The low temporal resolution of stochastic optical reconstruction microscopy (STORM) limits its ability to observe dynamic events in live cells. Further, the post-processing analysis and reconstruction algorithms have an important effect on super-resolution images. In this study, we report a new noise-correction principal component analysis method for single-molecule localization microscopy against fluorescent spot overlapping and excessive background noise in a single frame of images owing to high-density labeling and high camera-sampling frequency. The proposed method can improve the positioning accuracy of existing localization methods by pre-processing the raw images acquired by the single molecule localization microscopy before reconstruction. In addition, this method can accurately distinguish the overlapping molecules. Therefore, it is suitable for samples exhibiting a high fluorophore density. Thus, the proposed method improves the temporal resolution of super-resolution imaging, providing a powerful technical support for the STORM imaging of live cells.