Decoupling multi-source interference and time-varying parameters in laser echo signals for non-cooperative target detection

Light Detection and Ranging (LiDAR) has become a pivotal technology for non-cooperative target detection and characterization in space situational awareness, on-orbit servicing, and deep-space exploration, owing to its high spatial resolution and precise ranging capability. Nevertheless, complex dynamic space environments introduce coupled temperature-vibration interference and time-varying distance/incidence angle parameters, triggering severe nonlinear fluctuations in laser echo intensity. Conventional LiDAR reconstruction methods based on the ideal Lambertian surface assumption fail to decouple such nonlinear modulation effects, resulting in prominent deviations in material property inversion. To address this issue, this paper proposes an instantaneous state sampling and parametric decoupling method for laser echo signals from non-Lambertian surfaces. Specifically, a Wendland C 2 -Gaussian hybrid interpolation model is established to characterize the temperature-vibration coupling effect, and a momentum-enhanced stochastic Bregman proximal gradient (MSBPG) algorithm is developed for global optimization. Combined with the Cook-Torrance bidirectional reflectance distribution function for target surface modeling, the proposed method effectively decouples temperature-vibration coupling disturbances from time-varying spatial modulation factors. Validated on a semi-physical simulation platform with controllable temperature field and multi-axis vibration excitation, the proposed method achieves superior performance under typical compound interference conditions (353 K thermal radiation and 25 Hz mechanical vibration). Quantitative results demonstrate that the MSBPG algorithm reduces the training-set decoupling root mean square error (RMSE) by 34.2% compared with the conventional Bregman proximal gradient method, with a low test-set RMSE of 0.026. Moreover, the proposed approach decreases the RMSE of emissivity estimation by 65.6% relative to the traditional Lambertian-based correction method. This work provides a high-precision, physically interpretable technical solution for robust non-cooperative target recognition and dynamic material characterization in harsh space environments.