紫外图像弱小目标检测与跟踪

Weak or Small Target Detection and Tracking on Ultraviolet Images

工作在"日盲"波段的紫外信号探测,屏蔽了太阳紫外辐射这一最大背景噪声干扰源,可以有效地进行微弱紫外目标信号采集,但是仍然面临着目标/噪声难以区分、目标短暂丢失等问题。文中对微弱紫外目标的成像特性进行详细分析,根据紫外目标成像特征提出了一种紫外图像弱小目标检测跟踪方案。该方案首先采用基于连通域的聚类方法对紫外图像中的样本点进行聚类,标识出前景中的目标块和噪声块,然后利用真实目标的轨迹特性,结合卡尔曼滤波预测目标的运动参数,从而实现对目标的检测跟踪。仿真实验证明了该方案的可行性及有效性。

Operating on solar-blind band,the ultraviolet signal detection masks solar ultraviolet radiation as the maximum background noise,whereas the weak or tiny ultraviolet target signal can be effectively collected,some problems,such as distinguishing noise and target,and temporary loss of target still exist.The imaging characteristics of ultraviolet small target are analyzed and a scheme is proposed for tiny UV target detection and tracking.A connected domain based clustering method is used to extract the real objects and noisy blocks from background.Based on characteristics of target trajectory and target moving parameter prediction with Kalman filter,the real object detection and tracking is realized.The simulation experiment proves the feasibility and the effectiveness of the method.