How does Acrobat extract vector graphics not inside an XObject

[Graeme_Kidd@adobeforums.com]

I was wondering how Adobe extracts images?

I am currently using Java to access the content stream in a PDF with vector graphics. I can extract the path data but the individual images don't seem to be within a XObject or Form XObject so if there are 4 vector images within a page I can only export them as one file rather than 4 different images.

Despite this Adobe Acrobat was able to export the 4 vector images separately no problem. How was it able to do that when they don’t seem to be grouped? Then would I be able to copy that functionality by writing a plug-in?

Thanks

[Leonard_Rosenthol@adobeforums.com]

Without seeing a sample PDF that demonstrates this, I can't really comment.

Also, what function in Acrobat did you use to "export vector graphics"?

[Graeme_Kidd@adobeforums.com]

"Also, what function in Acrobat did you use to "export vector graphics"?"
I was using Acrobat 7.0.5 Professional and went to:
Save As -> Save As Type: HTML

It then saved all the images to a separate folder including the vector graphics.

The "Export All Images" functionality in the "Advanced" menu can’t do this as it complains that it can only extract bitmap or raster images.

[Leonard_Rosenthol@adobeforums.com]

In the SaveAsHTML case, we convert all vector information into raster. We aren't necessarily seeing them as separate "images" (hence the error message when trying to export images) but as separate "areas" on the page that need to be exported for use from HTML.

[Graeme_Kidd@adobeforums.com]

Ah OK that was very helpful,

So if there is nothing in the PDF that says this section of path data is part of an image you simply attempt to group the path data into areas. I imagine this could be done before or after the rasterisation process, either by looking for white spaces or applying some hit test on the positioning of the path data.

Thanks very much, that has cleared a lot of things up for me.

[Leonard_Rosenthol@adobeforums.com]

Pretty much, yes.

Some PDFs, however, DO have "tags" that identify individual images - look at the section of ISO 32000-1 on Tagging & structure along with Marked Content.