The particle velocities are measured by the high-speed particle image velocimetry (PIV) in the acceleration and fully developed regimes of a horizontal pneumatic conveying. Based on the measured particle fluctuation velocities, continuous wavelet transform and one-dimensional orthogonal wavelet decomposition were applied to reveal particle dynamics in terms of time frequency analysis, the contribution from wavelet level to the particle fluctuation energy, spatial correlation and probability distribution of wavelet levels. The time frequency characteristics of particle fluctuation velocity suggest that the small-scale particle motions are suppressed and tend to transfer into large scale particle motions from acceleration regime to fully developed regime. In the near bottom part of pipe, the fluctuation energy of axial particle motion is mainly contributed from the wavelet levels of relatively low frequency, however, in the near top part of pipe, wavelet levels of relatively high frequency make comparable contribution to the axial particle fluctuation energy in the suspension flow regime, and this contribution decreases as particles are accelerated along the pipe. The low frequency wavelet levels exhibit large spatial correlation, and this spatial correlation increases as the particles flow from acceleration regime to fully developed regime. The skewness factor and kurtosis factor of wavelet level suggest that the deviation of Gaussian probability distribution is associated with the central frequency of wavelet level, and the deviation from Gaussian distribution is more evident as increasing central frequency. The higher wavelet levels can be linked to small sale particle motions, which lead to irregular particle fluctuation velocity. (C) 2018 The Society of Powder Technology Japan. Published by Elsevier B.V. and The Society of Powder Technology Japan. All rights reserved.