Drop-out is a process whereby the pre-printed content (template data) on an image is removed, leaving only the data that was added to a form (filled data). When filled and template data overlap, the filled data will be reconstructed as accurately as possible. Because drop-out is an important part of forms processing, it's important to follow certain guidelines as described below.

Improving the Quality of Drop-Out

Drop-out problems are usually caused by poor registration, but some may also be due to distortion of the form.

If drop-out leaves pieces of the form across the entire page, the problem is probably poor registration. Please refer to the sections on identifying images for more information. If drop-out leaves pieces of the form in only one or a few areas, the problem is probably distortion of the images. In rare cases, a large difference between the thickness of lines in the filled image and the template image also contributes to dropout issues across the entire page.

When resolving distortion problems, the first thing to consider is the possibility that two separate revisions of the same form are being processed. Sometimes this can be caused by the creator of the form making a change without recording it, or the form's revision number having been changed. This can also be caused by differences in printers. The same form printed by separate printers can contain font changes that cause drop-out to fail. If this is the case, see the section on handling similar forms during identification for additional suggestions.

Possible Methods to Improve Drop-Out Performance

• Turn off PerformReconstruction if you do not need it.
• Try reducing the value of AllowableMisRegistration, but setting that value too small may cause some forms to drop-out poorly.
• With overlapping fields, consider using one larger field instead of multiple overlapping ones. The number of pixels contained in each field is the dominant factor in performance and each pixel that is part of two separate fields takes twice as long to extract.
• In the case of several very small fields that are near each other, assess the use of one large field that covers all the desired pixels. There is a small amount of fixed time used for each field and for very small fields that time can be higher than the time spent processing each pixel.
• If the field is large, but not all the pixels within it are used, study the possibility of using several smaller fields. this will eliminate time spent processing pixels that you will not use.
• Attempt to improve consistency between the thickness/brightness of lines in the filled and template images as early as possible. In other words, be consistent in the printing and scanning process.
  Because once the images are scanned, a template cannot be adjusted to match an inconsistent thickness/brightness in the filled images. And, in most situations, adjusting the thickness/brightness of a filled image after scanning can adversely affect the filled content which is used for later data acquisition.

Alternatives

Form drop-out is very similar to the line removal, comb removal, and virtual bulb operations within ScanFix. In some cases, you will get better results by using those technologies.

Comb Removal in ScanFix - before and after images:

 

Before Comb Removal

 

After Comb Removal