Wind turbines are costly and difficult to repair in the field of new energy. Monitoring their health status and mastering the characteristic of blade fatigue crack propagation have been an important topic. A combining method of optimized reassigned scalogram and wavelet energy coefficient is presented to analyze wind acoustic emission (AE) signals of turbine blade crack propagation. Basis function bandwidth of reassigned wavelet scalogram is calculated based on Shannon entropy. The most suitable basis function for AE signals of propagation crack is attained. Therefore, the optimization reassigned wavelet scalogram of crack AE signal has high amplitude energy distribution in time-scale plane. Then, energy spectrum coefficient can be used to the optimization reassigned scalogram. Experimental research proves that the proposed method has excellent time-frequency concentration and noise restraining ability. It is achieved to extract the time-frequency characteristics of the wind turbine blade crack propagation of acoustic emission signals clearly. Energy spectrum coefficient as feature vector can show the characteristics of signals. Moreover, this method can be applied for real-time pattern recognition of blades in complex environments.