 |
document processing
Ripper121 from Germany [2 posts] |
9 year
|
I have a document and need to read out the corssed rectangles.
How can i do that?
|
|
|
|
Steven Gentner from United States [1446 posts] |
9 year
|
See attached robofile that will show the checked boxed in all black.
Now you just need to figure out how you want the information organized ... i.e. you can get the cog of all the marked boxes and determine what they mean based on the coordinate but this assumes that the boxes are always in exactly the same spot. To ensure this, you will have to align the image based on some known object like the outer black square. Or perhaps just shift the image 0,0 coordinate to be the first box detected.
Note, the image doesn't need to be as large as when you scanned it in for these results. Smaller is faster!
STeven.
 program.robo
|
|
|
|
Ripper121 from Germany [2 posts] |
9 year
|
Thanks for the fast help.
The card will stick in a box, so that the points are always on the same coordinates.
I need also the boxes with the numbers.
How do i get the coordinates from the detected boxes?
|
|
|
|
Steven Gentner from United States [1446 posts] |
9 year
|
Ripper,
When I mentioned coordinates, I was talking about pixel coordinates. Even if the card was is placed in the same box you'd very likely not have the same coordinates since even the smallest movement will change the pixel coordinates.
A better way is to instead order the boxes in a consistent manner. I've attached something that does this.
Note, from here on you really need to take many more images to test. Once you have this database of samples you can better test what is needed. There is a good deal more to do to get this to be reliable which you will notice when you test multiple images.
The attached has the basics which should help you complete the process for the other squares. You'll have to download the latest RR version for this to work.
STeven.
program.robo
|
|
This Thread
