Strange checker board pattern in camera image

[Ville_H.]

Hello! 

I’m using  UI-1545LE-M camera manufactured by IDS, LabVIEW 8.6 and using LabView Interface for uEye industrial cameras V 3.32. When I’m using IDS LabVIEW example programs like 8-bit single capture or 8-bit live stream (available: http://www.ids-imaging.com/drivers.php?cat=5), there is a problem whit the image quality. At first glance the image looks ok, but closer look reveals that every other pixel is brighter than it should be and the image has kind of checker board pattern. I checked image memory values and an Image that should be completely black had values like this: pixel location 0,0 value: 0pixel location 1,0 value: 10pixel location 1,1 value: 0pixel location 0,1 value: 10

Using brighter image gave similar results and it seems that every other pixel has about ten units too high brightness value. When I use UEYE demo software to take images, this problem doesn’t exist and images look completely normal. Anyone know what causes this or how can I fix it?

[muks]

The image you are getting might be a bayer coded one. You have to convert this to u8 to view it properly.

[Ville_H.]

Forgot to mention that the camera sensor is Micron MT9M001 (Monochrome CMOS, http://download.micron.com/pdf/datasheets/imaging/mt9m001_1300_mono.pdf)

I was able to create a new image using IMAQ Bayer Color Decode and IMAQ Create Bayer LUT. I converted the 8 bit Grayscale image which had the checker board pattern to a 32 bit RGB image.  The new image looks good and  there isn't detectable  checker board pattern, but negative side is that the new image is 32 bit color image. Is there someway to convert Image straight to 8 bit Grayscale image?

[muks]

I was in a similar state and let me just search for the vi. In the meanwhile you can use extract luminance plane and get a greyscale image. Sorry if you are already doing that.

[muks]

Bruce very nice explanation here will be of use to you

[Ville_H.]

Thanks for replys!

I extracted luminance plane and it's workin pretty good now. Although, converting a grayscale image to a color image and then converting it back to grayscale dosen't feel optimal solution. If you can find VI that converts an image directly, I would appreciate it very much.

I also tryed to use IDS provided Black level compensation VI and it also removed most of the disturbances and there was no detectable checher board pattern, but the image looks a bit more grainy than using bayer decoding method.

[muks]

useful link

[BruceAmmons]

If you use a convolution filter on the original U8 bayer encoded image with the following elements:

1 2 1

2 4 2

1 2 1

It will directly convert it to a luminance image.  It isn't perfect, but it is darn close.  I recently discussed this in detail in another posting.

Bruce

[Ville_H.]

I tryed convolution filter method and it gave similar results than first decoding the image to RGB and then extracting luminance plane from it. So it's working pretty good now.