The recognition results depend a great deal on the image quality of your documents. If you're scanning very low-quality documents, Readiris will not be able to recognize them properly.
The reason for this is that Readiris works by converting pictures of text characters into actual text characters. If your documents are too dark, the characters fade into one another and create black spots. If your documents are too light, the text characters might be broken. Consequently, Readiris will not be able to recognize them.
Example 1: text is too dark
Here, the letter shapes begin to fill in and become fuzzy. Letters like a, e, and o are almost completely blurred.
Example 2: text is too bright
Here, the text characters start to show gaps and incomplete shapes.
To improve the recognition results:
Check the Scanner Settings and make sure your documents have been scanned in a resolution of 300 dpi and in color.
If that's not the case, re-scan your documents using these settings.
Then access the Image Adjustment options:
Click the Adjust icon on the Image and Layout Editing toolbar.
This option is selected by default. It smoothens out differences in intensity, creating a greater contrast between the text and the background. Sometimes smoothening is the only way to separate text from a colored background.
This option removes unwanted dither (noise) from black-and-white images.
Move the slider to increase or decrease the brightness of your documents.
The results on the binarized image are shown immediately. If the results are not satisfactory, use the Undo button to restore the settings.
Example 1: image is too dark
The image below is too dark and results in a plain black image. There is no text to be recognized.
In this case you need to brighten the image in order for the text to be displayed.
Example 2: image is too bright
The image below is too light and results in broken text characters. The text is hardy legible.
In this case you need to darken the image to obtain satisfactory results.
Move the slider to increase or decrease the contrast between the text and the background.
The results on the binarized image are shown immediately. If the results are not satisfactory, use the Undo button to restore the settings.
Example
The image below results in broken text characters.
In this case you need to increase the contrast to obtain satisfactory results.
Sometimes documents contain a lot of "noise" - little black spots that occur when scanning low-quality documents or using bad scanner settings.
To remove these little black spots, use the Despeckle slider. The more you move it to the right, the larger the spots that will be removed.
The results on the binarized image are shown immediately. If the results are not satisfactory, use the Undo button to restore the settings.
Readiris is now able to remove vertical and horizontal lines to improve the text recognition.
Note that the lines are not removed from the output documents, they are simply removed before doing the recognition.
To remove vertical lines:
Select Remove vertical lines.
Enter values for the 3 parameters:
Minimum length: minimum length (in pixels) of the lines to remove.
Maximum thickness: maximum thickness (in pixels) of the lines to remove.
Maximum gap: maximum line gap (in pixels) to close, in case of broken, interrupted lines.
To remove horizontal lines:
Select Remove horizontal lines.
The same parameters are available.
When you are done adjusting the image quality, click X.
Important note: the Image Adjustment options are applied the binarized image only. Once you have applied them, the original image is displayed again.