Study On Individual Tree Crown Extraction Based On High Resolution Remote Sensing

The forest which feeds up the mankind is one of the most important resources in the earth. As the consideration for nature protection is enhanced, more and more attention has been paid on the vital significance of forest. With the economic and social development and the gradual improvement of the people environmental protection consciousness, the importance of forest resources are increasingly widespread concern.Indicators of measure forest resources mainly include tree height, DBH, canopy widths, crown density, etc.The measurement and estimate of these forest parameters has the vital significance.The traditional method of forest resources survey and management was mainly field investigation and forest map assist, which not only consumed a lot of manpower and material resources,but had update difficulties and low accuracy.Remote sensing, geographic information system and global positioning technologies, providded a more efficient, more convenient, more accurate research methods for survey of forest resources and forest parameters. In recent years, with the emergence of QuickBird, IKONOS high spatial resolution remote sensing data,tre research of forest parameters, in addition to the traditional stand level, but also many research of individual parameters extraction, and the individual tree crown is a hot spot. Compared to the traditional MSS, TM image, the high spatial resolution remote sensing data not only contained intact spectrum information,but also a lot of information of texture, shape, location and other spatial geometry. This image could not only show big landform, but also clearly reflect the detail characteristics, which provide the possibility of individual tree parameters estimation.Using object-oriented image analysis technique, in an appropriate scale,with the image segmentation,QuickBird classification and information extraction could be well done.Based on this background,the study area was Yanjiang forestry station,which was located in Tahe country, Daxing munitions.Using the QuickBird image as the data source,we used visual-manual method and seeded region growing method to study the extraction of individual tree crown and analyzed the accuracy combined with the measured data,and proved the feasibility of this method,which provided the basis to the deeper study of forest resource.