A method is proposed for extracting a useful signal hidden in a determinist
ic chaotic signal using a wavelet multiscale decomposition algorithm that c
an separate two signals with different wavelet scales. The mask removal is
done without resorting to chaos synchronization. It is a noise reduction me
thod without knowing the dynamical model of chaotic system, and restricting
the modulation ratio of the chaos signal and useful signal. So this method
is effective for unmasking chaotic mask through the quantitative analyses
and the numerical simulations of three examples.