CODEWITHSUNDEEP
.netmatlabopencvvideo-processing
gpuguy
gpuguy

Reputation: 4585

Algorithm for speed measurement of moving objects in a video

You captured a video of a moving car using your camera equipped in a smart mobile phone. The video is very high resolution and assume that some loss-less encoding was done. (Obviously the size will be huge). It is being played at 60fps. What are the possible ways to measure the real speed of the car wrt to you from the video? (A perfect solution may not exist for all cases, so I am open for some reasonable assumptions under specific conditions, if any.).

If this is impossible, my next target would be to go for speed measurement of the car wrt to a stationary object in the video.

Edit:

In my case the Camera is always stationary, and the car is moving from left to right, and is moving at constant speed.

What method I have looked already

As the car moves, I move an object (a reference object) in front of the camera with some known speed. Then the video is having both the moving objects, one with known and another with unknown. by suitable calibration I can find the speed of the car. But I am looking for a solution where I dont have to move a reference object.

Upvotes: 2

Views: 5390

Answers (1)

Chris
Chris

Reputation: 8170

If your camera is stationary, and your car moves left-to-right at a fixed distance, then you should be able to track the car across the sequence, and work out how fast it is moving (in pixels per frame). Then you will need a one off calibration step where you work out what the conversion factor is to mph, for example you might work out that 100 pixels per frame = 30 mph, so you can then take your speed in pixels per frame and multiply by 30/100.

If you don't know the distance to the car (or rather, if the distance is not constant) then this method will not work, as a closer object will move across the image fast than a more distant object, even if they are moving at the same speed in real terms.

Upvotes: 2

Related Questions