how to detect background color of image ?

i want to make function, so if the background color if black like image 1, it will change to white.

But, if the background color image is white like in image 2, so it doesn't change.

How to make it ?? thanks

[MJ Thangaraj]

The image is Binary so it's obviously going to have only White and BLACK values.Check whether the background in white and then Complement the image .

Answer 1

[Image Analyst]

If you can make the assumption that the majority of the pixels in the image are the background color, then

[rows, columns] = size(binaryImage);
numWhitePixels = sum(binaryImage);
numBlackPixels = rows * columns - numWhitePixels;
if numWhitePixels > numBlackPixels 
    % Background is white.
    % Do nothing at all.
else
    % Background is black.
    % Make image all white, everywhere at every pixel.
    binaryImage(:) = true;
end

[ElizabethR]

thanks for answare my question. i try it but, the output is only image withh all of white pixel

[Image Analyst]

Yes, exactly. Because you said "background color if black like image 1, it will change to white." So if the background is black, and it changes to white, and if the foreground is white and it stays white, then of course your whole image is white, exactly as you specified.

If you want something different than that, then say exactly what you want. Like if you want the black background to become white and want the white foreground to become white then say you want to invert the image, like bw=~bw.

[ElizabethR]

thanks for answare, yes you are right, thanks for you explanation. Sorry, because my english is not good enough. I mean that i have image like image one, then the background will change into black and the foreground will change into white. And if the image is like image 2, then the backgroun and the foreground will not change. hanks you ^^

[Image Analyst]

OK, then this should do it

bw = ~bw; % Change black into white and white into black.

[ElizabethR]

okay .. Thank You so Much Image Analyst. God Bless ^^

Answer 2

[Walter Roberson]

Foreground and background are matters of intent. For example, often binary images are white for the parts that contain the information of interest, but binary images might be representing text and text is often represented in black (corresponding to books, which use dark ink on a white page.) Chess diagrams often use both black and white for the pieces. It is therefore not possible to detect which color is the "foreground" and which color is the "background" by computer program alone.

[ElizabethR]

thanks for answare my question Walter ^^

Answer 3

[Luis Rosety]

This is a very old question but I am learning Matlab and I got the same problem and just in case anybody else has the same question, I contribute with my own solution.

I realized it was quite straightforward the answer.

Assuming the input image is IM:

if(size(find(IM),1) > size(find(~IM),1))
    % IM is white background
else
    % IM is black background
end 

[DGM]

Or much faster:

nz = nnz(IM);
if nz > (numel(IM)-nz)
    % IM is white background
else
    % IM is black background
end 

You might also be able to consider the dominant value around the image periphery as some indicator of "background". I agree with Walter that the general solution requires knowledge of content and intent.