hippy wrote: ↑
Tue Apr 30, 2019 1:04 pm
The problem comes in how one defines 'crossing the line'. If it's the point an athlete's torso crosses the line as opposed to head, hand, foot, arm or leg, then how does one distinguish one from the other ?
I would say that regardless what first passes the line (foot, head, ...) that needs to be measured.
That is the reason why a single laser cannot be the solution.
You would need a dense net of laser beams as shown in bank or museum robbery movies.
I would use v2 camera placed on one side of the finish line with center of FoV direted along the finish line.
Then take a video with raspivid/raspiraw and high framerate.
It depends on the measurement resolution on what you can take.
If measuring resolution 1/100th second is needed, use raspivid with mode 7 and "--fps 100".
If measuring resolution 1/1000th second is needed, use raspivid 640x75 tool that captures at 1007fps
(v2 folder of code drop):
I would turn the camera so that 75 is horizontal and 640 is vertical.
640x75@1007fps can be stored and analyzed later, or during capturing as described in this posting:
"Re: raspiraw raw bayer data: how to use in callbacks for feature extraction and robot control"
I am working on making v1 camera take global shutter videos by global reset and cutting very bright light after few microseconds. Alternatively I have an Arducam ov7251 global shutter camera.
I did take videos with very sharp propeller blade moving at blade tip with 125m/s rotational speed.
Later I will take sharp videos of a flying bullet, this posting explains how a first Pi with camera will monitor on when bullet leaves rifle, and trigger a second Pi with camera to capture bullet frame exactly at the right time (bullet flies with eg. 375m/s, that is 0.375mm/µs):
640x75 tool creates a microsecond precision timestamp for each and every frame captured, no assumptions needed wrt at which time a frame was captured.
This is a sample 640x75 tool generated animated .gif, the mouse trap closes in 0.01s, which results in seeing the mouse trap bar in closing phase on 10 frames of the video captured at 1007fps: