From Friday, April 19th (11:00 PM CDT) through Saturday, April 20th (2:00 PM CDT), 2024, ni.com will undergo system upgrades that may result in temporary service interruption.

We appreciate your patience as we improve our online experience.

Machine Vision

cancel
Showing results for 
Search instead for 
Did you mean: 

Checker Board Pixel Check

Hi,

 

I am trying to verify the pixels on a small display (160x102) are all working correctly. One of the test images uses a checkerboard pattern to test for shorts between pixels. The goal is to add the captured image to an inverse checkerboard pattern and check for any anomalies. 

 

The .vi I have written so far applies a threshold, finds the edges of the screen, rotates, and extracts the ROI. 

 

The problem I am having is when i try to resample the captured image (roughly 928x595) to the resolution of the screen/checkerboard (160x102), it is littered with false positives regardless of the threshold setting or interpolation type.

 

This may be the wrong approach alltogether however.

 

Attached is a cropped sample of the original image, the image with a threshold applied, and the image after resampling.

 

Any advice or help would be greatly appreciated. 

Download All
0 Kudos
Message 1 of 4
(4,311 Views)

Another approach i am experimenting with is: "The checkerboard image is binarized using a morphology-based adaptive thresholding algorithm, and the pixels are highlighted using blob analysis and filtering routines"

such as used in an ETCs DisplayCheck, however it may take a while to get a function such as this functioning. 

0 Kudos
Message 2 of 4
(4,301 Views)

I think you are making your own life too hard:

- If you want to subtract, don't downsample your source image. You are losing information. Upsample your subtraction image.

- Do you really need this? I created a second, test image with 4 errors put into it (attached). When using "Golden Template Matching", I immediately found the errors in there. It depends a bit on your test image quality and how well you can align the images..
- You can always go the texture defect / Wavelot route for this..

 

goldentemplate.PNG

0 Kudos
Message 3 of 4
(4,269 Views)

@JustinCase24 wrote:

Another approach i am experimenting with is: "The checkerboard image is binarized using a morphology-based adaptive thresholding algorithm, and the pixels are highlighted using blob analysis and filtering routines"

such as used in an ETCs DisplayCheck, however it may take a while to get a function such as this functioning. 


Can you put some light on what a "ETCs DisplayCheck" is?  I have never heard about that...

 

Edit:  I found it by reading this article...

0 Kudos
Message 4 of 4
(4,227 Views)