Is there a way to calculate speed of cars on camera

[Smilingreen]

You just need 2 long range reflective photoeyes, shooting across the road to a reflector of some sort on the other side of the road, with a known distance between the two eyes which would be tied into inputs on a data acquisition card on your pc. You will have a time stamp on when the beam was broken on the first eye and a time stamp on when the beam was broken on the second eye.

 

[wittaj]

88 feet per second (fps) is equal to 60 miles-per-hour (MPH), so 30MPH is equal to 44fps.

I just happen to have two pavement cracks that are 44 feet apart. With an overview camera in the middle, I can see both cracks easily and a decent frames-per-second and using BI, I can stop the video and simply take the time it takes to move between the two points (front bumper to front bumper). Would it hold up in court - probably not unless it was a painfully obvious double the speed limit. But it is a good check on traffic speed.

 

[Flintstone61]

If you can't lead em with a tripod mounted m-60 they're speeding.

 

[sdkid]

Pre-Radar Speed Trap...
My grandfather, who died in 1946, was the sheriff of a county in South Dakota in the 1930's. He put two lines of dirt across the blacktop road outside the small town that was the county seat. He calculated how many seconds it SHOULD take to go from one line of dirt to the next. He stood there with a stopwatch, and if a car went across that span in a time shorter than his calculation, he stepped out into the road and waved them down to give them a ticket.

Calculating with camera feed would certainly be a challenge-- especially if the road is at some odd angle to the camera line of sight-- which is effectively Always. I like what @wittaj wrote above-- noting the roadway cracks 44 feet apart. That would be similar to what my grandpa did in the 30's (though I think his lines were a hundred feet apart or so). There is geometry, physics, mathematics, and statistics involved in this-- would be interesting to have a 95% confidence interval calculated that the speed was between XX and YY mph.

 

[wittaj]

Yep, some states highway patrol is speed enforced by airplane or helicopter, so you may see lines going across the lanes or part of the lane. They simply do a time measurement for a known distance.

With Blue Iris showing time stamps down to thousands of a second, with a perpendicular to the street straight side view of a car with the known distance markers on the left and right, in playback, one can freeze frame move the video until the front of the car (or tire) crosses the first line and then slowly moving forward until it crosses the 2nd line and would be fairly accurate. Would it hold up in court? Who knows. But I think you can be fairly confident if you have the ideal setup.

 

[sdkid]

Yep, some states highway patrol is speed enforced by airplane or helicopter, so you may see lines going across the lanes or part of the lane. They simply do a time measurement for a known distance.

With Blue Iris showing time stamps down to thousands of a second, with a perpendicular to the street straight side view of a car with the known distance markers on the left and right, in playback, one can freeze frame move the video until the front of the car (or tire) crosses the first line and then slowly moving forward until it crosses the 2nd line and would be fairly accurate. Would it hold up in court? Who knows. But I think you can be fairly confident if you have the ideal setup.

I did not know you could adjust the time stamp to that level of precision! Awesome!

 

[wittaj]

Some wrong terminology going on there LOL. Parralax deals with changing speeds of items up close to items moving behind it. Or that cool effect where a still picture has the people "moving" while the background stays still.

These cameras are not doing that. I think that person was referring to the idea of parallel lines and that the distance of two straight lines on the screen will change distance between the two depending on where it is on the field of view - the distances between the two parallel lines on a camera screen will be smaller the closer the lines are to the camera and further apart the further out you do a measurement.

If you set it up with a good camera and high FPS and have that perfect perpendicular field of view to the street with two known distances clearly marked on the street, then you can get fairly close.

This is not my image but is from a solution someone created. I just hand calculate mine for anyone I am interested in, but with Blue Iris, I can move it frame by frame and line up within a few inches the front bumper to both of the lines and then subtract the time it took to go between both marks.





Here are a few threads and other sites where people have done just that.

