Image Normalization in the range 0 to 1

How to normalize an image in the range 0 to 1?

Answer 1

[Image Analyst]

Use mat2gray() or im2double(). Hopefully you have the image processing Toolbox.

normImage = mat2gray(yourImage);
normImage = im2double(yourImage);

Also look at stretchlim() and imadjust().

[Image Analyst]

Sapam - are you still alive???

[Sapam Jaya]

i was sick..thanx

[Image Analyst]

So does my answer solve your question?

[Sapam Jaya]

i guess no,the value is coming 0.the thing is i have to do dwt then get the low frequency (LL)component.in this LL i have to do a log average transform.the result of this has to be normalized from 0 to 1.in this log average i have to use a rectangular region.is it same as rectangular window.i dont understand this window much so maybe my output is wrong

[Image Analyst]

You can do your log, and then pass that in to mat2gray() like I showed you. The image is already rectangular so I don't know why you say you have to use a rectangular window and don't understand it. There's nothing to do since it's already a rectangle, unless you want a different sized rectangle, for which you can use rbbox() or imrect().

[Sapam Jaya]

the rectangular window is as per the instruction..this clears much of the doubt..thanx

[Image Analyst]

Can you go ahead and Accept the answer to close it out? Thanks in advance.

[Andrea Daou]

How can these normalized images between 0 and 1 be saved in a folder ? Because after I normalized them and saved them the images obtained remain with pixel range between 0 and 255.

Thank you in advance!

[Image Analyst]

If you really want them to have fractional values between 0 and 1, that's not a standard image format so you'll just have to save them with

save('myFile.mat', 'yourImageVariable');

to save them in a .mat file.

[Andrea Daou]

Thank you for your help.

Answer 2

[mutant]

Old question but as of R2017b, rescale does exactly this.

B = rescale(A); % Normalizes image (0 to 1, by default)

You can even use this to scale to uint8, for example:

B = rescale(A,0,255); % Normalizes image to [0 255]

Documentation here:

https://www.mathworks.com/help/matlab/ref/rescale.html

Answer 3

[Azzi Abdelmalek]

If im is your image

im=(im-min(im(:)))/(max(im(:))-min(im(:)))

[Jos (10584)]

I suggest to do this in two steps to avoid the calculation of MIN twice ...

IM = IM - min(IM(:)) ;
IM = IM / max(IM(:)) ;

Answer 4

[Sajid Khan]

function image_normalized = imnormalize( image_orig, min_norm, max_norm)
    val_max = max(image_orig(:)); 
    val_min = min(image_orig(:)); 
    range = val_max - val_min; 
    image_normalized = (image_orig - val_min) ./ range; % Then scale to [x,y] via:
    range2 = max_norm - min_norm;
    image_normalized = (image_normalized*range2) + min_norm;
end

In this function, you can set min_norm = 0 and max_norm = 1 to normalize image to a scale of 0 to 1. If you have any other questions to ask, then you are welcome. I always use this function for normalization purpose. It even works if you have to increase the scale length.

[Image Analyst]

Your function basically does the same thing as the built in function mat2gray().

[DGM]

Rather, mat2gray() only allows the specification of the input levels, assuming the output levels are [0 1]. Sajid's function allows specification of the output levels, while using image extrema as the input levels.

So in that sense, it's more like rescale(), same syntax and everything.