用心翻译,请楼主费心审阅
The rapid development of acquisition technology of remote sensing data, and large numbers of high-resolution remote sensing images such as IKONOS, QuickBird, have greatly promoted a wide range of applications in a number of industry sectors. High-resolution remote sensing images provide more information (texture, shape, topology, etc.)than low-resolution remote sensing images. For the characteristics of high-resolution remote sensing data, some scholars put forward the object-oriented information extraction method. The method not only uses features’ spectral information, but also its geometry and structure information. This can make up for the disadvantages of traditional pixel-based information extraction method. With the improvement of classification accuracy, more feature information can be extracted from high-resolution images which can support all areas research better.
In the object-oriented high-resolution remote sensing image information extraction methods, image segmentation is the foundation, which directly determines the generation of image objects, while the choice of segmentation scale is one of the key issues. Different segmentation scales will lead to totally different results. For a specific feature, more effective information extraction can be performed at a proper segmentation scale. In recent years, scale effect has always been a hot research issue for researchers, and also the key issues of acquiring ground information from images at a high degree of automation ultimately; however, in all fields so far, there are no clear and common scale selection criteria. Therefore, the study of scale effect on the impact of segmentation and classification and the determination of optimized segmentation scale of features in specific images which can be used to generate the most reasonable object hierarchy of the image have become very urgent and of great significance.
