Edge detectionon jpgs

Hi I am currently using lead tools edge detection to pick apart bit maps that have been converted from a RAW format.  This is a picture coming off of a digital camera.  It works well and we are having no problems with it.  However, the camera that we are using is being discontinued and we are being forced to retreive jpg's from the camera now. 

After converting the jpg to a bmp, I am again using the same algoriths to map out the edges of the objects.  However, now it no longer works 100%  More like 25% and pretty much failing something in every attempt.  I was wondering what would be our best bet to get this back up and running?  I have tried sharpening the image before I run the edge detection but that doesn't seem to work very well either.  Is this just a problem with jpgs encoding?  Or is there a way lead tools can get around it?

Thank you,

Mark Kirchdoerfer

Mark,
One possible reason is JPEG compression as I will explain here.
Usually, raw formats are lossless, meaning the pixels stored are identical to what the camera's sensor captures. On the other hand, common JPEG compression is lossy, and its algorithm throws away some color components to reduce file size. (There's a lossless variation of JPEG and LEADTOOLS supports it, but it's not very common and I don't think cameras support it because it doesn't reduce size as much).

If that's the cause of the problem, the solution or workaround will depend on different factors such as the images contents, current algorithm you're using, and the exact differences between the raw and JPEG images. If you can't find a solution, we can try to help you if you send us the following:
- A typical sample image of each type (raw and JPEG).
- A small test project that performs the correct processing on your sample raw image, but gives poor results on the JPEG image.

If you want to send files, you can either post them here or email them to support@LEADTOOLS.com

Important notes:
1. If you post or email files, please put them in a ZIP file first.
2. Please don't send your full application.

yeah I understand the loss from the jpeg, and I guess one of my major issues is when I run it through the EdgeDetector 12, EDG_SOBEL_ALL, function it comes back with the edges in white but a bunch of multicolored pixels washing the  image, I believe these artifacts are what is causing my issues,  Since they are showing up in teh middle of a white field it's an issue.  I'll send you some files with the images in them.  If I could find a way to reduce the pixelization I think I'd be pretty much set.

Mark,
When images are compressed using JPEG, the color components are often pixlized more than the intensity component. This means if it's possible to perform your processing without the color components, you might get some improvement. If you want to try that, convert the image to 8-bit grayscale.