Lines often interfere with OCR, particularly when they cross characters. If your images are color and have lines of a different color than the data you are trying to OCR, the VirtualBulb method is typically the best approach. Virtual Bulb acts like a dropout scanner, dropping a given hue to white and all other hues to gray or black, generating a gray output from a color input image. Virtual Bulb is also capable of doing things that real dropout scanners cannot, such as drop multiple hues at once, compensate for dark ink without losing faint fill data, and even extracting a given color and converting everything else to white.

Example of Virtual Bulb:

If your images are bitonal, ScanFix® Xpress includes two different tools for removing lines with minimal damage to intersecting characters. The RemoveLines method removes horizontal and/or vertical lines of a given minimum length from bitonal images, with settings for dealing with broken lines and reconstructing characters and lines that cross the removed lines. The RemoveCombs method removes both combs and fill-in-the-box elements of forms, using spatial frequency analysis to find the "teeth" and remove them, while leaving the vertical strokes of the characters.

Example of Remove Combs:

The AutoRemoveHolePunch method automatically detects and removes hole punch marks from a bitonal image. No user parameters are required to be set.

The RemoveBlobs method can be manually configured to remove large, high density objects from a bitonal image. This includes allowing the removal of items such as punched holes and graphical logos that might interfere with other processing.  You must specify the area, size, and density of the blobs to remove.

Example of Remove Blobs: