Poll: Would you be interested in an inexpensive Wide-Angle lens for the Axis Webcam?

[Tom Line]

This year, some teams modified their axis webcams through a variety of means to add a wide-angle lens to them so that they could more easily track other people's trailers.

If an option were available for you to purchase a lense, prefabricated and easy to install, would you be interested? What price point would you consider?

(This is my first time trying to post a poll, so if it does not work, don't lambaste me too badly).

[sdcantrell56]

As one of the teams who succesfully implemented a wide-angle fisheye lens for the axis camera, I will say that there is already a widely available lens for ~$30. We purchased ours from Fry's as the axis camera uses a standard mount for the lens.

[Tom Line]

sdcantrell, was the lens that you guys decided on the one in the two lens sakar kit?

[sdcantrell56]

It was a one lens package. I dont believe it was listed online but I found it in the security camera section.

[Greg McKaskle]

http://www.edmundoptics.com/onlineca...productID=2196

The have several quality levels for lenses. This is the primary link I use.

Greg McKaskle

[Nick Lawrence]

A wide angle lens would be nice, for ~ $30.

Man, the Axis 206 has the narrowest FoV I've ever seen, save a pinhole in a wall.

-Nick

[Tom Line]

I talked directly with Axis, and their standpoint is that this camera has a Fixed lens and they are not planning on making any attachments or different lenses. That's one of the reasons I'm pursuing this. Thanks everyone for your responses.

[PhilBot]

Hi

I'm embarrased to say that even though I passed optics in physics I don't know how focal length relates to FOV.

Any recommendations on how to choose between so many options? What does the camera come with, and what direction do we move to get a wider view?

Phil.

Originally Posted by Greg McKaskle

http://www.edmundoptics.com/onlineca...productID=2196

The have several quality levels for lenses. This is the primary link I use.

Greg McKaskle

[Greg McKaskle]

A wide angle lens would be nice, for ~ $30.

Man, the Axis 206 has the narrowest FoV I've ever seen, save a pinhole in a wall.

-Nick

The URL I posted contains wide angle and narrow/tele ones too. The second tab gives a table of specifications for the various lenses. Note that the 4mm lens has a viewing angle of ~70 deg. I never tried to make an accurate measure of the Axis, and instead relied on tables like this along with field of view calculators such as http://www.123securityproducts.com/calculator.html.

Anyway, note that the wider the viewing angle the higher the Distortion. Even with the Axis 4mm lens, the vertical lines are curving out or becoming barrels. The wider you go, the stronger this effect. Processing an image looking at size and shapes gets trickier unless you correct for the distortion -- taking more processing and introducing new artifacts due to resampling.

This isn't the same thing as this poll, but I'd be curious if/when teams buy alternative lenses to see comparative photos compared to the stock lens. This will in the long run save lots of team money and lead to better naive choices. It'd be a great little white paper, but may be better done as a collaborative wiki or collective forum post.

A strong cautionary note.
The axis sensor doesn't contain an IR blocking filter. Instead, the filter is expected to be built into the lens. This is cool in case you want to change the lens and do some rudimentary IR imaging. Not so cool if you neglect to use a lens with an IR filter somewhere around 650nm or so, and then notice that your contrast and color saturation are lousy -- I speak from experience. And in the end it is pretty much a waste of money. I have heard of teams that were able to transfer the IR filter from one lens to another, but I question whether this is a good use of time over ordering what you need in the first place. Note that on the page I posted, only two lenses contain the filter. So in fact, while this table is useful for comparison, I no longer recommend purchasing the lenses here unless someone comes up with an easy process for the IR cutoff.

Greg McKaskle

[PhilBot]

Originally Posted by Greg McKaskle

Anyway, note that the wider the viewing angle the higher the Distortion.

Greg McKaskle

I wonder if anyone has considered making a spherical imaging sensor.....

Hmmmm. Off to patent search.

Phil.

[Greg McKaskle]

Funny you mention that. I think most RoboCup teams that are allowed use the camera pointed up towards a mirrored surface, hyperbolic is really common I think, so as to have coverage of the entire field. I spoke with a professor who built a cheapy version using a mirrored light bulb and a different transform, and they were able to use robot motion as a stereo-optical substitute with pretty good results.

