Research On The Methods For Images Display And Targets Detection In Infrared Early Warning System

Real-time display of wide-field-of-view video and targets detection are very important for the infrared early-warning system. Stable display of wide-field-of-view images in real time within certain azimuth, is helpful to improve system performance. While real-time dim target detection in low Signal-Noise Ratio (SNR) images can extend combat range and prolong response time, which is of great importance to improve survival probability.This paper mainly deals with the techniques of image mosaics and algorithms of targets detection based on background modeling for infrared early-warning system. Firstly, classic methods for image mosaics are analyzed, and a new image mosaics algorithm based on phase congruency is proposed, which detects feature points according to their phases' congruency, and computes transformation matrix between images by matching these feature points. This approach achieves better robustness compared with previous methods. Secondly, in terms of characteristics of infrared image, a clutter suppression and target detection method based on background modeling is proposed. A scheme combined of one-frame target detection and multi-frame target determination is adopted in this method, which detects targets using one-frame image through background modeling algorithm based on nonparametric regression estimation and determinates targets using multi-frame images though a fading recursive least-squares technique. This method separates the targets according to the low relativity of gray value between target points and local background, which is proved to have good performance in clutter suppression, and is easy to implement by hardware in real time.The experimental results with the real infrared images demonstrate the effectiveness and practicability of the proposed algorithms for the infrared early-warning system.