Background suppression

If we know something about the background – either absolutely or probabilistically- we can design ways to suppress it before we do the pattern recognition.

Here are some examples

For a grainy background:

http://canopus.physik.uni-potsdam.de/~axm/bgsmooth.html

Here is a more general approach

http://www.neuroinformatik.ruhr-uni-bochum.de/ini/VDM/research/computerVision/imageProcessing/setsOfFeatures/gabor/robustnessAndManipulation/backgroundSuppression/results.html

In medical applications, the background is sometimes recognizable enough to allow it to be subtracted.

http://apps.gemedicalsystems.com/geCommunity/mri/mr_i/softapps/SmartPrep2000.jsp?pkg=smartprep

In general, one can seek to segment the image from the background. Segmentation of images is a “hot” field of current research. Unfortunately, little of this work is on the web. And, of course, the more a priori information you have, the better you can segment. These links give you the feel of this work.

http://www.sccs.swarthmore.edu/users/01/xianglan/e27/objRecognition.html

http://www.nada.kth.se/cvap/abstracts/cvap218.html

http://www.owlnet.rice.edu/~elec539/Projects99/NSJS/edgedet.html

http://www.inrs-telecom.uquebec.ca/users/amer/research/msc_abstract.html