The composite C-beam will undergo springback deformation after forming. The calculation and analysis of the deformation amount using the scanned data are prone to mismatches, which make it impossible to accurately describe and evaluate the actual deformation amount. In view of this, a C-beam point cloud registration and deformation detection method based on the constant characteristics of the web area and the symmetry plane is proposed. According to the deformation characteristics of the C-beam, the springback deformation is quantitatively described, and the calculation method of the deformation amount is given. Principal component analysis and iterative nearest point algorithm are used to extract the symmetry plane of the C-beam to obtain the symmetry degree of the C-beam. Combined with the point cloud constraints of the web area with small deformation, the measurement data is registered with the model data, and the global deformation distribution of the C-beam is obtained. The simulation point cloud of C-beam with springback deformation is constructed, and the deformation amount is calculated by the proposed algorithm. The results show that the proposed algorithm can realize the calculation of the deformation amount, the relative error of the maximum springback deformation is 0.0461%, and the deformation amount is consistent with the simulated set value, and the detection of the C-beam deformation amount is realized.