Neural Network Use In Infrared Recognition

To the characteristic of infrared image, the paper adopts two methods for elimination noise, the two methods are compound elimination noise and deformation model, for improve the ratio of signal and noise, it puts forward two methods. Then it studies segment color images in the feature space to choose a method which bases on non-supervise, and the experiments have testified the efficiency. In the processing of recognition, respectively adopt Radial Basis Function(RBF) Neural Network. The method especially fits for the recognition of dynamic objects which are moving, revolving or resizing. Experiments have testified that the method has a good development in the future.