Vehicle speed determination from video

I would love to know how fast cars are driving past my home. With a 2.5yo and 7mo, what used to be of no concern to me, has suddenly taken on a new perspective. Assuming I knew the precise distance between two points, and calculated the time to travel between them, I'm sure I could manually...

ipcamtalk.com

Is there a way to calculate speed of cars on camera

If I spray painted a couple of marks (x?) feet apart on the street, is there a way to calculate the speed of the car?

ipcamtalk.com

AI vehicle speed detection solution

Software to detect cars or any vehicles in any video and calculate their speed in any video stream by estimating counting lines and Camlytics camera events API. CCTV IP, DVR, NVR supported. Offline video analytics solution.

camlytics.com

Raspberry Pi As Speed Camera

Wherever you stand on the topics of road safety and vehicle speed limits it’s probably fair to say that speed cameras are not a universally popular sight on our roads. If you want a heated ar…

hackaday.com

 

[sdkid]

My argument is that you can’t accurately determine someone’s speed from a security camera with low fps at night that’s almost diagonal to the reference point and then without using parralax his speed measurement would be off by 10-50 percent. Would this be correct?

I think you're on the right track...
If the pole and the mailbox were the landmarks used, a car coming from the right to left would "hit" the mailbox starting point FARTHER AWAY-- farther down the street (look where the line crosses the curb on the near side versus the far side), and would need different calculations for accuracy, or their speed would be OVER-estimated.
(not my pic--- random photo from some site)

 

[wittaj]

^Yeah, anything other than a perpendicular to the street field of view can be problematic, but even above, one could "calibrate" it by driving between the two points as you would measure them at known speeds to get a better estimation of the speed. Would it hold up in court? Probably not but who knows. Maybe they come out and do exactly that at different speeds caught on this camera for the case.

But when you have two cracks or marks in the road that do not move and you have a perpendicular angle and a VMS that you can crawl through the footage with the time elapse on showing down to thousands of a second, you can stop it when the front wheel hits the first line and write down the time. Then crawl it until the front tire hits the second line and write down that time and then do the math. It isn't perfect, but can get pretty close. If this were for an accident case, I am sure the prosecution would fight like you know what to use this basic math function to estimate a speed. Would have to be more accurate than measuring the length of skid marks to estimate the speed and that was used in court all the time...

 

[SyconsciousAu]

See in the case, the prosecution is using a low fps camera (approx 15) and on an extreme angle which has the cars taillight stretched over half a meter. They’re also not using any equations in the matter to back up their calculations, only speed equals distance over time which is correct but there is a lot more calculations that’s needed for their argument. They’re also saying the parralax error makes one car look 76 and the other 100 when they’re travelling the same distance apart over 70 meters. I’d suggest anyway calculating speed measurement to check this error

Not sure if your matter has finalised yet but my advice would be, given that looks like a scene reconstruction from a Crash Investigation Unit, is to get some really good legal advice and employ a local expert to check their numbers if you have any concerns. The Police who I suspect you are going up against will have specialist training and knowledge which will often lead to the court affording them the weight of an expert opinion. The downside of course is that experts cost a bomb and may very well confirm the data that you already have. There is also the question to be asked (and do not discuss this with your solicitor if you have one, trust me they do not want to know) were you going the speed alleged, or more?

The fact that CIU is investigating tells me that this isn't some sort of minor speeding matter and that someone was badly injured, or worse, killed. Charges may be as serious as manslaughter but as a minimum probably dangerous driving occasioning death/gbh with alternate of negligent driving occasioning death/gbh. These are the big leagues and jail is a realistic possibility if you are convicted following a defended hearing. If you know that you were going as fast as they say (and I cant emphasise enough that you do not tell your lawyer this because it limits the arguments they can lawfull make in your defence) you need to ask yourself if it is worth taking the risk that they can prove it? Otherwise it may be worth considering offering to plead guilty to the lesser offence, taking the discount (for our American readers in NSW Australia an early plea of guilty can get you up to a 25% discount on sentence), and giving yourself the best chance you have of getting out of there with a community based order.

As always IANAL and you should seek legal advice who, if they are competent, should have canvassed this with you already.

 

[bradner]

What about this somehow? under $100