A novel infrared small dim target recognition method based on multi-sensor information fusion using evidence theory and grey model

Multi-sensor information fusion technology owns efficient capability to recognize small dim targets from complex ground background in the remote sensing images. A novel small dim infrared target detection and feature extraction algorithm is applied firstly by using line average subtraction and block-threshold segmentation in dual-channel mid- and long-wavelength infrared images. The further correlation analysis on grey model is used to generate the basic probability assignment function. Then, Dempster-Shafer evidence theory of evidential reasoning is employed to classify the final target type. Experimental results indicate that this method performs more efficiently in target detection and recognition comparing with the classical algorithms.