Remote Sensing Image-based Target Identification And Location Technology Research And Practice

Vfith the rapid development of remote sensing technology in recent years, space remote sensing systems of the time can already supply users with magnitude image data of high spatial resolution high spectral resolution and high time resolution. "Obtaining military information from remote sensing images" has been a very important means for acquiring information and a significant component of modern wars. "Obtaining military information from remote sensing images" is mainly engaged in solving three special problems, namely "WWW" problems. "WWW" means what, where and why. "What" illuminates attributes and properties of the military objects, "where" indicates the location of them (accurate geographic coordinates), "Why" explains the significance of them. The research goal of the paper is to resolve "what" and "where" problems of military objects, which is how to recognize and locate roads^ waters and other important objects form remote sensing images. The research work of the paper can be divided into following parts:1) In order to fully integrate information from different kinds of remote sensing images and overcome information shortage in automated image interpretation using single kind of remote sensing image, the paper proposes some image processing methods and improves quality of image fusion efficiently.2) In view of DEM' s significance in remote sensing image processing, the paper discusses two kinds of DEM generation methods. One kind is to achieve DEM from interpolation of vector data , the other kind is to obtain DEM directly from remote sensing images. The later method realizes DEM acquisition in large area.3) In order to realize object location at the same time, this paper accomplishes orthorectification for several typical remote sensing images. As to the "TM" image, which is complex in imaging mechanism, the absolute registration scheme is applied to it during orthorectification.4) This paper realizes object location recognition commutatively. The paper realizes road and some area objects recognition semi-automatically or automatically.5) Realizing the combination and integration for qualitative, quantitative analysis and position in object recognition.