基于时频面旋转的线性调频信号增强

Enhancing chirp signal based on rotation of time frequency plane

估计淹没于噪声中的线性调频（ｃｈｉｒｐ）信号，是许多研究领域中的一个基本课题。本文提出一种基于时频面旋转的增强 ｃｈｉｒｐ 信号的算法，其处理过程是： 首先计算含噪信号的短时 Ｆｏｕｒｉｅｒ 变换，然后用由瞬时频率得出的信号的倾斜角度控制时频平面的旋转，在奇异值分解的基础上用低有效秩分离开信号子空间和噪声子空间，再将信号子空间旋转回去，最后经短时 Ｆｏｕｒｉｅｒ 逆变换，就得到去除了噪声的信号。仿真实验的结果表明了这种增强算法具有很好的保幅性能。

The problem of estimating the chirp signal immersed in noise is a fundamental one in many diverse fields. In this paper, a new algorithm for enhancing the chirp signal based on the rotation of time frequency plane is proposed. The algorithm works in the following procedure: first, calculate the short time Fourier transform of the noisy signal; this is followed by rotating the time frequency plane with rotation angle yielded from the estimated instantaneous frequency; then, on the basis of singular value decomposition, use a low effective rank to split signal and noise subspaces; finally, rotate the signal subspace back and perform the inverse short time Fourier transform. Simulation results show that the proposed algorithm has a good property of amplitude preservation.