提出了一种新的管道滤波方法并将其应用于红外图像序列中弱小运动目标检测。首先采用局部灰度特征分析的方法对图像序列的一帧进行逐点检测,检出候选目标;然后对每个候选目标建立管道,利用局部灰度特征的分析方法判断后续帧的管道区域内是否仍然存在目标,保留仍然存在目标特征的管道,清除不具备目标特征的管道。该方法不需要对每一帧图像都进行复杂的管道建立、维持、消除等操作,具有丢失目标后快速重新捕获能力。实测红外图像序列的实验结果证明了所提方法的有效性和抗干扰能力。

A novel pipeline filtering method was proposed and used to detect small infrared target in infrared image sequence in this paper. First, potential targets in one of frame in infrared image sequence were detected by feature analysis of local grey level. Then, the pipelines were built for every potential target. The filter course in later frames of the infrared image sequences was: judging whether there was still existing target feature in every pipeline; if yes, held the pipeline; if no, cleared the pipeline. True targets were detected in those pipelines with enough length. With this method, it is unnecessary to build, hold or clear pipelines in every frame in infrared image sequence, and this has strong re-capture ability when target is lost. The efficiency and anti-jamming ability of this method are testified by experiments.