A study of the potential of using worldview-2 of images for the detection of red attack pine tree

摘要

Forest disturbances in South China caused by pine wood nematode may result in widespread tree mortality. In order to decrease damage to forest ecosystem and huge loss to national economy, early detection, early diagnosis to individual infected tree is essential to forest management agencies. However field survey is hard to achieve the fine management requirements. Satellite remote sensing technology has the characteristics of landscape of coverage, convenient, and fast in formation acquisition, so it is one of the most important and most effective means of red attack monitoring. The support vector machine(SVM) classification algorithm have been proposed as an alternative for classification of remote sensing data. The study is based on a multispectral Worldview-2(WV-2) scene and uses support vector machine(SVM) methods. We compared the eight bands with three bands of the image based on SVM and came to the conclusion that WorldView-2 are suitable for individual tree identification. Three visible bands spectral data can also discriminate discolored individual tree successfully. In other words, three visible bands of remote sensing can meet the requirements of red attack pine estimation and extraction.