c# - Separation of background/foreground layers in a scanned document -
i need automatically remove mildly colored background of scanned document image ocr.
scantailor open source c++ gui-based app background separation among other things, cannot figure out how run last step removes background.
ideally, find code , either:
- port part c#
- modify c++ respond command line execution, performing step on given image
can me understand how can either?
or know other libraries can this? (any language/platform acceptable)
you referring thresholding, despeckling , noise removal techniques necessary in ocr applications.
the quality of results depends many different factors -
print quality of original scan quality image resolution background colours , patterns used. noise , other marks.
you may find ievolution.net library @ http://www.hi-components.com/nievolution.asp useful. has many image processing functions play with.
there many commercial engines available. there no 1 perfect function solve image processing problems. must adapt functions , parameter match images. http://www.recogniform.com/thresholding.htm
- best threshold converting grayscale black , white
- adaptive threshold binarization: post-processing removing ghost objects
- adaptive threshold binarization's bad effects
- fast threshold , bit packing algorithm ( possible improvements ? )
a google search show lots of results.
Comments
Post a Comment