Abstract: To address the problems of strong background noise, blurred defect boundaries, and coherent artifact interference in ultrasonic phased array inspection of lead seals in high-voltage cable accessories, which were caused by curved surface geometry, coarse-grained material scattering, and strong boundary reflections, an ultrasonic imaging denoising and artifact suppression method based on adaptive wavelet thresholding was proposed. In the method, an adaptive adjustment factor was introduced into wavelet threshold denoising to dynamically shrink wavelet coefficients at different scales. By combining amplitude fidelity of hard thresholding with the continuity of soft thresholding, background noise was suppressed while defect echo amplitudes and edge features were preserved as much as possible. Experimental verification was carried out on artificial hole-type defects in lead seals of high-voltage cable accessories with burial depths of 10, 13, 15 mm and 18 mm. Since experimental samples mainly focused on hole-type defects, the conclusions were primarily applicable to image denoising and boundary enhancement of hole-type defects in lead seals of high-voltage cable accessories; for more complex defect types, such as cracks and interfacial debonding, further verification using more practical samples was still required in future studies. Results showed that the proposed method significantly reduced background scattering noise and local pseudo-structures in ultrasonic images of curved lead-seal surfaces, and the continuity and interpretability of defect contours were enhanced. After denoising, an average peak signal-to-noise ratio of 25.79 dB and an average structural similarity index measure of 0.72 were achieved. These results indicated that the proposed method could effectively improve ultrasonic imaging quality of curved lead seals in high-voltage cable accessories, and a clearer image basis was provided for hole-type defect identification and subsequent quantitative image analysis.