提出了一种新的复杂背景下低信噪比红外弱点目标检测算法。根据红外弱点目标在图像中的三维空间特征,从空间认知的角度出发,将三维的灰度分布特征转化为二维的等高线曲线特征, 建立红外图像的等高线图(IECM)描述,利用图论中的树结构(等高线树)形式化地表达等高线的空间关系,在此基础上,给出弱点目标检测的等高线树检测准则,同时给出了等高线划分等级的选择方法。理论分析与实验结果表明,该算法具有良好的检测性能,且结构简单,利于硬件实时实现。在信噪比为1.4的情况下,对红外图像序列的检测概率为96.3%。

A new detection algorithm for weak dim target with strong clutter background and low SNR is proposed. The intensity distribution characters of the image are converted to contour lines according to the 3D spatial characters of the image, and the Iso-Elevation Contour Map (IECM) of the infrared image is represented. The formative tree structure (contour tree) of graph theory is used to express the spatial relationship of contour lines. The rule of detecting weak dim target is established. Theoretical analysis and experimental results show that this method can provide good detection performance and its logical structure is simple to be implemented in real-time system. The detection probability is 96.3% when the SNR of the image is 1.4.