How to extract XObject or Inline Images, Image Masks

Extracting Inline Images is discussed in tutorial Extracting Page Images, so let’s focus on XObject Images and Image Masks.

Extracting XObject Image

Open a sample document.

>>> from pdfreader import PDFDocument
>>> fd = open(file_name, "rb")
>>> doc = PDFDocument(fd)

Have a look at the sample file sample file. There is a logo on the first page. Let’s extract it.

>>> page = next(doc.pages())

Let’s check a dictionary of XObject resources for the page:

>>> page.Resources.XObject
{'img0': <IndirectReference:n=11,g=0>}

This stands for an XObject named img0, and referenced under number 11 and generation 0. The object has not been read by pdfreader still. We are lazy readers. We read objects only when we need them. Let’s see what the object is.

>>> xobj = page.Resources.XObject['img0']

We just read the object (__getitem__ does this implicitly) and now we may access its attributes.

>>> xobj.Type, xobj.Subtype
('XObject', 'Image')

Wow! It’s really an image. Should we care about it’s internal PDF representation? Of course no, let’s just convert it to a Pillow/PIL Image and save.

>>> pil_image = xobj.to_Pillow()

And here we are!


Try to open it and see any differences. It’s absolutely the same as in the document.

Now you can manipulate pil_image with usual PIL methods: rotate, convert, blur, split, inverse, merge and so on, so on, so on.

Extracting Images: a very simple way

A very simple way also exisits. Use SimplePDFViewer:

>>> from pdfreader import SimplePDFViewer
>>> fd = open(file_name, "rb")
>>> viewer = SimplePDFViewer(fd)
>>> viewer.render()

After rendering all 1st page images are on the canvas

>>> all_page_images = viewer.canvas.images
>>> all_page_inline_images = viewer.canvas.inline_images
>>> img = all_page_images['img0']
>>> img.Type, img.Subtype
('XObject', 'Image')

Now you can convert it with magic to_Pillow() method, save or do whatever you want!

Extracting Image Masks

Image Mask is just a specific kind of image actually. Except it is not always visible directly in your PDF Viewer. Nevertheless it can be accessed absolutely the same way.

Let’s have a look at the example from Extracting Page Images, and see what image masks it contains.

>>> from pdfreader import SimplePDFViewer
>>> fd = open(pdf_file_name, "rb")
>>> viewer = SimplePDFViewer(fd)

We use Image.ImageMask attribute to filter image masks from another images. Let’s go to the 5th page and take the first image mask:

>>> viewer.navigate(5)
>>> viewer.render()
>>> inline_images = viewer.canvas.inline_images
>>> image_mask = next(img for img in inline_images if img.ImageMask)

Now convert it to Pillow object and save:

>>> pil_img = image_mask.to_Pillow()

Have a look! What a beautiful QR-code!