In this paper, a novel homomorphic filtering method based on partial time-domaincoherence algorithm is introduced.By this way, we can quantitatively analyze the mixed signal and its signal components,decompose the mixed signal into expected signal components whether each of theirfrequency-bands overlaps or not. And each of the decomposed components can be eitheruseful or noise signal.According to partial time-domain coherence algorithm, let the mixed signal iscomposed of n multi-frequency signals, we can make the mixed signal decomposed into ncomponents in the condition that n-1 reference signals whose waveforms are similar to n-1signal components can be obtained regardless of their amplitudes and phases.And apply it in homomorphic filtering to decompose the product of two or moresignals by some subtle transforms.Finally design a software to examine the algorithm, by analyzing the results ofexperiment, we can draw the conclusion that this software can be operated easily andreliably. |