Need some help with lenses

[Alden IA 0]

I am using a cctv camera for a scientific application and need to get a wide feild of view with minimal distortion as the images need to be used for measurement. I bought a 1.8 mm lens but the fisheye effect is to distorte dfor me to get accurate results. What is the widest angle lens before it starts to distort? My problem is I have a maximum mounting height of 16 ft (the distance between the roof and the surface i am studying) but would like to get ~30 ft of view. The camera need to point straight down as I ma tracking motion.

[bike_rider 0]

You may be stuck here. A quick check with a field of view calculator says that if you are using a 1/3" sensor, 16 feet from the target with a 30 foot field of view requires about a 2.6mm lens (which you probably can't find).

 

Simply put the wider the angle, the greater the distortion. Your choices are to use a mast to mount the camera (100 feet would need a 16mm lens), cover a smaller area or use more cameras.

[Alden IA 0]

I guess my question is at what point does a lens start being a fish eye lens?

3mm? 4mm? I don't know but figured someone else might. The mast idea is out of the question because I am indoors with a fairly low roof for about 2/3 of the area I need to cover. Any help woul dbe appreciated.

[Metal Shaper Man 0]

Just by spot checking the difference between some of my cameras with inexpensive lenses, I find the fish eye effect stops at a 6 mm lens and begins below that.

[bike_rider 0]

I guess my question is at what point does a lens start being a fish eye lens?

3mm? 4mm? I don't know but figured someone else might. The mast idea is out of the question because I am indoors with a fairly low roof for about 2/3 of the area I need to cover. Any help woul dbe appreciated.

 

At some point it is subjective what is distorted. It you think of the field of view of normal vision, that's something like 45 degrees I'd guess. Once you get wider than that, things look weird. More cameras may be your only solution.

[Metal Shaper Man 0]

I agree with bike_rider. If I’m right and the fish eye stops a 6 mm, you would have to have two cameras to cover the field of view or distance you require.

[CameraGimp 0]

There are some 360 degree cameras that computationally correct for fisheye lenses in real time, I know www.grandeye.com make one that is available through OEMs, I wouldn't use that to measure distances too accurately though but it shows how it could be done. They warp a fish eye into right angle perspectives for observational cctv not scientific measurements. When you start thinking about all the variables it gets a bit scary but if you have the know how to work out distances from a normal lens you may be able to calculate movement accurately even with a fish eye. As long as you fix the lens and the plane of the object you'd have to change your ratios depending on how far you are from the center line. Now I wouldn't know how how to start doing this by computer but I could draw some contour circles on the monitor that would give a rough indication of scale.

 

The grandeye cam does it with a whacking hot GPU my way is cheaper.

[Alden IA 0]

I am in the process of laying out a grid to "calibrate" the fisheye. By knowing the coordinates I could apply a correction to the data set. I am tracking floats in a scale model of a river to establish the surface flow patterns and velocities. I would like to avoid going to a 6mm camera because I would have overlapping feilds of view and processing would become tedious.

[photys 0]

If memory serves Axis have a dome cam using a 1/2 inch chip. Covers 140 degrees

http://www.axis.com/products/cam_212/index.htm

 

I remember seeing a demo and noticing very little fisheye.

 

Hunting for the demo.......

 

Edit:

Heres one.

Dark right now, possibly in Taiwan

http://ebdemo.8800.org:212/

[photys 0]

Apologies, there is a lot of fisheye on that cam, sadly.