To address the issue of noise interference in mining blasting vibration signals, a joint denoising algorithm combining the Crested Porcupine Optimizer (CPO), Improved Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (ICEEMDAN), and an enhanced wavelet thresholding method is proposed. First, the CPO algorithm was employed to globally optimize the key parameters of ICEEMDAN, adaptively decomposing the measured blasting vibration signals from an open-pit mine into a series of intrinsic mode functions (IMFs). A noise identification threshold was then constructed using multiscale permutation entropy (MPE) to screen out high-frequency noisy IMF components. These components were processed with the improved wavelet thresholding method and subsequently recombined with IMFs below the threshold to achieve denoising. Comparative experiments with CEEMDAN-MPE and ICEEMDAN-traditional wavelet thresholding methods demonstrate that the proposed method improves the average signal-to-noise ratio (SNR) by 39.33% and 19.93%, respectively, and reduces the root mean square error (RMSE) by 2~3 times across three sets of vibration signals. Furthermore, three-dimensional time-frequency energy analysis reveals that the main frequency energy distribution remains unchanged before and after denoising. These results indicate that the proposed method not only effectively eliminates noise interference but also fully preserves the main frequency energy characteristics of the original signal, demonstrating superior performance and engineering applicability.