Join us for an engaging webinar, as we unravel the potential of AI for revolutionizing document management.
Watch Now
Enable your employees to remain productive throughout the document management process.
Read More
Learn how SmartZone uses a regular expression engine integrated into the recognition engine to achieve the best possible accuracy on data that can be defined by a regular expression.
Docubee is an intelligent contact automation platform built to help your team success
What quality should my images be for processing form data and recognition using FormSuite?
In all cases, you want to have your images as clear and as clean as possible. For any particular procedure, please consider the following:
OCR and ICR: Capture images in at least 300 DPI resolution. Ideally, working in black and white allows the objects of interest on your image to be better defined and recognized. Free the image form all noise as much as possible. As if a human were reading it, you want the text objects on the image to be as legible as possible. For ICR, ensure that the characters are printed (no cursive text, etc).
Barcode recognition: As with OCR and ICR, capture images in at least 300 DPI and working with black and white content can provide excellent results. Ensure that the bars in the barcodes are clearly defined on the image and are not malformed (for example, the barcodes should have the proper start and stop sequence, etc). Clear as much noise from the image as possible.
Forms matching and registration: As with the prior 2 items above, capture your documents in at least 300 DPI. Ensure that your resolution is consistent between your form templates and incoming batch images. Form templates should only contain data that is common to every image that is being processed (i.e. Form fields, the text that appears on the blank form itself, etc). The template should not have filled-in field information as this will affect the forms matching process.
What are the best quality images to use when processing form data and recognition?
In all cases, you’ll want to have your images as clear and as clean as possible. For any particular procedure, please consider the following: OCR and ICR: Capture images in at least 300 DPI resolution. Ideally, working in black and white will allow the objects of interest on your image to be better defined and recognized. Free the image form all noise as much as possible. As if a human was reading it, you’ll want the text objects on the image to be as legible as possible. For ICR, make sure that the characters are printed (no cursive text, etc). Barcode recognition: As with OCR and ICR, capture images in at least 300 DPI and working with black and white content can provide excellent results. You’ll also want to make sure that the bars in the barcodes are clearly defined on the image and are not mal-formed (for example, the barcodes should have the proper start and stop sequence, etc). As always, clear as much noise from the image as possible. Forms matching and registration: As with the prior 2 items above, capture your documents in at least 300 DPI. Make sure that your resolution is consistent between your form templates and incoming batch images as well. Form templates should only contain data that is common to every image that is being processed (i.e. – Form fields, the text that appears on the blank form itself, etc). The template should not have filled-in field information as this will affect the forms matching process.
When should I apply image cleanup operations on my document images?
There are a number of cleanup operations that you can use to make an image more suitable for a particular application. What you observe visually on the image and how you perceive its impact on your project is the most important. For example, if you’re noticing very many random specks on your image, and you’re planning to use OCR, then you may want to try a depseckle or blob removal operation first. If the content in your image looks a bit slanted, you could try a deskew or rotate operation. In some cases, using a line removal operation on forms that have grid fields could be helpful also. The amount of image cleaning you may need to do can very from project to project. There’s not a one shot cleaning operation that will always work for all images. But, observe the nature of the noise and interference in your images to determine what general parameters appear to provide the best results.