Development Of Forest Remote Sensing Image Classifier Based On Support Vector Machine

Remote sensing technology is a comprehensive sensing technology for detecting long distance targets through sensors.It has been widely used in many fields,such as geographical conditions survey,disaster resistance and flood control monitoring,urban planning and surveying and mapping,cultural heritage and ecological environment protection,natural resource detection and so on.In view of the problem of remote sensing image processing that needs to be solved in the process of forestry informatization,image segmentation and classification technology has gradually become the focus of development.Especially with the development of remote sensing imaging technology,the resolution of remote sensing images has been continuously improved,and higher requirements for image classification technology have been put forward.In this paper,the background,significance,difficulties and future trends of this study are analyzed,and several classical methods of image segmentation and classification are introduced.After analyzing the shortcomings of traditional classifier in the face of high score images,the support vector machine model is selected to study the classification process of remote sensing images.In this article,ENVI+IDL two development platform,complete the support vector machine classification model for high resolution remote sensing image,and train and test the classifier by dividing the sample set to verify the performance of the classifier.The main contents of the article are as follows:1.Aiming at the problem of high resolution forestry remote sensing image classification,the remote sensing image data and small class data are superimposed,and the sample sets are divided into the training support vector machine classification model,and the feature extraction process of the training samples is optimized.In view of the shortage of small class data and lack of prior knowledge,the theory of migration learning is introduced and the prior knowledge is fully utilized to combine the marked training samples with unlabeled training samples to form a new training sample set,and the feature combination is selected as the classification basis,and a preliminary support vector machine classification model is constructed.2.Through the continuous training of the preliminary model,the parameters in the classification model are optimized,and the optimal parameter set suitable for the experimental environment in this study is finally formed,thus the optimal classification model is constructed.