A new multi-related-targets approach is proposed to optimize swimmer tracking.
The classical tracking approach based on dynamic fusion is presented.
The swimmer is considered as a bloc of connected sub-targets.
The occluded sub-target can be localized according to the visible one.
Direct Linear Transformation approach (DLT) is used to calibrate the videos.
Our system is tested on swimmers in national and international competitions.
Two evaluation criteria: tracking percentage and instantaneous speed accuracy.
The findings show that our new approach is more accurate than the classical.