针对红外图像中海天交接处目标能量较弱, 难以获得精确检测的问题, 本文提出一种基于杂波抑制的海平线弱小目标检测方法。该方法通过对红外图像特征分析, 检测海平线像素特征, 去除障碍物干扰, 确定海平线区域; 对海平线区域采用改进的均值差滤波与背景建模, 去除海平线杂波; 杂波滤除后, 计算两个区间自适应阈值作为滞后阈值, 从而检测出海平线弱小目标像素。实验结果表明, 本文提出的算法能够有效去除海平线杂波, 确定海平线区域, 有效提取海平线像素, 从而准确检测出海平线弱小目标, 具有较高的检测率与较低的虚警率, 验证了该算法的有效性。

This paper proposed a method based on clutter rejection for small dim infrared target detection from horizon region. Firstly, the horizon pixels were detected by analyzing the features of infrared image and removing obstruction of islands, seaside or buildings, so as to determine the horizon region. Secondly, the horizon region was processed with improved mean subtraction filter and building background models to remove the horizon line clutter. Finally, it calculated the adaptive threshold of two region as the hysteresis threshold, so the dim target pixel could be determined. The experimental results show that the proposed method can effectively remove horizon line clutter to detect horizon pixels, and accurately detect the small dim infrared targets. It produces low false alarms and high detection rate, which validate the method.