Image Processing Algorithms to Extract Swimming Tumble Turn Signatures in Real-Time

Vision based methods provide a powerful technique for accessing swimmer performance. Currently, manual analysis of recorded images is used to determine performance-relevant metrics, which incurs significant time and effort for the analyst. Real-time automated image processing algorithms to measure a swimmer's performance for tumble turns is detailed in this paper. The algorithms yield performance parameters including contact time, foot orientation, distance between feet, lane position and depth of contact with the pool wall. Pressure sensors have been employed to measure pressure exerted on the wall and wearable LED markers are used to track movement of the swimmer during the turn. The system significantly improves the efficiency of the analysis process.