Wavelet analysis has been used for studying dynamic behavior of fluidized beds, which proved effective in resolution of time series into different scales of components with distinct structure and in identification of transition from the dense phase to the dilute phase. By examining wavelet spectrum functions of various dynamic signals measured from fluidized beds, it is indicated that the signals can be decomposed into three scales of components: micro-scale (particle size), meso-scale (cluster size) and macro-scale (unit size). The principal component method was employed for phase separation from concentration signals measured by the optical probe. In this method, the maximum scale parameter s(0) of the wavelet spectrum function was chosen as the optimum scale parameter. The principal component method can reduce the computation time significantly and remain the benefit offered by the direct method described in our previous publication (Ren & Li, in: L. S. Fan, T. M. Knowlton (Eds.), Fluidization, Vol. IX, Engineering Foundation, New York, 1998, p. 629.). The method was also extended to detect the boundaries of clusters in 2-D digital images acquired from fluidized beds.