The full-field vibration displacement acquired via digital image correlation method is widely applied in aerospace structur al testing and monitoring because of its advantages of high environmental adaptability and non-contact full-field. However， structur al damage identification based on full-field vibration displacements faces two critical problems： Current modal analysis methods pos sess low computational efficiency when dealing with high spatial resolution displacement fields； The baseline-free damage identifi cation method based on modal shape is difficult to extract damage features and has poor anti-noise performance. In order to solve those problems， a frequency domain modal analysis method based on kernel decomposition and joint principal component analysis， and a damage localization method based on pseudo-excitation are proposed. Singular value decomposition is employed to process the full-field displacement fields for obtaining the kernel functions and their coefficients， which contain the local damage characteris tics. On this basis， a frequency domain modal analysis enhanced by joint principal component analysis is adopted to evaluate the noise-robust and high spatial resolution modal shapes. The disturbance of local dynamic equilibrium equation caused by structural damage is equivalent to a pseudo excitation force for damage detection. In addition， the local proximity of damage features and sparse spatial distribution of measurement noise are harnessed to optimize the damage localization accuracy via a hierarchical cluster ing method. Multi-modal information fusion damage index is proposed to improve the accuracy of damage localization. Numerical and experimental results demonstrate the effectiveness of the proposed method.