一种改进的基于奇异值分解的信源数目估计算法

信源数目估计问题在盲源分离中具有重要的意义。研究了传感器数目大于信源数目时的源数估计问题。首先分析了用奇异值分解法进行信源数目估计的优势与不足，然后提出了一种改进的基于奇异值分解的信源数目估计算法。该算法首先对含噪混合信号进行奇异值分解，然后检测信号分量与噪声分量之间的转折点，将信号分量与噪声分量区分开来，从而得到信号源的数目。实验仿真表明，该算法在低信噪比以及采样点数较少时仍然具有好的性能。

An improved algorithm based on SVD for blind estimation of the number of sources

Estimating the number of sources is an important problem in blind sources separation. This paper mainly studies the problem of estimating the number of sources when the number of sensors is greater than the number of sources. The advantages and disadvantages of using singular value decomposition(SVD) to estimate the number of sources are analyzed firstly, and then an improved algorithm based on SVD is presented.In the algorithm,the singular values of the noisy mixtures are obtained by using SVD, and then the turning point between the signal components and the noise components is detected, which can distinguish the signal components and the noise components to get the number of sources. The simulation shows that the algorithm still has good performance in low signal-to-noise ratio(SNR) and fewer sampling points.