Abstract: An adaptive K-distribution shape parameter estimation scheme based on fuzzy statistical normalization processing (FSNP) is proposed. In the proposed scheme, the reverberation is normalized from strongly to slightly by the FSNP method firstly. Then the shape parameters are estimated based on the different strength normalized data sets. Lastly, the estimated shape parameter value based on the proper normalized data is chosen as the final estimated value. Simulation results show that the proposed method has good shape parameter estimation performance for both inhomogeneous and homogenous K-distributed reverberation. Estimation results based on the same region reverberation from an active sonar with bandwidth of 4000 Hz show that the minimum value, the maximum value, the standard deviation, and the range of the estimated shape parameter values are 1.13, 25.76, 4.36, 24.63 respectively for inhomogeneous reverberation, and 2.54, 20.78, 3.84, 18.24 respectively for homogeneous reverberation. The standard deviation and the range of the estimated shape parameter values decrease significantly compared with the traditional shape parameter estimation methods for both inhomogeneous and homogeneous reverberation.