Modal-based damage identification methods, particularly those utilizing mode shapes, excel in damage localization due to their spatial distribution characteristics. Traditional contact-based sensors often struggle with achieving high spatial resolution, while optical dynamic measurements, though high-resolution, are susceptible to environmental noise. This study proposes a novel multicomponent information separation (MIS) approach to overcome these challenges. The method decomposes mode shape information into noise, damage-related features, and waveform trends. The methodology begins with optical dynamic measurement to capture high-resolution mode shapes, followed by two-dimensional continuous wavelet transform (2D-CWT) for multiscale analysis. Adaptive wavelet scale selection via scale correlation analysis (SCA) improves noise separation, and an iterative weighted least squares fitting (IWLSF) method accurately fits the waveform trends. Finally, a noise-robust and baseline-free damage indicator (DI) is constructed through a difference operation to eliminate waveform trends, with damage localized at DI abrupt change positions. Validated through numerical simulations and experiments on damaged aluminum plates, the method demonstrates superior noise robustness and localization accuracy for multiple minor damages compared to existing wavelet-based approaches, showcasing its potential for advancing damage detection techniques.