Investigation Of Forest Types Extraction Technology Using Multiple Classifiers Combination In Tahe

Remote sensing technology invention in the 20th century, after half a century of research, theory and technology have made progressed ripely, remote sensing image classification technology is always accompanied by the development of remote sensing technology.Now remote sensing technology has become one of the most active in the modern era of science and technology. Remote sensing technology has played a significant role in economic development, environmental protection, disaster monitoring, resource development.With the development of remote sensing technology, land use/cover types of information with high precision, fast, automatic extraction, is an important research direction of computer classification of remote sensing images. There are a lot of remote sensing image classification method, but in practice, each classifier for each feature different types of recognition, some classification of certain types of feature recognition accuracy is high, while the other part classifier other feature types of high recognition accuracy, since there are some differences between the classifier, so that no single best classifier. But using a combination of multiple classifiers method for improving the accuracy of remote sensing image classification provides a new way. Multi-classifier combination can make up each classifier, but also have the advantage of each category, so that the combination of multiple classifiers study has important implications for the field of remote sensing image classification.This test based on Entropy classifier combination method will be applied to the extraction of forest types of information. The main findings are:(1) Proposed the use of entropy method to determine a combination of classification rulesEntropy Law will be introduced to the field of remote sensing image classification technique is proposed based on a combination of classification rules Entropy Law. First select the sub-classification of the study area are classified, get each classifier results and accuracy, the use of entropy method for each classification model obtained information entropy, the degree of variation of parameters of the single classifier results to determine the combination classification the right of each sub-classifier weights, weighted to pray for the formation of a new classifier.(2) Multiple classifiers combination of remote sensing image classification and compareIn order to improve the accuracy of remote sensing image classification, multi-classifier combination of ideas, multiple classifiers combination of both an effective way to make up for the relevance and complementarity between classifiers. A combination of the overall classification accuracy of 75.57% than the single classifier improved 3.85%, each of which has been improved classification type of broadleaf forest classification accuracy of 82.32 percent increase of 2.87%, for conifer classification accuracy of 66.45% improved 4.82% of the coniferous forest classification accuracy of 75.49% increase of 4.1%. By comparing the combination classifier and single classifier and the accuracy of the results, indicating that the combination of multiple classifiers ways to improve the accuracy of remote sensing image classification, also showed entropy method is applied to the field of remote sensing image classification is feasible.