How to remove least significant bits in a pixel.

i read a image

img=imread('lena.jpg');

now i need to remove 3 least significant bits form every pixel of the image, how can i do that?

can anyone please provide a solution.

Answer 1

[Prasanna]

Hi Niranjan,

It is my understanding you want to remove the 3 least significant bits (LSBs) from every pixel in the image. Each pixel in an 8-bit image has values from 0–255. In binary, this is 8 bits. To remove the 3 LSBs, you essentially set them to 0, keeping only the 5 most significant bits (MSBs). This can be done using bitmasking by keeping a binary mask of 248 and using “bitand(pixelValue, 248)” removes the 3 LSBs.

img = imread('lena.jpg');

% Initialize output image
img_clean = zeros(size(img), 'uint8');

for channel = 1:3
    img_clean(:,:,channel) = bitand(img(:,:,channel), 248);
end

imshowpair(img, img_clean, 'montage');

In the above code “bitand(img, 248)” applies a mask to zero out bits 0–2 of each pixel value. For more information, refer to the following documentations:

• bitand: https://www.mathworks.com/help/matlab/ref/bitand.html
• imshowpair: https://www.mathworks.com/help/images/ref/imshowpair.html

Hope this helps!

[Image Analyst]

A more general version that is more robust and will work with other than uint8 images and color images is

% Initialize output image
img_clean = zeros(size(img), class(img));
for channel = 1 : size(img, 3) % Works for color as well as gray scale.
    img_clean(:, :, channel) = bitand(img(:, :, channel), 248);
end

[Walter Roberson]

@Image Analyst

That code will not work with float images.

(Mind you, with floating point images, it is a bit questionable as to what "the last three bits" means.

if isa(img, 'integer')
    mask = intmax(class(img)) - 7;
    img_clean = bitand(img, mask);
elseif isa(img, 'logical')
    img_clean = false(size(img));
elseif isa(img, 'single')
    img_clean = round(img .* 2^29)./2^29;
elseif isfloat(img)
    N = 2+log2(eps(ones(1,class(img))));
    img_clean = floor(img .* 2.^(-N)) .* 2.^N;
else
    error('unhandled image type: %s', class(img));
end

Answer 2

[Image Analyst]

A simple method that works for both color and gray scale images is to use bitand:

inputImage = imread('lena.jpg');
% Set last 3 bits to 000.
% Works for both color and gray scale images.
outputImage = bitand(inputImage, 248);
imshowpair(inputImage, outputImage, 'montage');