I'm looking for a framework/library that will preprocess scanned images and prepare them for further OCR process. For example, it will automatically detect and remove background noise, will improve text contrast etc.
Is there anything available on the market for this purpose?
The LEADTOOLS Document Imaging library contains various image processing and document clean-up functions. Some of them are specifically designed for OCR pre-processing. Examples are DotRemoveCommand, DeskewCommand and InvertedPageCommand. You can try by either downloading the full SDK trial edition, or the stand-alone OCR click-once demo. (Disclaimer: I am an employee of this toolkit’s vendor).
For example, the DotRemoveCommand can be used to automatically remove dots and specks of various sizes, the code will look like this:
RasterCodecs codecs = new RasterCodecs();
RasterImage image = codecs.Load("image.tif"));
DotRemoveCommand command = new DotRemoveCommand(DotRemoveCommandFlags.None, 1, 1, 10, 10);
command.Run(image);
