Image extraction

[meghanaviyyapu]

There is one image in the PDF but page.get_images() gives the following output
[(18, 0, 354, 118, 8, 'DeviceRGB', '', 'I0', 'DCTDecode'), (19, 0, 1181, 1772, 8, 'DeviceRGB', '', 'I1', 'DCTDecode'), (18, 0, 354, 118, 8, 'DeviceRGB', '', 'I0', 'DCTDecode')]

Can you let me know how the image file and bbox coordinates can be extracted?

In the internal structure it is represented as Form XObject with /BBox [ 0 0 595.276 76.53543 ] but the bounding box is not getting drawn with these values.
example_010.pdf

[JorjMcKie]

The method page.get_images() reports what is contained in the PDF page object definition. This is not necessarily a list of images actually displayed by the page. Please read the documentation about that method and section "Block Dictionaries" within the TextPage chapter.
Page 1 of your example displays only one image. Execute first page.clean_contents() and then page.get_images(), and you will see only one item. This is because cleaning synchronizes the object definition with the appearance source of the page.
Otherwise, detecting the bbox of an image works via page.get_images_rects(xref).

[meghanaviyyapu]

Thanks. May I know how page.get_images_rects(xref) computes the bounding box values?

[JorjMcKie]

This happens based on page.get_image_info() combined with page.get_images - see the source in utils.py.

[meghanaviyyapu]

Is it possible to calculate bbox values with width and height of image, page coordinates?

[JorjMcKie]

Is it possible to calculate bbox values with width and height of image, page coordinates?

What does that mean? If you are asking for the relationship between image dimension (width, height) and the bbox dimension on the page, then please look at the documentation here.

[meghanaviyyapu]

Had a look.
Would like to know the internal working of "page.get_image_bbox("fzImg0", transform=True)" or how the bbox values can be calculated without using page.get_image_bbox()

[JorjMcKie]

Well, there are the source codes in utils.py and fitz.i. The main logic is in get_image_info(xrefs=True) which uses the page's TextPage. Using the MD5 code of the extracted image, a match is tried with the items in page.get_images().

[meghanaviyyapu]

Thanks for the explanation