Last Updated on
Component: OCR & Recostar Deskew Property
Affected Version: Ephesoft Transact 4500
There may be a scenario where the images are skewed and when uploading the skewed image while setting up KV Extraction and properly defining Key and Value regex expressions still doesn’t give you proper results.
You may see mismatch / change in order of text in data on what actually exist on image to open eye or even see data not getting extracting when clicking on Test Extraction button on the KV rules UI.
Reason for this behavior:
This is observed only in case images are skewed and due to a bug Recostar Deskew option is unable to deskew this skewed image due to which we are seeing differences in results.
One of the workaround to come over this is to use -deskew 40% option in ImageMagick parameters, please note that you need to have ImageMagick selected for TIFF conversion process. Once this deskew property is set then the skewed PDF will get deskewed with this property in place and will produce correct results during Test Extraction functionality and when document uploaded as a Batch through drop folder or upload Batch functionality.
You will still not see correct results on KV Screen because internally it doesn’t use ImageMagick for conversion.