To address the lack of objective and effective evaluation methods for the covert performance of underwater acoustic communication systems，an approach that relies on wavelet packet transform is proposed in this paper. The proposed method extracts signal feature vectors and utilizes a similarity algorithm to assess the covert capability of signals. The proposed method capitalizes on the superior time-frequency resolution and adaptive-frequency-band selection of the wavelet packet transform to differentiate intercepted signals from environmental noise. It extracts feature parameters of intercepted signals，specifically the wavelet packet energy percentage，wavelet packet energy entropy，and wavelet packet scale entropy. These parameters are used to construct the feature vectors，and the cosine similarity algorithm is applied to determine the covert performance of signals. Simulation experiments and sea trials demonstrate that the concealment performance yielded by the proposed method aligns closely with the actual changes in covert performance concerning signal-to-noise ratio. As a result，this method offers an effective reference for evaluating and designing covert underwater acoustic communication systems and grading the acoustic countermeasure performance.