I want to compress large tiff file (1.5gb) to atleast 100mb or less to the same file type (tiff). I need to maintain the original resolution and metadata for co-ordinate points. Is there any built-in matlab function that i can use to solve this?

does imwrite2tif function help?

Answer 1

[DGM]

We can't afford to change depth, resolution, or use lossy compression; nor can we discard any extraneous metadata. Given that, the ability to reduce the filesize depends mostly on the image content and whether the original image is already compressed. Since nobody ever knew either of these things, nobody would have known if it were a realistic goal.

Assume you had a very simple synthetic image. I'm going to ignore metadata and assume a simple uint8 single-channel intensity image.

% pick a fixed image geometry
N = round(sqrt(200E6)); % approx 200MB uncompressed

% a simple 8x8 checkerboard pattern
A = im2uint8(checkerboard(round(N/8)));
dumpfilestats(A) % it compresses a lot

bigfilesize = 200.0534

smallfilesize = 0.5363

ratio = 373.0517

% a photograph of the same size
A = imread('cameraman.tif');
A = imresize(A,[N N]);
dumpfilestats(A) % it compresses less

bigfilesize = 199.9969

smallfilesize = 16.1685

ratio = 12.3695

% a random noise image of the same size
A = randi([0 255],N,N,'uint8');
dumpfilestats(A) % it doesn't compress at all

bigfilesize = 199.9969

smallfilesize = 200.0583

ratio = 0.9997

function dumpfilestats(A)
    % say you had an uncompressed version
    fn1 = 'testbig.tif';
    imwrite(A,fn1,'compression','none');
    S = dir(fn1);
    bigfilesize = S.bytes/1E6
    
    % you could compress the same data
    fn2 = 'testsmall.tif';
    imwrite(A,fn2,'compression','deflate');
    S = dir(fn2);
    smallfilesize = S.bytes/1E6
    
    % how much would it compress?
    ratio = bigfilesize/smallfilesize
end

Obviously, it the original file is already compressed, that changes what's possible. If it's lossy compressed, then the whole data integrity boat has already sailed.