Using Machine Vision to determine Saw blade damage

[StevenZi]

Hello,

I am using Machine Vision in LabVIEW to determine the amount of carbide that is on a sawzall blade tooth. To do this I am using a template image of a tooth with all of the carbide intact and using pattern matching VI's to determine how much carbide is intact on the inspected images of other blade teeth. I am using the score of the match to determine how much carbide is intact on each tooth. After going through some runs, I am unsure if the pattern matching VI's are capable of doing what I want. Would geometric matching VI's work better for finding the amount of Carbide on the teeth?

I have included a photo in the zip file showing what part of the tooth is carbide.

Thanks in advance 

[Worle]

I would suspect that using pattern matching is not going to give you a good measure of remaining carbide.

Looking at your pictures I have a couple of points, I am by no means an expert this sort of image analysis but this might get the conversation started for you.

- Image analysis is made so much easier with good quality images, in quite a few of the images in your sample the tooth is not entirely in focus. If you can adjust your cameras aperture settings to get the full tooth in focus whilst still having adequate lighting it will make your life easier.

- I would potentailly be more interested in different angles, Ideally on looking directly down on the blade and one looking from the side. If you had these two views you could analyse the profile of the blade and estimate loss of carbide that way (This is assuming clean breaks that go all the way though the tooth which you do have in a couple of your images)

- If you have two cameras well synchronised and accurately positioned you could possibly produce 3D data from the steroscopic images, this might be more involved than you are willing to go because analysing the 3D data would be a sizeable project on its own

- If you are stuck with your images I would be trying to detect the edges of the tooth, by measuring these edges you might be able to calculate material volume. This is possibly going to require better timing than you currently have to ensure that the tooth is in the same position relative to the camera each time. It will also be very sensitive to the type of break so it won't be easy.

[StevenZi]

Thanks for the advice.

[StevenZi]

Hello,

I am using IMAQ Pattern Matching VI's to determine the amount of Carbide, a material used for cutting, that is left on the teeth of a saw blade. I am using a template image of a blade tooth with all Carbide intact, and searching images of other blade teeth to see how well the images match. I am attempting to use the score of the match to determine the percent of Carbide that is left on the blade tooth. After going through some runs, I am unsure if the Pattern Matching VI's are capable of doing what I want. Is there something obvious I should change in my code? Would geometric matching VI's or other Vision VI's work better?

I have included a photo in the zip file showing the part of the blade tooth that is carbide.

Let me know if you have any questions.

Thanks for help

[David22JRA]

Hello StevenZi

Thank you very much for sharing the code. Have you tried to use the pattern matching examples? The code works pretty well and presents the score between the original image and the template you use.

Find Example (Browse)>>Toolkits and Modules>>Vision>>Pattern Matching

I hope this can help

[StevenZi]

Which example are you recommending?

[AllisonSCJ]

Duplicate from https://forums.ni.com/t5/LabVIEW/Pattern-Matching-VI-s-or-Geometric-Matching-VI-s-or-Other/m-p/40679...

Please continue your initial thread, rather than opening a new one.