基于高精度边缘方向求解的红外弱小目标检测

Infrared Dim Target Detection Based on Highly Precise Edge Orientation Searching

针对边缘方向误差较大导致各向异性SUSAN算法背景抑制能力降低的问题,提出了一种高精度边缘方向求解算法来确定实际边缘方向;并引入独立强度模型来确定短轴的尺度以实现滤波器长、短轴自适应地确定。实验表明:高精度边缘方向求解算法的引入能显著提升红外背景重建的精确度,从而提高了算法的背景抑制能力,经过该算法滤波后的图像其信噪比、检测率都获得显著提升,虚警率明显降低,能有效提高系统的单帧红外弱小目标的检测能力。

The ability to restrain infrared background would decrease if the edge orientation is quite different from the real edge orientation. Aiming at solving this problem, a method of highly precise edge orientation searching algorithm is proposed to ascertain the real edge orientation. An Intensity-Dependent-Spread model is also introduced to ascertain the scale of V-axes so that the U-axes and V-axes of the filtering can be ascertained adaptively. The experiments show that the highly precise edge orientation searching algorithm can remarkably improve the precision of rebuilding infrared background, so the ability to restrain infrared background can be improved. The SCR and the detection rate will be remarkably improved after processing by our algorithms, and meanwhile the false alarm rate will remarkably decrease. So the algorithm can effectively improve the ability of the system to detect the infrared dim target in a single frame.