滑动置信度约束的红外弱小目标跟踪算法研究

发布时间：2018-03-08 09:50

本文选题：信息处理技术　切入点：红外小目标　出处：《兵工学报》2017年09期 　论文类型：期刊论文

【摘要】：为了提高红外视频弱小目标的跟踪精度,提出了滑动置信度约束的弱小目标跟踪方法。在快速自适应中值滤波的红外图像背景抑制技术的基础上,设计了正交变换和置信域约束的轨迹预测,利用加权参数增强目标函数的收敛性能,提高下一位置初的预测准确度;通过轨迹相邻点的位置差计算搜索窗口的大小,搜索与之相匹配的特征点进行关联处理,完成对初预测点的筛选;以滑动轨迹置信度检验为准则判决轨迹的真实性,并进行目标轨迹更新以实现对弱小目标的准确跟踪。通过红外弱小目标视频对所提算法进行了实验验证,结果表明,该算法对红外弱小目标的跟踪轨迹误差有较小的均方偏差与均方差,在噪声消除和对图像整体信息保护方面都具有良好的性能。
[Abstract]:In order to improve the tracking accuracy of dim and weak targets in infrared video, a sliding confidence constrained tracking method is proposed, which is based on the background suppression technique of fast adaptive median filter. The trajectory prediction with orthogonal transformation and confidence region constraints is designed, the convergence of the objective function is enhanced by weighted parameters, the prediction accuracy of the next initial position is improved, and the size of the search window is calculated by the position difference of the adjacent points of the trajectory. Search the matching feature points for correlation processing, complete the screening of the initial prediction points, and judge the authenticity of the trajectory based on the confidence test of the sliding trajectory. The proposed algorithm is verified by the infrared dim target video, and the results show that the proposed algorithm can be used to track the small and weak target accurately, and the experimental results show that the proposed algorithm can be used to track the small and weak target accurately. This algorithm has small mean square deviation and mean square deviation for tracking track error of infrared dim target, and has good performance in noise elimination and overall information protection of image.
【作者单位】：北京科技大学自动化学院;
【基金】：武器装备“十三五”预先研究基金项目(61404520101) 中国博士后科学基金项目(2016M600922) 中央高校基本科研业务费专项资金项目(FRF-TP-15-117A1) 高分辨率对地观测系统重大专项项目(2015年)
【分类号】：TP391.41

【相似文献】