Research On Techniques Of Detection And Recognition Of Target In Dual-band Infrared

Small target detection and recognition based on infrared image is a one of the most important technology in infrared guidance weapons. In order to improve the ability of target detection and recognition, this paper study small target detection and recognition based on dual-band infrared image from three aspects.Firstly, we analyzed the characterization of small target, combining with the target imaging modeling, which is the basic of the detection and recognition of small targets. Limited observational data can’t reflect the feature of target, so it is necessary to produce the target database through modeling and analysis of the target characteristics. A dual band infrared target imaging model is established, considered the characteristics of the detector, the target properties, and environmental factors, then the simulation images and target database is generated. With the example of three typical targets we analyzed the image characteristic that can be used for target detection and recognition.Secondly, a fusion algorithm of small target detection based on adaptive decision threshold and a multi-level fusion algorithm of spotted target detection on local energy are proposed. Usually, the prior knowledge about the infrared image is unknown, and the fusion detection center considers each detector having the same performance, which leads to detection performance degraded when one of the sensors severely degraded. In order to solve this problem, we proposed a fusion algorithm of small target detection based on adaptive decision threshold. The core idea is that adjusting each band detection threshold dynamically under the false alarm rate, so as to ensure the stability of fusion detection performance. Also, according with the features of spotted target, we propose a multi-level fusion algorithm of spotted target detection on local energy. The results of experiment show that, this algorithm can improve the detection performance significantly when the two detectors have the same performance. The algorithm can still keep stable performance when one of the detectors degraded severely.A dim target fusion recognition system had designed based on the dual-band infrared sensors. The image of the warheads and decoys is small and has less feature information which makes it difficult to identify the real target. For this proplable this article proposed a fuzzy comprehensive target recognition algorithm. This target recognition system can fusion the judgment of each band and then give the final recognition result. The results of the experiment show that the target recognition system increases the accuracy rate of target recognition effectively.