High sensitivity camera module with no color or ir filter

[levi47]

Would it be possible to offer a camera module that omits the color and ir filter for a highly sensitive black and white sensor? I have an application in mind that requires as high a light sensitivity level as possible and will only be dealing with monochromatic light. I also thought this sort of module may be useful for people looking for highly sensitive security cameras.

[riklaunim]

Some machine vision camera vendors provide their software stack for ARM-Linux (Point Grey, IDS Imaging... more?) so you could use USB2 mono camera to get those features

[mahjongg]

Would it be possible to offer a camera module that omits the color and ir filter for a highly sensitive black and white sensor?

no, thats not technically possible, removing the ir filter was just uncommon, but possible, removing the RGB (bayer) filters from the CCD IS impossible, because they are an integral part of the CCD chip.

https://en.wikipedia.org/wiki/Bayer_filter

and

https://en.wikipedia.org/wiki/Image_sen ... separation

[riklaunim]

Would it be possible to offer a camera module that omits the color and ir filter for a highly sensitive black and white sensor?

no, thats not technically possible, removing the ir filter was just uncommon, but possible, removing the RGB (bayer) filters from the CCD IS impossible, because they are an integral part of the CCD chip.

You just take a mono sensor and not a color with Bayer filters. So yes, it is possible to offer a new product with a mono sensor but not possible to do that with existing color one. And it's CMOS in this case, not CCD.

[jbeale]

I know that Canon has made a few true-monochrome versions of their prosumer DSLR cameras intended for astronomy (no bayer filter):
Canon EOS 20Da (in 2005) and Canon EOS 60Da (in 2012). Or, you can buy a Leica M monochrom for about $8000.

If you really want to live on the edge, some people have done a modification I would not have thought possible, check this out (!)
http://petapixel.com/2013/08/04/scratch ... bw-photos/

[levi47]

Does anyone know if omnivision or any other maker has a cheap cmos module of similar specs that does not have the bayer filter applied, for security cameras? I figured there would be a large market for these type of modules where high light sensitivity(3x+) was important and color imaging was not. If nothing is available as a raw sensor does anyone know of a site that lists simple usb camera's by min lux values?

[jbeale]

The Omnivision OV7221 is B/W and claims good sensitivity, but it is only 640x480 pixels, and I don't know if it's a current product- never seen one for sale. http://www.ovt.com/products/sensor.php?id=73
RS Electronics sold an eval board for it (RS 714-1486) but that product is listed as discontinued.
http://sa.rsdelivers.com/product/omnivi ... 41486.aspx

I believe the majority of the security camera market is now so-called "day/night" cameras which do have a bayer filter, and a switchable IR filter so you get color images during the day, and mostly-IR images at night. The assumption is that you will have some kind of IR illumination at night. The "night" mode images are forced to black-and-white in software, although they are almost pure black and white already, because all the colors in the bayer filter array are mostly transparent at IR, and for that reason there is little advantage in that application to a pure-monochrome imager without a bayer filter array.

I know that people doing astrophotography often want a pure-monochrome, no-bayer sensor, and you can buy cameras for that purpose. Usually not cheap.

Here's a board-level B/W OEM camera with USB, but I have no idea where you could buy one.
http://www.videologyinc.com/cameras/cmo ... C=overview

[levi47]

It looks like it will be very unlikely that they will release a high sensitivity module that will allow the kind of direct capture control i am looking for for this project, but i can dream.
Off the direct topic but related:
I wonder, if i know the approximate wavelength of the monochromatic illumination light if i could use some sort of binning of the surrounding color pixels and weight them by the sensitivity of each subpixel to that wavelength.

[jamesh]

It looks like it will be very unlikely that they will release a high sensitivity module that will allow the kind of direct capture control i am looking for for this project, but i can dream.
Off the direct topic but related:
I wonder, if i know the approximate wavelength of the monochromatic illumination light if i could use some sort of binning of the surrounding color pixels and weight them by the sensitivity of each subpixel to that wavelength.

It simply would not be cost effective to do all the work to get a sensor like that working, since the market is pretty small.

In the long run, if the CSI-2 HW spec gets released, people will be able to write their own drivers, but it's a big job since the control stack for the camera is not OSS, and that's >200k lines of code. The tuning might also take a while.

[6by9]

In the long run, if the CSI-2 HW spec gets released, people will be able to write their own drivers, but it's a big job since the control stack for the camera is not
OSS, and that's >200k lines of code. The tuning might also take a while.

Mono sensors with inbuilt AGC shouldn't need tuning. Let's see what Gordon docs manages to get permission to release from of Broadcom.

[jamesh]

In the long run, if the CSI-2 HW spec gets released, people will be able to write their own drivers, but it's a big job since the control stack for the camera is not
OSS, and that's >200k lines of code. The tuning might also take a while.

Mono sensors with inbuilt AGC shouldn't need tuning. Let's see what Gordon docs manages to get permission to release from of Broadcom.

I did wonder if it would need tuning. I presume some work would be needed but not to the extent of a full colour sensor.

[bjswift]

Has anyone messed with the color filter scrape-off method recently?

I tried it out with a spare camera module yesterday, but buggered up some of the bond wires around one corner and bricked it. So given license to just go to town on it, I was seemingly successful in using fine tweezers to scrape the filter/microlens layer off. Using a sharpened wooden stick (from a cotton swab) or a plastic spudger didn't work because the materials were too soft. I'll try some more when I get a chance with a slightly harder wood, or a softer metal(?) because while the tweezers were effective, it did leave some visible scratches that I'm sure don't need to be there.

I'll also have to read up and try to be more careful in opening the module case to get to the sensor. Any advice there?

[jbeale]

You're a braver man than I am, bjswift. But do keep us informed if you manage to make it work!

[alain1405]

Has anyone managed to get a monochromatic sensor at a reasonable price?

[Gavinmc42]

Might want to try removing the bayer filter from a Sony PS3 eyecam.
It is only VGA but it has bigger, more sensitive pixels and secondhand is cheap.
IR filter is in the lens, so just swap to lens without filter.

There are companies that offer bayer pattern removal for $$$$.
The trick is probably finding a solvent that dissolves the RGB filter but not the silicon.
Maybe Acetone?

Or get one of these
http://www.x20.org/color-night-vision/

Sony do have better image sensors for night time.
http://www.sony.net/Products/SC-HP/sens ... arvis.html
You would have to use the raw mode as these are not supported by the gpu.(yet?)
Look for 2 lane MIPI type to use the same connector.

Another option is the parallel type of data output cameras, typically in the <3Mp range.
I have used DMA to get data directly from a image sensor without using a dedicated sensor interface on other micros.
Parallel mode on GPIO pins, usually around 26MHz?