Is it possible to make a custom Bayer filter in LabVIEW?

[JRaetz]

Hello all,

I am working to connect a JAI/Pulnix TMC-9730CL camera for image processing in LabVIEW and Vision.  Currently the only Bayer filters I can find in LabVIEW are the 4 built in filters that are 50% green.  This camera however uses a 75% green array.  The array is organized like this:

GBGG

GGGR

GBGG

GGGR

Is there anyway to specify the arrangement of my own filter, or can LabVIEW only use the 4 defaults?  I've looked into how the filters are set up, and they seem to be an array of numbers.  Can I change the numbers to be my filter, or behind the scenes do those correspond to only the pre-defined filters?  Any help that can be offered will be much appreciated.  I am fairly new to LabVIEW, but won't mind delving into any C code if that's what is necessary.

Thanks in advance,

Jordan

[JRaetz]

Anybody have any insight into this?  Is there any other way to even make a work around for this?  I've been trying to use 2 different Bayer filters to separate the images, extract 2 of the 3 colors, and combine them, but it will still never turn out to be the filter I need.  Once again, my camera uses this array:

GBGG
GGGR

Labview only provides the ability to filter in these four ways:

RGRG

GBGB

GRGR

BGBG

BGBG

GRGR

GBGB

RGRG

Thanks,

Jordan

[Chris_VH]

Hey Jordan,

I definitely do not know of a quick way off the top of my head to do this, but I would almost guarantee that it can be done. As you've stated: we normally deal with the 4 set Bayer filters, as thats what almost every camera uses. 

If anyone has dealt with this, their help would be greatly appreciated, but if not, I will look into it for you.

[BruceAmmons]

This is a crazy camera!  The quality of your blue and red components will be extremely low and blurry compared to the green components.  Why did they put so many green pixels on there?

The only way I can think of doing a Bayer filter for this is to write your own.  This has been discussed in other posts on the forum, so I will give the highlights here:

1.  Apply three color masks to obtain three separate images from the original image.  Each image will represent one color plane.

2.  Apply a filter that fills in the missing data for each color plane.  If you need color correction, include the factors in the convolution.

3.  Combine the three color planes to get the final full color image.

Bruce

[Andrey_Dmitriev]

---

Bruce Ammons wrote:

This is a crazy camera!  Why did they put so many green pixels on there?

---

Probably "G" means "Great" instead of "Green"? 

I'm sorry for offtopic,

Andrey.

[pschmal]

Hi all,

I have taken this project.  I wrote a linear interpolation algorithm to process the raw pixel intensity.  It works OK, but of course linear interpolation has its short comings.  Also, I think it is comp. intensive when running through matrix in LabVIEW.  Is there a way to do splines or faster interpolation natively?

This is not that uncommon of a color filter array... unfortunately it does not look like LabVIEW has native support for it.  The Bayer filters are handled thru DLL.