The extracted text is incorrect with OCR for .NET

Aspose.OCR Product Family
1

I tried the demo “Extract text from images” of OCR for .NET with a simple image, but the extracted result is incorrect.

Following is the content of the image(and the format is BMP):

"Please browse a valid BMP or TIFF image file and press a button to check the output. Note: Arial and Times New Roman (16pt and 32 pt) fonts are supported."


But the extracted text is :

“P+^-^+^-^- ^ir*=-^^+^- ^ ^++^ era= **i=+rr +^v^J^- stew an^ vi^-+^-+^- ^ rumen x^ <^^-<+^^- **=rt=r- water Aha+ an^ ^^v^-+^- ^)^-~ ^v^^
!r^+^- ^i^- +^-=vv=–^-”
2

Hi Even,

Can you please attach your sample file for analysis? Sorry for the inconvenience.

Best Regards,

3

I haven’t the permission to upload the sample file, you can just save any text to image in BMP or TIFF format, and use the demo to extract the text, then you will get the incorrect text as I described before.

4

I want to add attachment, but it the dialog tell, I don't have the right to udpate, like below:

You do not have permission to upload or link to files. Please contact your system administrator.

5

Hi Even,

I have tested the above mentioned text with 28-36 pt sizes and extracted text has over 90 percent accuracy. Unfortunately current release has some issues with smaller font sizes and you can get the best output when font size is 28 pt or greater or Arial 16.

Our developers are working on this issue and recognition of smaller font sizes will improve gradually with the future releases. Sorry for the inconvenience.

P.S. You need to pack your sample in a ZIP archive to attach with the forum thread.

Please feel free to contact us in case you have further comments or questions.

Best Regards,

6

Hi There,

Any updates on handling smaller font sizes?

Cheers
Zardax
7

Hi Zardax,

We are making some major changes in Aspose.OCR product to improve performance and recognition quality and because of that, 1.2.0 release has been delayed. This issue will also be addressed once we release next version of Aspose.OCR. Sorry for the inconvenience.

Best Regards,

8

The issues you have found earlier (filed as ) have been fixed in this Aspose.Words for JasperReports 18.3 update.