<p>Parametric identification for health monitoring of civil infrastructures using wavelet-based multi-channel signal processing is the theme of this work. For this purpose, the wavelet transform of a measured signal, referred to here as a channel, is adopted, which offers excellent time-frequency localization. However, the scalogram obtained from this tool is often smeared, ultimately affecting the quality of the identified parameters. Moreover, individual channel provides different values of modal parameters, leading to subjectivity in system identification. To address these issues, two different algorithms have been proposed in this paper involving a multi-channel version of the continuous wavelet transform using a reallocation technique, a.k.a synchrosqueezing. The first option proposes a multi-scale, multi-channel non-parametric regression scheme for unique modal identification offering improved robustness against noise and inconsistencies across sensor channels. In contrast, the second extends an existing multi-channel singular value decomposition of the coefficients obtained from synchrosqueezed transformation leveraging an eigenvalue-based approach to enhance modal coherence and reduce uncertainty in extracted parameters. These methodologies significantly improve accuracy in detecting modal frequencies and tracking structural anomalies compared to conventional individual channel methods. The proposed algorithms are validated using different examples, i.e., sinusoids, chirp, laboratory test data of a 3DOF system and damage identification of a simulated beam. Finally, a full-scale reinforced concrete bridge is considered, where ambient vibration data from different sensors are used to investigate its in-situ structural condition. All these results highlight the efficiency and accuracy of the proposed algorithms for the health monitoring of existing structures.</p>