Abstract: With the rapid development of commercial spaceflight, the demand for rocket flight attitude estimation based on external measurement data is becoming increasingly urgent. This paper, based on an in-depth study of the characteristics of High-Resolution Range Profiles (HRRP) of rockets, employs a truncated mixed Gaussian model based on the idea of parametric representation to complete the reconstruction of rocket HRRP. This approach achieves rocket flight attitude estimation under two conditions: with and without a template library. Tests using simulated HRRP sequences and actual flight HRRP sequences demonstrate that the reconstructed HRRP sequences have the advantages of high signal-to-noise ratio and strong stability. By employing methods such as multi-envelope detection, range template matching, and length projection, stable implementation of rocket flight separation detection and attitude estimation based on HRRP data is achieved.