Real-Time Image Processing with Labview

[amila]

Hi All,

I started to use Labview couple of months ago. I need to do real-time image processing. My application is to measure displacement of an object. I use a Logitech CamPro 9000 webcam to capture the movement. As it capture the movement, real-time image processing should be done so that the system know how much displacement has been done.

I found a VI for capturing video using webcam as I attached. How would I process the images? Can I do the processing directly or I have to convert the streaming video to readable format like JPG or something?

I'm using LabView 8.5.1 and have NI Vision Assitant 8.5.

Thanks for your time.

Regards,

Amila

[crelf]

The first thing you need to do is define what you mean by "real time". I'm not sure your definition and the industry standard definition are the same thing.

[amila]

QUOTE (crelf @ Jan 21 2009, 09:19 PM)

The first thing you need to do is define what you mean by "real time". I'm not sure your definition and the industry standard definition are the same thing.

Sorry.. I might not precisely understanding the term. What I need to do is acquire image of a moving object, do the image processing and measure the movement of this object. This system will like a feedback system. I need to know the current location of the object and after this object reach the desired position, I should stop acquiring images.

I attached an updated VI here. I manage to acquire image, covert the image from RGB tu HSL and apply edge detection. Next step is to detect the edge position. How should I do this?

Many thanks.

Amila

[crelf]

QUOTE (amila @ Jan 22 2009, 11:50 AM)

Next step is to detect the edge position. How should I do this?

Check out the VIs in the "Vision and Motion" > "Machine Vision" > "Locate Edges" functions palette.

[Neville D]

I wouldn't use JPEG images for edge detection. The discrete cosine transform used in JPEG compression is brutal on straight lines.

Use png or tiff if possible.

Neville.

[amila]

QUOTE (Neville D @ Jan 22 2009, 06:21 PM)

I wouldn't use JPEG images for edge detection. The discrete cosine transform used in JPEG compression is brutal on straight lines.

Use png or tiff if possible.

Neville.

Neville,

I'm really new to this LabView and my question may sound silly. How do I know that images being captures are in JPEG format? Where can I check it? And how can I convert it to PNG format?

Many thanks.

Amila

[Neville D]

What I meant was don't save the images from the camera in jpeg format and then try to detect edges after reading in the jpeg. Use the best resolution you can get from the camera, acquire the image, perform your processing and then if you want save the images (for off-line processing) in png format.

N.

[amila]

QUOTE (Neville D @ Jan 22 2009, 10:53 PM)

What I meant was don't save the images from the camera in jpeg format and then try to detect edges after reading in the jpeg. Use the best resolution you can get from the camera, acquire the image, perform your processing and then if you want save the images (for off-line processing) in png format.

N.

Ok.. I understand now.. Thank you.

QUOTE (crelf @ Jan 22 2009, 06:04 PM)

Check out the VIs in the "Vision and Motion" > "Machine Vision" > "Locate Edges" functions palette.

I'll try this today and let you know later. Anyway, many thanks.

Have a nice weekend..