红外弱小目标检测中的相邻背景预测算法研究

Adjacent Background Prediction Discussion in IR Small Weak Targets Detection

红外弱小目标检测的关键问题是背景预测,背景预测的好坏直接关系到红外目标的提取。考虑背景预测过程中,待预测像素仅仅和相邻像素相关,提出了一种相邻背景预测的算法,并针对背景起伏和平缓情况进行详细的分析。实验结果表明无论是在背景起伏较小的图像,还是在背景起伏较大的图像,该算法都具有良好的目标检测性能。

Background prediction is the key problem in IR small weak targets detection. Its performance influences the detection results directly. Taking into account that the pixel to be predicted only relates to the adjacent pixels, this paper introduces an algorithm named adjacent background prediction and analyzes the performances of the algorithm in processing heavy clutter images and smooth clutter images. Experimental results indicate the robustness of this method both in processing heavy clutter images and normal clutter images.