Page Background Color

[meghanaviyyapu]

Can PyMuPDF extract background color of a page or text?

[JorjMcKie]

No, because you cannot define this in a unique way, can you?

But you can create a pixmap of the bbox of some text and check whether it is unicolor (only one color), whether there is a predominant color: pix.color_topusage(). If this returns that for example 95% of the area is colored the same way, then the text will probably not be readable.

[meghanaviyyapu]

Ok thank you

[meghanaviyyapu]

Can you let me know how a pixmap of bbox of text can be created using PyMuPDF?

[JorjMcKie]

I assume you want to check whether some given text is actually visible, right?
First thing you need to know is the bbox (boundary box) of that text.

Then do pix = page.get_pixmap(clip=bbox). As mentioned above you now can check pix.is_unicolor. If true, the text is definitely invisible.

To lower this very strict condition, you can alternatively compute top_color = pix.color_topusage(). This a tuple (ratio, pixel), where ratio is a float 0 < ratio <= 1 indicating a relative frequency of color pixel.
pixel is a bytes object representing a pixel value. If ratio is close to 1, then "almost all" pixels have a color pixel, which would mean that the text is "almost" invisible.
If ratio == 1, then this is equivalent to being unicolor.

If you need to check many text bboxes on the same page, it makes sense performance-wise to create the pixmap of the full page once only, and then check pix.color_topusage(clip=bbox) for each of the bboxes. This approach avoids computing multiple pixmaps (which can be time consuming).

[meghanaviyyapu]

I want to extract the background color of text with bbox value. Can you let me know if the color of background of selected bbox can be extracted?

[JorjMcKie]

Same answer: there is no such thing like a background color. The best approximation is what I showed you.

[meghanaviyyapu]

How to get RGB value from pix.color_topusage().?

[JorjMcKie]

the pixel value content

[JorjMcKie]

I now have a decent internet connection again. Second item in the tuple (ratio, pixel) is a bytes object of length 3 (4) if the Pixmap is RGB(A). So the items of this object represent the integers red, green, blue (, alpha).

[meghanaviyyapu]

Thanks so much for the explanation