A Research On Target Recognition Algorithms For Remote Sensing Image

These days more and more attention is given to Automatic Target Recognition technology for remote sensing images. This paper deals with the problems on satellite remote sensing search and information fusion,mainly about two kinds of target recognition: large linear targets such as airfield runway and harbor and small targets such as planes and warships.The paper is organized as follows:Firstly,an overview of remote sensing image processing and automatic target recognition technologies are given;Secondly,a comprehensive comparion of line detection methods is done and then an improved algorithm to detecte parallel straight lines based on Hough transformation is presented and applied to the recognization of airfield runway and harbor.Thirdly,it puts forward an algorithm to detect harbors based on boundary straight line phase.At last,an intergrated algorithm to recognize small targets such as planes and warships is developed,which efficently combines the advantage of Moment Invariants and projection figure matching. The result of multi-feature matching are fused by means of fuzzy intergration or D-S evidence theory. Experimental results show the performance and efficiency of the algorithms.