CODEWITHSUNDEEP
matlabimage-processing
user1234
user1234

Reputation: 437

What is the need of normalisation in image?

I saw this code in the internet

imgN = double(img-min(img(:)))/(max(img(:)-min(img(:))));

to normalize the image in 0..1 . What's the meaning of it? Also its shows this error

Error using  / 
Integers can only be combined with integers of the same class, or scalar doubles.

Upvotes: 1

Views: 191

Answers (1)

hbaderts
hbaderts

Reputation: 14316

In MATLAB, many functions make assumptions about the range of values in an image. Usually, images are stored either as uint8 or double. In the case of uint8, the range is 0...255 (which is all possible values in a uint8.). For double, MATLAB chose a value range from 0...1. To use functions like imshow, imwrite and others correctly, your double image has to be scaled correctly, otherwise some pixels are not displayed correctly.

There is a small bug in the code: It is basically a division a / b, where a is double(img - min(img(:))) and b is (max(img(:)) - min(img(:))). If your input image is of type double, this will work fine. However if it is uint8 or any other type, the numerator a will be converted to double, while the denominator b stays e.g. uint8. You will have a division double / uint8. MATLAB tells you in the error message, that it can not mix types in division. An easy solution is to also convert the denominator b to double:

imgN = double(img-min(img(:))) / double(max(img(:)-min(img(:))));

Upvotes: 4

Related Questions