04-01-2014 08:05 AM
Hello,
I am doing first time image processing on labview and need help. I wanted to stitch the 100 images without overlap in labview for my application without vision software with simple labview with array and cluster manipulation for the project. For example I have image with "Labv" and in second image I have "view" and while stitching both images I wanted to discard of image part which appers twice in the image while joining and image should appear like "Labview" without two times v in the image for example "Labvview". I hope the question is clear other wise please let me know.
Note: Stitching of Image withput overlap(without labeled vision and other toolkit just using simple labeled)
I am looking forward for your reply.
Thanks in advance....
04-02-2014 02:15 AM
Hey
Where is the tutorial link in your post I am not able to see the link can you attach one more time and if you have better solution let me know.
Thanks.
04-02-2014 03:28 AM
Hello Ajaxy,
I do not fully understand what you mean with your last note.
What do you mean with labeled vision?
This could be a starting point:
https://decibel.ni.com/content/docs/DOC-4595
You could use the same principle as in the explanation.
If you don't want to use the Vision Development Module, then you could try to
- convert your image to its "raw data/pixel values".
- Stitch those together for all 100 images.
- Afterwards convert the "raw data/pixel values" again to the original image format.
04-02-2014 04:36 AM - edited 04-02-2014 04:39 AM
Hello Good morning,
Sorry Its atomatically corrected by labview dictonary its not labled its LABVIEW.
I dont have the image processing tool kit for processing data so i am looking for simple array manupulation and then converstion of image.
This option is not work for me. If you have any other solution please provide here.
Thanks....
04-02-2014 02:43 PM - edited 04-02-2014 02:43 PM
Hello,
As noted in my previous post:
If you don't want to use the Vision Development Module, then you could try to
- convert your image to its "raw data/pixel values".
- Stitch those together for all 100 images.
- Afterwards convert the "raw data/pixel values" again to the original image format.
Please let me know which part is unclear.
PS: What is actually the reason why you do not want to use the Vision Development Module?
Is this the licensing cost?
04-02-2014 09:36 PM
Discard overlaping part is hard. because it's hard to identify the overlaping part. With or without the vision module.
It's easy to do with the Microsoft ICE. You specify the images, ICE do the rest.
04-03-2014 05:10 AM
Hello,
I must have overlooked that part in the original post concerning the overlapping part.
Somehow I didn't see that. My apologies for that.
Next to the tool mentioned by zou, there are many others available online:
http://en.wikipedia.org/wiki/Comparison_of_photo_stitching_software
Making that from scratch would indeed require quite some knowledge (and algorythms following out of that knowledge) for the identification of the overlapping parts.
If it is only 1 Dimensional, then it would be a bit easier, but it could still be quite challenging.
How much Vision-related background do you have?
Have you done anything like this before in other programming languages?
How much experience do you have with LabVIEW? (followed courses, self-training, other LabVIEW projects, ..)
As a first starting point it might be interesting to have a look at the following pages:
- Manually defining an overlapping zone:
http://forums.ni.com/t5/Machine-Vision/Image-stitching-create-panorama-image-calculate-homography/m-...