PASTd算法应用于机动多目标角度跟踪

PASTd Algorithm Implementation for Multiple Moving Angle Tracking

Ryu et al.提出一种基于Kalman滤波和信号子空间的机动多目标角度跟踪算法,在该算法中需要计算信号子空间矩阵W的投影矩阵,因而需要计算N×N维复数逆矩阵,这主要是由于W的列向量间不正交。提出一种用Kalman滤波预测的角初始化W的方法,使得在用PASTd算法时W能够更快地收敛于列向量为正交向量的矩阵,从而避免了计算N×N维复数逆矩阵,既降低算法的运算量,同时跟踪性能也得到提高。

Ryu et al. recently proposed an effective multiple moving target angles tracking algorithm using Kalman filter and signal subspace,in which the premiss is the projection matrix of Signal subspace matrix W,that is,we need to compute the N×N dimension complex inverse matrix, for W is not a matrix with exactly orthonormal columns. In this paper, we proposes a new approach using the predicted angle obtained by Kalman filter to initialize W,which makes W converge to a matrix with exactly orthonormal columns rapidly. As a result,the computation of N×N dimension complex inverse matrix is avoided, namely,the caculating amount is reduced while the tracking performance is enhanced.