A new method for recognizing point targets

An information fusion method based on the combination of artificial neural networks (ANN) and Dempster-Shafer's evidence theory to identify point targets is introduced in this paper. Two kinds of targets fly together in high space from far away, so that each of them occupies only one pixel in the image obtained by the detector and has not any shape characteristics. These two targets are supposed to have identical motion and visible characteristics but different construction, so the infrared characteristics is the only characteristics available. Six and twelve infrared features are adopted, respectively. A BP networks is used to identify these targets at first, then an information fusion method. It is found that the later has a much higher identification rate.