Image Processing
- To: mathgroup at smc.vnet.net
- Subject: [mg123392] Image Processing
- From: pjf <paul-fons at aist.go.jp>
- Date: Mon, 5 Dec 2011 05:16:01 -0500 (EST)
- Delivered-to: l-mathgroup@mail-archive0.wolfram.com
We have a collection of about 500 GB of 16 bit images that represent diffraction data taken by a CCD. In a field of approxiamtely 10^6 pixes, we need to find patterns within a relatively large radially symmetric background function (stemming from an amophous background). A "data point" typically consists of a group of 3-5 pixels that are about 20% above background. We want to be able to compare the intensities of data points from different images so we need to be careful about techniques that will loose this intensity information. If we perform a maximum intensity projection on the thousands of images we have, we can see (as expected) that the data points (spots) like on well defined narrow radially symetric rings (Debye-Scherrer rings). I was hoping to solicit suggestions for how best to track down the needle in the hay type signal we are looking for. Wavelet processing came to mind. Are there any suggestions out there? Thanks.