为解决低信噪比快速运动点目标的检测问题，文中提出了一种基于顺序形态滤波的点目标检测算法。首先采用百分位p=1的顺序形态滤波提取目标的广义内边缘，找出少量候选运动目标点，然后利用图像流模型建立目标的运动约束方程，根据目标像素在图像序列中运动的连续性和轨迹的一致性原则，在多帧连续图像中检测出真实的目标运动轨迹。实验结果证明，该方法能够快速、可靠检测信噪比小于3的运动点目标。

To resolve the movingpoint target detection problem under the condition of low SNR, a new point targetdetection algorithm based on generalized edge extracted by order morphology filtering ispresented. Two steps of point target detection are adopted. First, background suppressionand generalized inner-edge of targets extraction are performed by order morphologyfiltering whose percentile is selected to pick out a small quantity of possible movingpoint targets. Then the motion constraint equation of the targets is set up according tothe image flow model. By use of the model, the true target motion trajectory is detectedin multi-frame successive image. The experiment results show that the method caneffectively and reliably detect the moving point target with SNR less than 3.