CSBP model.For the problem that the standard cuckoo search algorithm has slow convergence speed and low precision in the late stage,an adaptive step length strategy is used to improve and ROSENBROCK function is used to test the optimization ability of the improved algorithm.In order to verify the optimization effect of cuckoo search algorithm on BP neural network,traditional BP neural network,BP neural network optimized by standard cuckoo search algorithm,and BP neural network optimized by improved cuckoo search algorithm were used for the experiment of remote sensing image classification,and Compare the Classification Results of the ThreeAlgorithms.It was found that the overall classification accuracy of remote sensing image classification using the BP neural network optimized by improved cuckoo search algorithm was 88.3927%,and the Kappa coefficient was 0.8646.Compared with the original BP neural network and the BP neural network optimized by the standard cuckoo search algorithm,the classification accuracy of the new algorithm was improved by 9.42% and1.58%,respectively.