Infrared image target recognition of complex background based on curvelet neural network

To improve the recognition rate of the complex background infrared image target, an infrared image target recognition method based on curvelet neural network is proposed. In this method, first of all, all sample images and test images are decomposed by fast discrete curvelet transform. Curvelet coefficients of different scales and various angles are obtained. The low frequency coefficients are applied as characteristic parameter to the SOM neural network for training. Finally, the trained SOM neural network is used for target recognition. The method is not only able to reduce the amount of data to be processed, but also to improve the recognition rate. Simulation results show that the proposed method is superior to the other recognition methods in performance and its recognition rate can reach more than 95 percent.