Image processing without Vision and IMAQ ?

[drakones]

Hello,

I am new around here. I've been working with Labview for a while.

I am working on a project that involves image processing (actually some stereo image matching algorithms, that is). Since the algorithms that I've tried to embedd are way too slow for big images, I want to implement a ROI tool that allows me to select a rectangular area in one image, and apply the algorithm only on that area of the images.

The problem is that I don't want to use IMAQ and Vision. I don't have access to them, so I just want to use the default functions of Labview. Any help here, would be greatly appreciated. I am sumbled upon this problem, cuz I don't know how should I do this ...

Another question would be on the slow speed image loading in labview. For an image of 2000x2000 pixels, per se, it takes some time to load it. Is it normal?

Thank you,

Laurentiu A.

Romania.

Edited July 5, 2010 by drakones

[Antoine Chalons]

The problem is that I don't want to use IMAQ and Vision. I don't have access to them, so I just want to use the default functions of Labview. Any help here, would be greatly appreciated. I am sumbled upon this problem, cuz I don't know how should I do this ...

Ufff, it's like saying : "I want to go from Paris to New York, but I don't want to take an air plane", it's technically possible but you're not really chosing the easiest way.

Basically you have to know the format of your image, read the binary data and convert it to an 2D array of pixel (that will be your image), and then you have to redevelop the equivalent of the IMAQ analysis functions. Good luck!

Another question would be on the slow speed image loading in labview. For an image of 2000x2000 pixels, per se, it takes some time to load it. Is it normal?

This is not "normal", can you show the code you use to load your image?

By the way, what is the format of your image and how do you acquire the image?

[drakones]

Hi Antoine,

Thank you for your time.

Attached you can see the code for loading the images. I load them from disk, so I don't acquire them from external hardware (cameras). It verifies if the extension of the file is either jpeg, png of bmp. I can send you the whole vi, but I think it looks pretty scary right now .

I know how it looks what I'm trying to do. I already have the 2-D arrays of the images and decoded the RGB palete into separate colour codes.

As for ROI, I was thinking of a sort of overlay or mask over the picture tool, and then use some mouse events or drawing a rectangle on a transparent layer, and then taking the coordinates of the drawn rectangle and use them as the coordinates of the "selected" are in the original picture. I know it may sound stupid maybe, but as I told you, I don't have access to the specialised modules . Only the cracked versions, but of course those won't help me further .

[Francois Normandin]

As for ROI, I was thinking of a sort of overlay or mask over the picture tool, and then use some mouse events or drawing a rectangle on a transparent layer, and then taking the coordinates of the drawn rectangle and use them as the coordinates of the "selected" are in the original picture.

Hi Drakones,

for simple things like ROI or color threshold, you can venture into doing it with coordinates and events directly on the 2D Picture. But I agree with Antoine that if you need to fit patterns or do particle analysis, it's not worth starting from scratch.

Here's a little something to start with. (LV 2009)

ROI on 2D Pict.vi

[drakones]

Thanks a lot, François!

I will look into it. The problem is that where I am at this moment, I only have access to LV 8.5. I have to wait some time to gain a remote access to my home computer with LV 2009, and then I will check the vi you posted. I will give you a feedback after I'll check it.

Thanks again!

[Antoine Chalons]

One side question :

Why can't you use IMAQ & Vision Development Module? It would really make your life easier... :-o

[drakones]

I just checked the vi, it's just exactly what I was needin'. Well, with no cropping, but I can manage modifying that.

Thanks a lot, guys!

One side question :

Why can't you use IMAQ & Vision Development Module? It would really make your life easier... :-o

Well, I have license only for the LabView 2009, and no other modules included. Ask my superiors why they are so cheap .

I don't like getting complicated either, but I have to manage only with what I have. I know what Imaq and Vision can do ... And anyway it's a good way to practice with LV

[Dragoon235]

This is a bit of a stretch, but if you have matlab, you could use a matlab script node. They have a very large standard image processing library. But you might as well use Matlab for the project. Similarly, you could also use c code nodes, or compiled dlls, but this is overkill for labview. You'd probably do better developing natively in something else, rather than writing dlls for labview.

As for LV practice, I hate to rain on your parade, but do you know how many man-hours go into a toolkit for Labview (or for any programming language, for that matter)? I assure you that you will spend more time repeating other's work than doing your own. Remember, your time is money, too. Assuming you are an engineer, two weeks of your time is definitely more costly than a license. If you're a grad student, I can sympathize (I wear both hats), and I've heard the "make it work" message.

so much for "how to paint a house without a paintbrush"...

- Jon