成像导引头运动点目标检测

在仿真分析的基础上，针对红外成像空空导弹导引头的应用特点，对现有基于形态学的算法进行改进，提出一种新算法。新算法通过计算局部图像的相似度，鉴别背景点与目标点，提高检测和识别概率。算法分四步：(1)进行背景预测，通过消除背景，获得输入图像中相对背景较亮的部分；(2)利用自适应阈值分割，消除大量低灰度噪声点、背景，获得候选目标；(3)利用点-航迹关联，根据点目标运动的连续性剔除噪声；(4)对原始图像中以候选目标为中心的子图像进行相邻帧的相似度计算，根据子图像匹配程度剔除剩下的候选目标中的背景，从而检测出真正的目标。仿真结果证实了新方法对低信噪比复杂背景中点目标检测与识别的有效性。

Point Targets Detection Method for Imaging Seeker

A new algorithm based on morphological operation and sub-image matching was introduced.First,gets the positive impulses smaller than the diameter of structuring element by gray morphological top-hat transformation,so that,the point targets remain in the result image.Second,uses a Constant False Alarm Rate (CFAR) threshold to get candidate point targets.Third,the method based on the continuity and regularity of target movement are used to remove the most of noises and part background leakage.Finally,uses sub-image matching to remove background leakage from candidate targets,which has been tracked for given frames.The new algorithm benefits from the low computational complexity and parallelism of gray-scale morphological operation and accuracy of the image matching.Simulations show the new method is effective.