The Feature Extraction And Recognition For Small Target Under Complex Background

How to distinguish the target from decoys is a difficult and challenging problem in information processing of interceptive missile. At the present time, there are a lot of unsolved question in the system of detection and recognition for small target, because we can't get enough of evidence that characteristics about IR target picture of small target and decoy and so on, including geometric shape and texture. In the thesis, the technique of recognition for warhead and decoy in the IR image sequences is discussed, we get the result effectively.Firstly, we discuss the important and necessity of pretreatment for infrared small target detection under complex background during the recognition of warhead and decoy in IR image sequences. And we give the example of small taget processing under actual complex background, According to the statistic characteristics of IR background clutter, a new combinability structure element algorithm for background suppression is presented to improve its signal-to-noise ratio; According to the moving characteristics of small targets in IR images sequences, a judging method based on neighborhood is discussed, emulational results show this method can distinguish small target from sequences effectively.Secondly, we elaborated the infrared image characteristic of target(the target is warhead in the thesis) and decoy during the moving, Based on the difference analysis about the characteristic of warhead and decoys, four primary recognition characters—grayscale character,temperature character,motion character and shape character are brought forth, At different imaging phase, different character is chosen to determine which object is the warhead and which one is decoy. And all the elementary judgments are fused under the D-S evidence theory into a good recognition result.Finally, the work in this thesis is summarized, and the key problems and direction of the future research are outlined.