how can normalize the data between 0 and 1??

回答(2 个)

mat2gray() would normalize to exactly 0 to exactly 1.
But what value do you want instead of 0? Should the smallest values be mapped to eps(realmin), which is about 1E-324 ?

3 个评论

the value should be more then zero but not exactly 0 and less than one. it may take any value as you said eps(realmin) also fine
@ananthi: Accepting an answer means, that the problem is solved. Then most readers will not care about the thread anymore. Is the problem solved?

请先登录,再进行评论。

Jan
Jan 2017-2-23
编辑:Jan 2017-2-23
A cheap adjustment of the edges:
x = randn(100, 1);
xn = (x - min(x)) / (max(x) - min(x));
xn(xn == 0) = eps; % Or: eps(realmin)
xn(xn == 1) = 1 - eps;
Or consider the limits during the normalization: [EDITED, first version failed]
xmin = min(x);
xmax = max(x);
range = (xmax - xmin) + eps(xmax - xmin);
xn = (x - (xmin - eps(xmin))) / range;
% Or:
% xn = (x - (xmin - eps(xmax - xmin))) / range;

4 个评论

thank you jan simon.. this logic is working
Hi can I ask what is the function of eps(xmax-xmin)
It's the data that you want to rescale.

请先登录,再进行评论。

类别

标签

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by