RoboCup has very standard lines on the field and markings on the robots to make it a bit easier for vision, but it would be a cool system even if you only had a few items to reference for orientation.

Greg McKaskle

[dlavery]

Originally Posted by PhilBot

I wonder if anyone has considered making a spherical imaging sensor.....

Hmmmm. Off to patent search.

Spherical reflective camera systems (both monocular and stereo) are fairly common. Several groups (including NASA, Carnegie-Mellon, Stanford, and MIT) did a lot of early development of the technology more than 15 years ago. Here is one example of a commercialized system from back in 1997.

-dave



.

[MrForbes]

Originally Posted by PhilBot

Hi

I'm embarrased to say that even though I passed optics in physics I don't know how focal length relates to FOV.

Have you ever used an SLR camera with a zoom lens? the focal lengths are marked on the lens barrel and align with a pointer on the zoom adjustment, you can see the FOV change as you zoom in and out. A point and shoot camera might have the focal length range printed on the lens housing, and you can zoom from one to the other and see what happens.

(shorter focal length = wider FOV)

[Greg McKaskle]

What does the camera come with, and what direction do we move to get a wider view?

The camera has a 1/4 inch sensor and the default lens is a 4mm focal length. If you go to the URL of a calculator such as
http://www.123securityproducts.com/calculator.html
and type in the distance to the object along with your lens description, and it will tell you the extent of the image at that distance.

For example, with the default lens and sensor, it calculates that you have a viewing angle of 73 degrees and at 30 feet from the camera, the image goes from floor to twenty feet above, and the distance from the leftmost to rightmost object in the image is 27 feet. Meaning that from midfield, the camera image contains the entire driver area.
Type in a new lens focal length, and it updates the description of what is in the image. If you want to jog those neurons and look at the math behind the calculator, you may want to lookup angle of view rather than field of view on wikipedia. They are closely related, and the angle of view seems to be the better documented term.

Greg McKaskle

[Ed Law]

I know this is an old thread. The URL posted http://www.edmundoptics.com/onlineca...productID=2196 now links to a product page that has lots of lens with different FL and with IR filter. However these lens are threaded for M12 X 0.5. How do I attach them since the Axis M1011 has no threads? Is there another source I can get a lens I can replace the original lens with? How do I replace it?

Originally Posted by Greg McKaskle

The URL I posted contains wide angle and narrow/tele ones too. The second tab gives a table of specifications for the various lenses. Note that the 4mm lens has a viewing angle of ~70 deg. I never tried to make an accurate measure of the Axis, and instead relied on tables like this along with field of view calculators such as http://www.123securityproducts.com/calculator.html.

Anyway, note that the wider the viewing angle the higher the Distortion. Even with the Axis 4mm lens, the vertical lines are curving out or becoming barrels. The wider you go, the stronger this effect. Processing an image looking at size and shapes gets trickier unless you correct for the distortion -- taking more processing and introducing new artifacts due to resampling.

This isn't the same thing as this poll, but I'd be curious if/when teams buy alternative lenses to see comparative photos compared to the stock lens. This will in the long run save lots of team money and lead to better naive choices. It'd be a great little white paper, but may be better done as a collaborative wiki or collective forum post.

A strong cautionary note.
The axis sensor doesn't contain an IR blocking filter. Instead, the filter is expected to be built into the lens. This is cool in case you want to change the lens and do some rudimentary IR imaging. Not so cool if you neglect to use a lens with an IR filter somewhere around 650nm or so, and then notice that your contrast and color saturation are lousy -- I speak from experience. And in the end it is pretty much a waste of money. I have heard of teams that were able to transfer the IR filter from one lens to another, but I question whether this is a good use of time over ordering what you need in the first place. Note that on the page I posted, only two lenses contain the filter. So in fact, while this table is useful for comparison, I no longer recommend purchasing the lenses here unless someone comes up with an easy process for the IR cutoff.

Greg McKaskle