Yet another half finished project

Some years ago I had the idea of being able to reproduce a 'Hawkeye' system on a single domestic camcorder. I believed that it would work but wasn't absolutely sure until I produced a prototype.

To be able to do this requires some fairly complex 3D mathematical transforms and it is an awfully long time since I used anything like that level of maths so it was a struggle. Added to which the video processing and encoding technology was way behind what we have today but more or less it works.

This is one the analyses.

 

 

All we need to know here is the height of the stumps (pretty much standard) and the distance between the stumps - 22 yards for adults. We can then create a transformation matrix to convert our model of what the ball is doing to the real world view from the camera - in effect we are determining the viewing point height and distance, zoom and hither and yon planes in 3D from a single 2D camera.

Once we know this matrix we can transform any 2D point - such as the bowler releasing the ball into our 3D matrix, thus by selecting a couple of points on the video, typically release point and pitch point we know exactly what the ball is doing. We have to of course add in some basic projectile theory and equations, an estimate of the coefficient of restitution of the mat (how much the ball will bounce), and some assumptions about air resistance.

From this we can produce the Hawkeye predicted path and the actual pitch map.

One day I will restart this project because I think it has huge potential as a coaching aid.