Rapid Collecting And Processing Of Ground Sample Data And It's Expand Application

The sample data is the basic data for supervised classification of remote sensing images.A large number of high-quality sample data play an important role in improving the accuracy of remote sensing image classification,obtaining the area of land cover types and monitoring land cover change.In the large-scale research area,relying on the current sample data acquisition method,it is relatively difficult to obtain a large amount of data,and less sample data becomes a short-board for large-scale remote sensing image classification.Therefore,how to carry out rapid sample collection and processing is a key issue to solve the problem of small samples in large-scale remote sensing image classification.This paper proposes a method of sample collection using GPS camera,and introduces the process and precautions of sample collection in detail.The GPS photo data rapid processing system is developed by IDL language,and the development process and usage method are elaborated.The correction of the midpoint and the supplement of the sample data improve the accuracy of the sample database and increase the number of samples.At the same time,the sample data was optimized.After purification,expansion and filtering,the training samples were formed,and the sample separation degree was calculated to test the sample quality.Finally,the random forest image was used to classify the remote sensing image by land cover.Sample training method.The overall accuracy of the classification results is 79%,the Kappa coefficient is 77,and the F1 score is 72.42.The results show that the computer automated training samples have a good effect,but at the same time there is still much room for improvement.