Abstract
Professional long drivers can hit golf balls to a speed of up to 350 km/h. Extremely short exposure time is required to capture the objects moving that fast without blur. The very high frame rate is also required to capture the sufficient sequence of object movements within the field of view in order to predict trajectory, which means a sacrifice in frame image resolution at fixed data throughput. This paper presents a high-speed stereo vision system that is able to track the golf ball motion at the speed of up to 360 km/h under normal indoor lighting conditions. Binocular camera operated at 810 frames/s with 35- $\mu \text{s}$ exposure time, requiring no additional strobe lights in order not to compromise user experiences. Ball markers were designed for spin measurement. To detect balls and identify markers from highly underexposed low resolution images, new algorithms were developed and implemented on field-programmable gate array board with an advanced RISC machines CPU. The experimental results showed high detection rate and high accuracy with overall processing time less than 300 ms.
Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
