High rise buildings are key components of urban infrastructure, and accurate monitoring of their dynamic responses under wind and seismic loads is essential for structural safety assessment. Ground based interferometric radar system has become an effective tool for displacement monitoring due to its high precision, non-contact operation, and full field measurement capability. However, practical measurements are often contaminated by thermal noise, atmospheric phase delays, and multipath effects, resulting in strong nonstationary and nonlinear characteristics that limit the accuracy of dynamic parameter identification and micro deformation extraction. To address this issue, this study proposes an adaptive denoising method based on Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) and a dual criterion of energy and correlation. The method first applies CEEMDAN to suppress mode mixing and separate multi frequency components. A joint energy–correlation criterion is then introduced to adaptively distinguish noise dominated and structural response dominated intrinsic mode functions (IMFs), enabling effective noise removal while preserving essential signal information. Using micro deformation data from the CCTV Tower as a case study, the proposed method is compared with wavelet denoising and Kalman filtering. Results demonstrate that the proposed approach effectively suppresses noise while preserving waveform characteristics, peak amplitudes, and energy distribution, providing a reliable solution for radar based structural monitoring of high-rise buildings.