Photomerge algorithm

Hello,



I am trying to stitch together 16-bit grayscale images (1000x1000) that are taken under 40X microscope. I can photomerge 10X images without a problem, but I can't photomerge 40X images. Does anyone know what I need to optimize during image acquisition to enable the photomerge feature? I am wondering how the photomerge algorithm is written - based on contrast? Similar pixels? Please help! Thanks!
Photomerge algorithm
Which version of Photoshop are you using? CS2 vs. CS3 is a huge difference.



Photomerge (CS3) works by identifying ''features'' in each of the images, and then matching up. The algorithm is called SIFT (Scale Invarient Feature Transform). For details, see:



http://en.wikipedia.org/wiki/Scale-invariant_feature_transform



It's hard to tell without seeing the images in question, but it could be the higher magnification images don't produce the kind of locally distinct features the algorithm needs to get traction on. Increasing the overlap and/or the contrast (e.g., via histogram equalization or similar techniques) may help.
Photomerge algorithm
Thank you for your reply.



I am using CS3. The funny thing is that I tried Photomerge after processing images with ''Auto Contrast'' as well as about 95% overlap, and the images still could not merge. I will take a look at SIFT. Thanks!

Hello,

I am new to this forum... I don't know how to create new topic...



Anyway, We have currently have a license for Adobe Photoshop CS2 and the main feature we are using there is the PHOTOMERGE Command.



We like to manually arrange the pictures during photomerge (as in CS2)... But in photoshop CS3, I can not find the way to Disable the ''Attempt to align selceted layers based on content''.



Does anyone know how to do this in Photoshop CS3?



Thanks.

From R.Vinoya

%26gt;I am new to this forum...



I think you may also have the wrong forum. Are you using the Photoshop

SDK (this is a tool for programmers to write plug-ins and automation

for Photoshop)?



Aandi Inston