一种适于信号压缩正弦模型的匹配跟踪方法

在基于包方式传输的网络中进行高频信号的实时传输时，利用信号模型对信号进行压缩有强大的生命力。基于帧结构的匹配跟踪原理，提出了一种有效而灵活的正弦模型分解方法。利用匹配跟踪算法，通过迭代抽取每帧信号的一组正弦成分，把这些成分通过帧重叠相加就可以重构原来的信号。使用高频域的快速傅氏变换算法（FFT）能有效快速地实现整个模型。本文提出的方法可用于信号压缩和调整，由于抽取的正弦成分是以人耳感觉的好坏排序的，因此该方法特别适合于音频信号的逐级压缩。

A Flexible and Efficient Sinusoidal Modeling Using Matching Pursuits Suited for Signal Compression

Signal models provide a general framework for attacking a wide range of challenges in the delivery of real_time audio over packet networks. In this work, we developed and implemented an efficient and flexible method for sinusoidal modeling using frame_based matching pursuits. Working on frames of the input signal, a set of sinusoidal components for each frame is iteratively extracted using extensions to matching pursuits algorithm. The frames overlap_add to reconstruct the original signal. The model formulation allows for an efficient implementation in the frequency domain via the Fast Fourier Transform (FFT). The developed sinusoidal model finds use in many applications including signal compression and modifications. In addition, this formulation is particularly well suited for scalable compression since the extracted sinusoids are ordered in terms of their perceptual significance.