A novel discrimination mechanism based on GMM for high-precision motion estimation of instability space targets

CL Guo and RM Sun and ZQ Zhou, MEASUREMENT SCIENCE AND TECHNOLOGY, 36, 115110 (2025).

DOI: 10.1088/1361-6501/ae1c5e

Measurement noise, being inherent and unavoidable, poses a significant challenge to high-precision motion estimation of non-cooperative space targets, especially in a linear measurement system. To address this issue, we propose a novel discrimination mechanism based on the posterior probability of potential corresponding points, modeled via a Gaussian distribution. This mechanism determines noise intensity through a key metric that maps positional correspondence between the current point and potential matches across an entire point cloud sequence. This mechanism makes it highly effective for the motion estimation of dynamic targets under noise interference. Under conditions of low-noise interference, the expectation maximization algorithm is employed for iterative optimization by refining the mean-mapping process and reconstructing virtual points to replace the original measurement points, yielding high-precision motion parameters. Under high-noise interference, the improved grey wolf whale optimization algorithm is used to optimize the key quantity values of corresponding points and correct the initial values deviated due to noise interference, achieving a significant improvement in motion estimation accuracy. Experimental results show that our method achieves a accurately and effectively suppress the influence of different noise interferences on motion estimation, and in most cases of different noise intensities, the error reaches the order of 10-2.

Return to Publications page