Research On Detection Algorithms Of Small Target Based On Background Prediction In Infrared Clutter

Background prediction algorithms is the correlation between the pixels to predict the background, but the noise in images are independent which can not be predicted accurately, thus the existing background prediction algorithms can not eliminate the interference of noise. In response to this limitation, here a self-participate prediction algorithm (SPPA) is proposed in this paper. This algorithm can be used with any existing background prediction algorithm. The predicted pixel has the current noise information in its input value, through corrected the prediction value obtained by existing algorithms with the input value of the predicted pixel, SPPA algorithm can improve the accuracy of background prediction, reduce the interference of noise in the residual images, and improve the signal-clutter-rate (SCR), and finally enhance the detection probability.In this paper, the SPPA algorithm has been justified theoretically and simulated with real infrared images. Experimental results indicated that this algorithm has good effect on suppression of background clutter and reduction of the noise interference, the signal-to-clutter ratio of the images processed by this new algorithm is improved almost 2 or 3 fold than those processed by existing predictors directly.