-
Semper Vaporo
-
- Posts: 331
- Joined: Sat Oct 09, 2010 11:51 am
by Semper Vaporo » Tue Feb 18, 2020 4:58 pm
I have done this manually a few times. I know the distance from two points in the street (cracks in the pavement) and can time cars passing the two points.
But, note that the best SightHound can do with timing is in tenths of a second (the frame rate is 10 frames per second). Thus it would have up to 0.1 second error in the time measurements -- a car might have not quite reached the trigger point (crossing a line) at the start point in one frame, but in the next frame it might be well past that point... so at what time did the car pass the trigger point? The same can happen at the other end (the stop point).
Also, due to parallax error, the distance traveled will vary based on whether the car is on the far or near side of the street.
Manually I overcame those problems by measuring the actual distance traveled over 20 to 30 frames (the more frames the better). But SightHound cannot do that, you are stuck with fixed points and snapshots of the movement where the moving object may not be positioned at the start and stop points the moment the frame is captured. I was able to say the car was 9ft. past one crack in the street in frame 1, and 11ft. from a crack 60ft. away from the 1st in the 20th frame, thus it took 2 seconds to traverse 50ft. thus was traveling at 17 MPH. A bird 5ft. from the camera might travel only 10ft. in 2 seconds because it is closer to the camera flying only 3.4MPH. SightHound cannot tell if the object was in the street or right in front of the camera.
But I still had a problem of the car being blurred due to motion so it was hard to tell exactly how far it was from the cracks in the street, so my speed calculation was not all that accurate.
BTW, double-check your calculations in your drawings... if a car takes 10 seconds to traverse a distance and you say that is 10 MPH, then a car taking only 8 seconds for that distance would be going FASTER not slower.
Semper Vaporo,