Infrared Dot Object Multi-Spectrum Data Fusion Identification Research

When the target such as missile coming with high speed ,it's image is a dot on the focus face of detector. Its infrared spectrum distributing information can be detected. And the infrared spectrum information can also identify the target. So we can use the infrared spectrum information to recognize the target and we can increase the reaction time of the fighting object in the war. Because the spectrum information which is got from the detector is uncertain and un-integrated , the recognition result lies on the fusion-recognition method.Here we developed a multi-period multi-spectrum pattern recognition approach for dot object by using ANN parallel nets and D-S theory in the paper. The forestage of the fusion model completes target presort and its post-stage is used to multi-period uncertainty inference and the whole set distribution of probability. Because of the parallel ANN, the approach avoids the multi-sample training question for single ANN. In this article we realize multi-spectrum infrared dot target classification through the approach...