1 Topic by mjwall

  • mjwall
  • -
  • Offline

I'm pretty new to all this so please let me know if there is a better place to ask the questions....

Say I place a single camera and record an object (swimmer) which passes across the frame and only moves in the X and Y direction at a fixed distance in the Z direction, say 1m. If I also have a reference object which is a 1m (real world length) long stick hanging perfectly vertically (Y axis) in frame at a Z distance of 2m. Using this information about the stick, is it possible for me to calibrate the Y co-ordinate system and apply that to the swimmer (at 1m Z distance) as they pass the camera to measure something in the Y direction, such as their depth in the water? I've seen some calculations using the lens focal length to compute the apparent size at a given distance.

My other thought was to use control points on the swimmers body, such as stickers at the ankle and knee. If I know the distance between the stickers then I can calibrate both the X Y coordinates, particularly if the known dimension is oriented diagonally. I guess this is also better because the Z distance is pretty much irrelevant, I can directly compare measurements between two different trials where the swimmer is slightly further or closer to the camera without extra math.

Sorry, I guess this is a pretty basic concept but I'm pretty unprepared smile

MIke

2 Reply by mjwall

  • mjwall
  • -
  • Offline

I've done some digging into this and it is very interesting.

Due to the way the 3D world plane maps to the 2D image plane it should be theoretically possible, in ideal conditions, to use a distant object to to calibrate dimensions for an object placed closer in Z direction (or vice versa).  An object in world plane of known height h at a known depth z will have an apparent height in the image plane proportional to h / z. So, an object 2m tall placed 2m away from the lens has the same apparent size as an object 1m tall placed at 1m distant - for the same camera configuration. Therefore if you calibrated the length of the 2m object (at the 2m distance) in kinovea as being only 1m in length then you should be able to directly measure the physical dimensions of any new object at a distance of 1m from the camera. 

Not entirely sure why this might be useful t anyone. But I do know that neurodivergent monotropism is fun smile