0
$\begingroup$

I am looking for possible professional terms. In a binary (1 or 0) image, normally, the areas with "1"s are where the information is. I wonder if there is any professional name for it in image processing. Thank you!

$\endgroup$
4
  • 1
    $\begingroup$ Why should we have a specific name for these? Also, the 0 pixels a priori have the same information content as the 1 pixels. So, I'm not even sure it makes sense what you're asking. If you're actually dealing with visual content, you can simply make an arbitrary color mapping, for example 1 = yellow, 0 = red, and call the "1 areas" simply "yellow areas", but that doesn't make it any less arbitrary. $\endgroup$ Apr 18, 2018 at 7:03
  • $\begingroup$ @Marcus Müller. I new to image processing. Sorry for not making the question clear. What I am wondering is that in the binary image case, is there any way to refer to the 1 pixels without defining them (such as 1 = yellow, 0 = red)? This question arises in my study of morphological operations in binary image. For example, the dilation operation "expands" the 1 pixels in the binary images, but to express it this way seems a bit off. That is why I am searching a possible name of it in image processing. $\endgroup$
    – ZHUANG
    Apr 18, 2018 at 8:57
  • 2
    $\begingroup$ Foreground and Background pixels are the specific terms used. This is really fundamental terminology found in any textbook on the topic. I don't understand the motivation behind the question (?) $\endgroup$
    – A_A
    Apr 18, 2018 at 9:33
  • $\begingroup$ I agree with A_A about the lack of motivation. Notice that there's no law that says "0 is always a background pixel"; it's just what often happens. $\endgroup$ Apr 18, 2018 at 9:35

2 Answers 2

1
$\begingroup$

This problem born when you use pixel intensity as physical mass, for example to calculate mass center or inertia moments, in this case the background should be 0 and object should be 1.

Btw your question is not clear because in a binary image {0,1} have the same information content and generally there are not difference.

There is not a name for 1 pixel and 0 pixel, often in common convention 1 is white and 0 is black, but it's not a general rule. Maybe you can call 1 high value pixel and 0 low value pixel but I think that are not common names and they can make confusion in some contexts.

$\endgroup$
11
  • $\begingroup$ Sorry for not being very clear. I am new to image processing. I am just wondering if there is any name referring to 1 pixels or 0 pixels in binary images. $\endgroup$
    – ZHUANG
    Apr 18, 2018 at 9:02
  • $\begingroup$ Generally 0 is black 1 is white $\endgroup$
    – Andrea
    Apr 18, 2018 at 9:16
  • $\begingroup$ @Andrea is it? not in my world. Or, to be more precise: it is a common convention, yes, but I'd still be careful to point out that it's still arbitrary and to document that 0 is black and 1 is white. $\endgroup$ Apr 18, 2018 at 9:30
  • $\begingroup$ Yes, in general it's false, I know but I meant as common convention ("generally" not ever) $\endgroup$
    – Andrea
    Apr 18, 2018 at 9:42
  • $\begingroup$ "There is not a name for 1 pixel and 0 pixel" -- This is simply not true. $\endgroup$ Apr 18, 2018 at 22:49
-1
$\begingroup$

A binary image conventionally distinguishes "foreground" from "background", or "objects" from "background".

More often than not, the foreground pixels, which form the objects, have a value of 1, but some libraries will use a value of 255 (storing the binary image in a 8-bit unsigned integer array). Conversely, the background typically has a value of 0.

The reason to select 0 for background and a non-zero value for foreground is that (at least in every programming language I know) 0 is evaluated as false and non-zero as true. This makes subsequent logic more intuitive.

I have personally never come across a scientific paper on image processing or analysis that reverses this convention, but I have seen plenty of papers that print background as white and foreground as black -- When most pixels are background, this makes the foreground stand out better in print.

Nonetheless, I'm sure there are programs out there that use 0 for foreground. It's not written in stone. It is simply the very common convention.


(Edit to preserve the comment I wrote to the other answer.)

Note that these terms and conventions go back to the beginnings of the field of digital image processing, and probably earlier too.

Azriel Rosenfeld was one of the pioneers of the field. In this book from 1976, "Digital Picture Processing" (Ch. 8), he says

[...] the function which has the value 1 for points belonging to the object, and value 0 elsewhere.

And a little later

[...] deciding whether a given point of the picture belongs to an object or to the background.

$\endgroup$

Not the answer you're looking for? Browse other questions tagged or ask your own question.