一类矩阵奇异值分解的快速算法及在时变滤波中的应用

Fast algorithm for computing singular value decomposition of a class of matrices and its application to time-varying filtering

研究了镜象对称矩阵的性质，提出了计算该类矩阵奇异值分解的一种快速算法。计算机模拟结果表明：快速算法的计算速度比一般算法快３～４倍，并可节省一半内存。将该算法用于线性时变滤波器的设计，取得的结果是令人满意的。

Some properties of mirror-image matrices are investigated and a fast algorithm for the singular value decomposition of this class of matrices is proposed. Computer simulation results show that the computation required by the fast algorithm is reduced 3 times than the original one. Design problem of linear time-varying digital filters by using above fast algorithm is also considered.