A distributed sensor-based method for tracking and localization of space target groups

Abstract

Low-orbit infrared sensors are an important means of space exploration and hold significant importance for space security. However, the detection range of individual satellites is limited, presenting challenges in fulfilling the task of continuous indication of targets. Joint exploration using multiple sensors is a more effective choice. Therefore, we propose a multi-target tracking and localization algorithm based on cooperative detection using multiple infrared sensors. This algorithm enables an integrated process from image plane tracking to three-dimensional spatial localization of small target groups. Firstly, we propose a multi-target tracking method using an improved discriminative correlation filter as the tracker. The method sets an energy concentration threshold based on the characteristics of infrared small targets to suppress background noise. Simultaneously, the minimum Euclidean distance and velocity similarity between consecutive frames of targets are used to associate the trajectories, effectively reducing association errors. In addition, an adaptive extended Kalman filter algorithm is synchronized to predict the target positions, addressing the challenge of small targets being easily occluded. Subsequently, an adaptive weighted covariance intersection fusion algorithm is employed to integrate multi-sensor information of tracking, effectively mitigating the issue of reduced localization accuracy caused by instability or tracking errors in individual sensors. Experimental results show that the mean Optimal SubPattern Assignment of the proposed tracking method is less than 0.2 pixels in simulated multi-target scenarios. The proposed multi-sensor fusion algorithm ensures localization accuracy within 44 m for detection ranges exceeding 4000 km. This highlights its potential in the fields of space exploration and target indication.