Merging/Splitting Tracking of Resolvable Group Targets Based on ISODATA Algorithm

Resolvable group target tracking under merging/splitting conditions is a difficult task due to group structure changes. This article proposes an algorithm for tracking, merging, and splitting of groups founded on the ISODATA technique for iterative self-organizing data analysis and the generalized labeled multi-Bernoulli filter. The criteria for detecting group target merging and splitting were established: subgroup splitting is detected based on distances between group members and their corresponding subgroup centroids, while merging is determined by distances between the centroids of different subgroups. The proposed method can dynamically estimate the states and number of group members, the subgroup structures, and the number of subgroups within resolvable group targets. Simulation experiments demonstrate that the algorithm effectively estimates the states, quantity, internal structure, and number of subgroups of resolvable group targets in scenarios involving group merging and splitting.