Image Bitrate conversion

[UW-Ming]

Hello guys,

Hopefully this will be a very easy resolution.

I'm trying to follow the example as follows here:  http://digital.ni.com/public.nsf/websearch/F9B71B8E8266D90686256EDE006A4DE8?OpenDocument

However my box looks as:  http://mywebspace.wisc.edu/ymyang/web/myview.jpg  (lower right corner) and I can not seem to edit the inputs to make it look like the example.  Can anyone give any insight into this?

Also for the record, I'm using labview 8.5 however the ask-for-help wizard only went up to LV 7.1

Thanks!

Ming

[UW-Ming]

Quick extra note:  LV's IMAQ-Create only seems to handle 8 or 16 bit inputs (for the monochrome image i want to handle), yet it claims support for 8, 10, and 16 bit grayscale images.  Even if I get the image converted to a 10 bit mono image, when i pass it through vi's will it still retain its proper size?  I need to do fast gaussian fittings and things with said image and do not wish to work with a bloated (16 bit) image size, nor do i want to lose resolution (8 bit).

thanks again,

Ming

[VGA_CD-ROM]

Hi UW-Ming,
It looks like you've already done most of the work, you're just missing a step or two. The Image Out indicator shown in your block diagram looks like it's not a display, based on the property node you show in your screenshot. If you go to your front panel and look at the indicator there, does it look like an image display indicator with area to show your output image?  If not, go ahead and delete it and replace it with an Image Display from your Vision controls palette on the front panel. Then when you go back and create a property node for the display in the block diagram, you should see the properties for 16-bit display mapping as shown in that KnowledgeBase article. Then you'll want to right-click the property node and choose "Change All to Write" so that they accept the inputs you will specify ("Given Range", 1023, 0)

As for image processing steps such as your Fast Gaussian methods, you will unfortunately need to keep the data as a 16-bit image for those steps. This is because of the way the image is stored in memory, and each pixel can only be stored in multiples of whole bytes. So if you're working with an 8-bit monochrome image, you can store each pixel in just one byte and be OK, but once you go up to 10 bits, you need to use two entire bytes to store each pixel, so you're forced to use a 16-bit image type for that, unfortunately.

[UW-Ming]

Thanks Vijay,

That reply helped a lot.  I was actually more worried about using the image for later processing than just display, so if I should keep it 16-bits I will do that.  As far your comment regarding the 16-bits for gaussian methods, are there any functions you happen to be familiar with that can perform such functions (of finding the center of a bright spot) without having to convert the image into matrix form?  The only code I've found are IMAQ mean, and IMAQ centroid, which arent as robust as a gaussian fit.

Thanks,

Ming

[VGA_CD-ROM]

The IMAQ Centroid function is the best one that I know of for performing that kind of function without converting to a matrix first (and with NI Vision you can also convert matrices back to arrays and back to images as necessary).

You can also take a look at the NI Vision Concepts Manual for more information on image processing theory, which can be found by going to Start » Programs » National Instruments » Vision » Documentation » NI Vision. On Windows XP, the concepts manual is usually found in the following directory: C:\Program Files\National Instruments\Vision\Documentation