How can i get the actual dpi of a picture in pdf

[R4Y-R4Y]

Hello
so i am using PyMuPDF to extract pictures for ocr scanning
i want to check the dpi of pictures so i wont scan pictures that are basically unusable for tesseract
from what i saw in the docs there's pic.xres or pic.yres but it's always by default 96
I've tested it with two pictures in documents but its always the same
can someone please help me :)

[JorjMcKie]

I hope you are extracting the "pictures" via doc.extract_image(xref). Given this, if you make a Pixmap of the image or, equivalently, a Pillow Image, then you will get access to the DPI value stored in it. This is all that is available. Depending on the packages, a few default assumptions are made if no, or an invalid DPI is stored.

[R4Y-R4Y]

okay so i have tested this with this do here
here's the current code :

def scanned_dc(pdf):
    for j in range(pdf.page_count):
            img_pdf = pdf.get_page_images(j)
            for image_page in img_pdf:
                xref = image_page[0]
                res = pdf.extract_image(xref)
                fp = io.BytesIO(res["image"])
                x = Image.open(fp)
                print(x.info["dpi"])

and i got this error:

File "********************************************************************", line 34, in scanned_dc
    print(x.info["dpi"])
KeyError: 'dpi'

idk what to do :)

[R4Y-R4Y]

also it worked with another doc and it said it had 72 dpi instead of the default 96 but couldn't convert it to numpy with this function:

def pix2np(pix):
    """
    Converts a pixmap buffer into a numpy array
    """
    # pix.samples = sequence of bytes of the image pixels like RGBA
    #pix.h = height in pixels
    #pix.w = width in pixels
    # pix.n = number of components per pixel (depends on the colorspace and alpha)
    im = np.frombuffer(pix.samples, dtype=np.uint8).reshape(
        pix.h, pix.w, pix.n)
    try:
        im = np.ascontiguousarray(im[..., [2, 1, 0]])  # RGB To BGR
    except IndexError:
        # Convert Gray to RGB
        im = cv2.cvtColor(im, cv2.COLOR_GRAY2RGB)
        im = np.ascontiguousarray(im[..., [2, 1, 0]])  # RGB To BGR
    return im

it gives me this error:

  File "****************************************************************", line 16, in pix2np
    im = np.frombuffer(pix.samples, dtype=np.uint8).reshape(
AttributeError: 'dict' object has no attribute 'samples'

the stars is just for hiding the files

[JorjMcKie]

'dict' object has no attribute 'samples'

Obviously pix was not a Pixmap!

[R4Y-R4Y]

yeah i found out how to change it by simply changing its type to np.array x = np.array(x)
thanks for your help you're really a savior and helped me a lot :)

[JorjMcKie]

pleased to hear that!

[JorjMcKie]

idk what to do :)

neither do I, without any way to reproduce the situation.

[JorjMcKie]

Ah got the file.
Yes that image has no embedded dpi info. That dict is empty. If this occurs (use try/except to catch), assume 96 like (Py-) MuPDF does.

[R4Y-R4Y]

oh that's interesting
so its value is None or sth like that?

[JorjMcKie]

No the PIL image has image.info = {}.

[R4Y-R4Y]

No the PIL image has image.info = {}.

so basically an empty dictionary
cool
thanks for the support :)