基于运动分类和二维卷积的随机Hough变换航迹起始

Track Initiation of Randomized Hough Transform Based on Motion Classification and Two-dimensional Convolution

随机Hough变换具有占用存储空间小、计算量低的优点,常用于低杂波环境下的航迹起始。针对随机Hough变换在密集杂波环境下存在峰值簇拥、计算量大导致航迹无法正确起始的问题,文中提出了一种基于运动分类和二维卷积的随机Hough变换航迹起始方法。首先,利用距离约束限制无关点迹做随机Hough变换;然后,对不同运动朝向的点做随机Hough变换,产生的特征参数分别用矩阵存储,为避免测量噪声导致航迹特征参数得不到积累,对每个特征参数矩阵做二维卷积;最后,将过阈值门限的特征参数点进行聚类处理完成航迹起始。验证结果表明:所提方法能在密集杂波环境下有效地起始多目标航迹,不仅成功率高,而且计算量小,具有较好的工程应用前景。

Randomized Hough transform has the advantages of small storage space and low calculation amount,so it is often used for track initiation in a low clutter environment.Aiming at the problem that randomized Hough transform in the dense clutter envi-ronment has the disadvantages of peak clusters and large amount of calculation which cause the track start uncorrectly,a random-ized Hough transform track initiation based on motion classification and two-dimensional convolution is proposed.Firstly,the dis-tance constraints are used to limit irrelevant points to do randomized Hough transform.Then the feature parameters generated after randomized Hough transform on points of different moving directions are stored in different matrices.In order to avoid measurement noise causing the track feature parameters not to be accumulated,every feature parameter matrix is subjected to two-dimensional convolution.Finally the feature parameter points that have passed the threshold are clustered to complete the track initiation.The verification results show that the proposed method can effectively initiate multi-target trajectories in dense clutter environment,with high rate of success,small calculation,and good engineering application prospect.