Oh wow, my brain was completely off (I just rolled out of bed). I'm just now seeing Toël's detailed dump of the PDF image info.
Thanks again! On Fri, Mar 11, 2016 at 8:17 AM, Tilman Hausherr <[email protected]> wrote: > Am 11.03.2016 um 17:16 schrieb Vince Harron: > >> Hi Toël, >> >> Thanks for your reply. But I guess my question is more about the pdf >> file. Is my code extracting the image out of page 2 pixel perfect or is >> it >> resampling the page? >> > > The code is fine (for 1.8). Google uses two different sizes. No idea which > one came first. > > Tilman > > > >> >> >> On Fri, Mar 11, 2016 at 1:06 AM, Hartmann Toël < >> [email protected]> >> wrote: >> >> Hi, >>> >>> The dpi information embedded in the image is 300 for EzFQJ9v.png but on >>> US08000000-20110816-D00001.png it is 72. >>> I extracted the image of the head only from both the pngs and get two >>> different pixel size: >>> >>> the head in EzFQJ9v.png is 1722x1593, the head in >>> US08000000-20110816-D00001.png is 1331x1231. >>> >>> I would say that Google has a resized image and changed the dpi info to >>> 72. >>> >>> The image info for the pdf page is: >>> position in PDF = -1.2, 0.0 in user space units >>> raw image size = 2560, 3300 in pixels >>> displayed size = 614.4, 792.0 in user space units >>> displayed size = 8.533334, 11.0 in inches >>> displayed size = 216.74667, 279.4 in millimeters >>> dpi = 300 dpi (X), 300 dpi (Y) >>> >>> >>> >>> >>> /Toël >>> >>> On 11 mar 2016, at 09:14, Vince Harron <[email protected]> wrote: >>> >>> Here is the original patent from the US Patent and Trademark Office: >>>> >>>> http://pimg-fpiw.uspto.gov/fdd/00/000/080/0.pdf >>>> >>>> I'm extracting images as follows: >>>> >>>> List<PDPage> list = document.getDocumentCatalog().getAllPages(); >>>> >>>> String fileName = srcPdfFile.getName().replace(".pdf", "_cover"); >>>> int imageNumber = 0; >>>> for (PDPage page : list) { >>>> PDResources pdResources = page.getResources(); >>>> >>>> Map pageImages = pdResources.getImages(); >>>> if (pageImages != null) { >>>> >>>> Iterator imageIter = pageImages.keySet().iterator(); >>>> while (imageIter.hasNext()) { >>>> String key = (String) imageIter.next(); >>>> PDXObjectImage pdxObjectImage = (PDXObjectImage) >>>> pageImages.get(key); >>>> >>>> pdxObjectImage.write2file(srcPdfFile.getAbsolutePath().replace(".pdf", >>> >>>> String.format("-D%05d.png", imageNumber))); >>>> imageNumber++; >>>> } >>>> } >>>> } >>>> >>>> The image I extract from page 2 looks like this: >>>> http://i.imgur.com/EzFQJ9v.png >>>> 2560x3300 (300dpi) >>>> >>>> Here is the same image from Google Patents >>>> >>>> >>>> >>> https://patentimages.storage.googleapis.com/US8000000B2/US08000000-20110816-D00001.png >>> >>>> it's only 1446 × 2037 (~224dpi) >>>> >>>> The Google image is cropped a bit compared to the PDF page. When I trim >>>> the my PDF page image down to match the same area as the Google image, >>>> >>> the >>> >>>> my extracted image is still much higher resolution than the Google >>>> extracted image (1934 × 2550) >>>> >>>> Assumption 1) Google is using the same data source as me (PDF) >>>> Assumption 2) Google wouldn't downscale technical diagrams in patents >>>> because they might lose important detail >>>> >>>> If my assumptions are correct, I must be extracting the image >>>> >>> incorrectly, >>> >>>> upsampling the ~224dpi image to 300dpi. Is that what's happening? >>>> >>>> Thanks, >>>> >>>> Vince >>>> >>> >>> --------------------------------------------------------------------- >>> To unsubscribe, e-mail: [email protected] >>> For additional commands, e-mail: [email protected] >>> >>> >>> > > --------------------------------------------------------------------- > To unsubscribe, e-mail: [email protected] > For additional commands, e-mail: [email protected] > >
