Pattern Recognition for solar cells for different sizes

[lazer]

Hello every one,

Thank you for all help with time and time provided by forum team. Now again I am stuck with programing again. What I am trying here is to detect the solar cell even if the reference image cell is entirely different to the actual image. In programing, it can be work by checking the the shape but here in labview, I am struggling to implement my concept I need help. I have attached the sample picutre which I did with the help of Eden ( member of forum). This picture will explains what I am actually trying to do.

I tried to do this by comapring the images which gives mean image and this new image can be used to detect but this concept is not working. Can someone help me please. The code is attached and orgnl image is one to detect the cells. I will upload my reference image on my next post. 

My next stage aim is very similar, which is instead of different size it would be different intensity of images and rotated iamges. Can you explain how and which Vi should I look for to achieve what I am trying to do. I really want to give a go myslef before coming here. 

Many thanks

Lazer

[lazer]

Reference image is attached but reference image can be any image. 

[IanS]

Hi,

In what way is it not working? Is it not returning any matches?

Looking at the code I see you are using the matching VIs already, which is good. These will look for a shape and match it. It already searches for rotated matches and different intensities, so you don't have to change anything here.

It looks like you haven't got a setting for the minimum score, it may be that the matching scores are too low and are being missed. I would suggest trying a few of these.

What sort of experience do you have of using the IMAQ functions? There is a lot of information in the help files about what each function does, and is a good place to start looking for more details. You can find these under Program Files>>National Instruments>>Vision>>Documentation. (maybe program files (x86)).

There is a NIVision Concepts help, which gives a lot of details on how to do vision analysis in general, and the Labview help gives details about the vision VIs in labview.

Let me know if you have some more questions.

[lazer]

Hello Ian,

Apologies for late reply. What I am trying to do, let me explain in steps

1) Locate/detect the solar module out of image

2) dtect the crack and spots from the solar image. 

What I did? I look at this project systamatically, and break down into more components and at the moment I am trying on stage 1 and I have only 5 weeks left to do this project. I have detected the solar module sucessfully as you can see sample picuture however to make the software more reliable, I break down this into following stages

1) What if the image is not straight. In this case make a code which straight the image 

2) What if intensity of the image does not match to reference image. 

3) What if destination image is different to reference image ( different size of the cells, could be different in widht or lenght)

To achieve stage 1 I used matching VIs, which is working perfectly. Regards to your reply, In what way it is not working? I am trying to make a code for detecting the image even destination image is different to reference image but code still be able to detect the solar cells and highlight the all the cells or requested number of matches. Hopefully, my this post will clear all the concept what I want to achieve and my question was is there any way to complete this different size detection task, if yes, then how to do it and can you provide an example which can guide me through.  

Regards to Documents I am looking at this and found very useful. I will wait for you kind reply. Thanks 

Best wishes,

Lazer

[lazer]

Hello Ian, 

As you advised me about getting the score and I tried to make this VI. What I try to achieve here is, detect the shape and area of the cell. However this isn't working quite well can you have a look at this please. 

Many Thanks 

Lazer