一种新的背景预测方法在红外弱小目标检测中的应用

Application of background prediction in infrared weak and small targets detection

针对空中红外弱小目标的检测,提出了一种新的背景预测方法。该方法对云层边缘区域的点采用亮暗点分类,寻求最相似点的方法进行预测;对非边缘区域的点采用基本背景预测法进行预测;最后经过背景对消和阈值分割,将弱小目标检测出来。实验结果表明,该方法能够提高对起伏背景预测的准确性,减小由于云层边缘预测不准确而引起的虚警,从而能够更加有效地检测到弱小目标。

Proposing a novel method based on background prediction to detect aerial weak and small targets in infrared images. For the points in the area of the cloud＇s edge, we adopt bright and dim point classification and seek most similar point as predicted point. For the non-edge area, we use basic background prediction method to predict. Then detect weak and small targets through eliminating background and threshold segmentation. The experimental results indicate that the method can enhance the predicted accuracy of the edge of cloud and eliminate false alarm points caused by the edge of cloud. Using this method we can detect weak and small targets from IR images more effectively.