CODEWITHSUNDEEP
ocrrdfexifxmpdublin-core
Vasily A
Vasily A

Reputation: 8646

Any conventional standards for storing OCR data/metadata in JPEG images?

I want to organize a collection of scanned documents (receipts, bank statements, etc.) by adding their metadata and text content (OCR'ed) into the same jpeg files. Is there any more or less commonly accepted way of storing such data? Any commonly used schemas?
For metadata, for example - I found a Dublin Core scheme, but most of the fields I want are not there, and I'm not sure what's the good way to add custom fields - can I just use them like if they existed in DC or XMP scheme (i.e. <dc:myfield>myvalue</dc:myfield> or <xmp:myfield>myvalue</xmp:myfield>), or I have to define my own scheme by adding xmlns:myScheme="http://myScheme.uri" and then use it as <myScheme:myfield>myvalue</myScheme:myfield> ?
Also, in all the examples I found, this data is stored inside <rdf:Description> which is inside <rdf:RDF> which is inside <x:xmpmeta> - is it a standard requirement? I don't see it in the XMP specification for storage in files...

For now, based on the examples, I plan to embed something like this:

    <?xpacket begin='' id='W5M0MpCehiHzreSzNTczkc9d'?>
    <x:xmpmeta xmlns:x='adobe:ns:meta/' x:xmptk='MyTool v 0.0.1'>
        <rdf:RDF xmlns:rdf='http://www.w3.org/1999/02/22-rdf-syntax-ns#'>
            <rdf:Description rdf:about='' 
                            xmlns:dc='http://purl.org/dc/elements/1.1/'
                            xmlns:myDoc='http://some.custom.uri/'>
                <dc:format>image/jpeg</dc:format>
                <myDoc:doctype>scan</myDoc:doctype>
                <myDoc:originalfilename>20190519121225_003.jpg</myDoc:originalfilename>
                <myDoc:originalimagewidth>1684</myDoc:originalimagewidth>
                <myDoc:originalimageheight>2788</myDoc:originalimageheight>
                <myDoc:langOCR>EN-US</myDoc:langOCR>
                <myDoc:acquisitiondatetime>2019-05-19T12:12:25Z</myDoc:acquisitiondatetime>
                <myDoc:documentdate>2019-01-02</myDoc:documentdate>
                <myDoc:pagesindocument>6</myDoc:pagesindocument>
                <myDoc:page>2</myDoc:page>
                
                <myDoc:textcontent>
          Bank
          statement
          02/01/2019
          Page 2 of 6
                ( Here goes raw OCR content 
                    as multiline text )
                </myDoc:textcontent>
                
                <dc:subject>
                    <rdf:Bag>
                        <rdf:li>bank</rdf:li>
                        <rdf:li>statement</rdf:li>
                    </rdf:Bag>
                </dc:subject>
            </rdf:Description>
        </rdf:RDF>
    </x:xmpmeta>
    <?xpacket end='w'?>

Does it make sense at all? I'm sure many people already worked on similar tasks, I don't want to reinvent the wheel...

Upvotes: 3

Views: 262

Answers (0)

Related Questions