<p>Locked-segment-type rock slopes (LSRS) are large-scale rock slopes controlled by one or more locked segments, which critically affect the overall stability of LSRS. However, it is generally challenging to identify the location of the locked segments before a landslide occurs due to their highly concealed nature. In this paper, we proposed a comprehensive solution for inverting the long-term deformation of LSRS by combining Multi-Temporal Interferometric Synthetic Aperture Radar (MT-InSAR) and geometric deformation calculations to facilitate the early identification of the locked segments. First, we utilized MT-InSAR technology to obtain multi-track line-of-sight deformations of the slope. Second, we developed an innovative method for the spatiotemporal registration of multi-track deformation data. Third, we established a slope coordinate system to decompose the true deformation. Finally, we applied singular spectrum analysis (SSA) to separate the trend and seasonal components. Using the Xinmo landslide as a case study, our research is the first to accurately identify the locked segment of the landslide through remote sensing data. We observed the locked segment 7&#xa0;months before the landslide failure from the decomposed deformation map. Additionally, our study reveals the spatiotemporal characteristics of LSRS deformation influenced by anti-dip joints, and SSA highlights the varying acceleration effects of rainfall on slope-parallel and slope-normal deformations. The proposed methods demonstrate high accuracy and applicability in exploring the fine deformation mechanisms of locked segments and provide valuable insights for the early identification of locked segments in large rock slopes.</p>