Machine Vision

cancel
Showing results for 
Search instead for 
Did you mean: 

Tool Advice

Hey guys,

 

I need to inspect and image(grayscale) of a battery and need an advice on the type of tool which is more efficient and fast..

My requirement basically is to check for top cover which is the black inner ring on the image and the label which is the white outer ring.

The image is overlayed with CWIMAQRotatedRectangle object and on each rectangle I want to determine if its either black or white

Would it be efficient to use the histogram tool on every single rectangle and make a decision if its a pass or fail base on the CWIMAQHistogramReport object.

The application has to make a decision within 1.2 sec which I think is possible.

 

Any help greatly appreciated....

Thanks

0 Kudos
Message 1 of 5
(3,690 Views)

I would consider the annulus tool for this application, unless you need to analyze each section (rectangle) independently.  The annulus would give you the entire ring in one shot.  What do failures look like?  Would the annulus still detect failures?

 

There are a number of tools that give you average intensity in an ROI - brightness, quantify, histogram.  I would try each one and see which one is fastest.  I would guess quantify would be faster than histogram.

 

Bruce

Bruce Ammons
Ammons Engineering
0 Kudos
Message 2 of 5
(3,664 Views)

Hi Bruce,

 

Thanks for the advice... I tried to use the annulus tool but sometimes it can only detect the part of the top covers or labels coz labels would sometimes be oval when it shrinks.

 

Also I need to check the top cover and label coverage on the cell.

 

And yes I think I need to inspect every single rectangle.

 

Please see attached reject images...

 

Thanks again

Riscoh

Download All
0 Kudos
Message 3 of 5
(3,659 Views)

I might consider spokes for analysis.  If you were to extract the intensities along each spoke, you could find if the line was all dark, all bright, or find the transition from dark to bright.  This would allow you to determine if the label edge is within tolerance all the way around.

 

Bruce

Bruce Ammons
Ammons Engineering
0 Kudos
Message 4 of 5
(3,650 Views)

Thanks Bruce

 

I'll give it a try

0 Kudos
Message 5 of 5
(3,643 Views)