Prediction Of Coal Surface Deformation Based On Gene Expression Programming

Hyperspectral remote sensing,because of its high spectral resolution characteristics,made the already wide-band remote sensing that can't detect substances can be detected in hyperspectral remote sensing,thereby enhancing the ability to detect surface features destination information,favored by many domestic and foreign research scholars in recent years.Many researchers used efficient and powerful SVM search capability evolution to solve the problem that the traditional classification method required that you obtain the feature category labeled target problem to a certain extent.However,the most commonly used support vector machine image classification method is limited to selecting parameters such that classification was lower in accuracy and speed.In view of this situation,this paper optimizes vector machine support parameters with hierarchical cluster analysis,combined with the characteristics of hyperspectral remote sensing image data,which was introduced in SVM support vector description of the algorithm,the use of space metric separability nature has been an improvement support vector machine classification method,this paper applies this classification to classify hyperspectral remote sensing images,the main work was as follows:(1)For the shortcomings of image classification method of support vector machine on the kernel function and parameter selection,this paper used hierarchical cluster analysis select support vector machine parameters.(2)The combination of high spectral remote sensing data features introduced in SVM support vector description of the algorithm,the use of space metric separability properties,the hyperspectral remote sensing image data is contained in a super ball as small as possible,it put forward improved support vector machines image classification method.(3)In the.NET platform,it achieved improved support vector machine algorithm for image classification with C # programming language.(4)It improved the image classification support vector machine applied to hyperspectral remote sensing image classification,and the classification and support vector machines used image classification method comparative analysis,then the classification and common traditions classification of BP neural network classification methods and K-means classification methods were compared.By comparing the experimental results,the conclusion can be obtained:improved support vector machine classification of hyperspectral remote sensing achieved overall accuracy and high speed,which can explain proposed improved SVM method for hyperspectral remote sensing image can be classified,it provided a new method for the classification of hyperspectral remote sensing images.