Abstract: Traditional elliptical probabilistic damage imaging uses signal amplitude difference or energy difference as characteristic parameters, which is not sensitive enough to damage identification and has poor noise resistance. In order to improve the sensitivity of damage recognition, it was proposed to use the energy relative variation deviation (ERVD) of wavelet packet as the damage factor. A single Lamb wave mode was excited by an air coupled ultrasonic probe in the composite laminate with an appropriate incidence angle of the probe. Wavelet packet decomposition was used to decompose the collected scanning signal. According to the change of signal characteristics before and after structural damage, a specific frequency band was selected. The calculated damage index was used for elliptical probability damage imaging, and the imaging effects of different damage factors were compared by simulating different environmental noise environments. The experimental results show that the wavelet packet energy ratio deviation as the damage factor has strong sensitivity to damage recognition sensitivity and noise resistance. Using this damage factor for air coupled ultrasound probabilistic damage imaging can improve the location and imaging effect of composite material damage.