In 2015, the University of Bath completed a refurbishment of the 50 m swimming pool used to train top athletes. Part of the re-fit included an imaging system commissioned by British Swimming and UK Sport for the purpose of advanced training, which is used to improve swimming techniques, including those competing at the Rio 2016 Olympics.
The vision hardware was supplied by Multipix Imaging whilst the bespoke software was written by Dr Simon Goodwill, a research engineer in Sports Engineering at Sheffield Hallam University. He has developed a range of customised software applications which link data acquisition hardware with camera systems which are being used to help improve an athlete’s performance.
The vision hardware is primarily based on cameras with a resolution of 2048 x 1000pixels @ 50fps. There are 12 cameras along the length of the lane and 1 camera each end of the lane where the swimmer enters the water. In addition, there are three cameras placed high above the pool for an overhead view of the lane.
Each camera is using Kowa LM5JCM, 5mm megapixel lens, as it was important to achieve the widest field-of-view possible due to the change in refraction index from air to water. To ensure camera protection, each is housed in a AutoVimation ORCA IP67 enclosure.
All cameras must be synchronised which, until very recently, would have required additional trigger cabling to run to each camera and a common trigger source. For installation, where space is constrained, the thought of running multiple cables over 50m+ distances is not ideal. However, this was simplified greatly with the use of GigeVision cameras supporting PTP and PoE. The extremely neat solution means only the CAT6e cable is required to run from each camera back to a NETGEAR switch which is connected to the 19” rack PC for image capture, analysis and replay. PTP (Precision Timing Protocol) was introduced to GigeVision 2.0 and is a way to synchronise cameras using the Gige cable interface and eliminates the needs for traditional triggering via a dedicated hardware port on the camera.
This is a great example of where industrial vision components are being used in sport analysis and the growing trend of non-industrial imaging solutions.