interp2 changes the value of pixels also

I am using interp2 as below to defrom my image. It imeans I want the places of some pixels change according to us and uy. The problem is that when I use interp2 it changes the value of pixels also. In my image the value of pixels are only 0, 1 , 2 , 3 but when I read the image I get after interp2 some values such as 2.67 appears also. I do not aim to change the value of pixels. I only want to change the places of pixels with the previous values (defrom the image). Why this happens and is there any other approach?

donoutdef = interp2(xx,yy,double(donut),xx+ux,yy+uy)

Answer 1

[Star Strider]

‘In my image the value of pixels are only 0, 1 , 2 , 3 but when I read the image I get after interp2 some values such as 2.67 appears also. I do not aim to change the value of pixels.’

There are several ways to interpolate. If you want the interpolated values to be those of the closest pixel in your image, choose 'nearest' as the method. See the documentation section on method (link) for details.

You can also use the round, floor or ceil functions on fractional values to force them to the closest integer values.

Answer 2

[KALYAN ACHARJYA]

Yes, those are the interpolated values. You have the image having pixels values are 0, 1 , 2 , 3, when you interpolated new pixels, the values of new pixels are depends on method of interpolation use. I dont see any issue here.

What is original size of the image?

What image size you are expecting after resize?

[Golnoush Asaeikheybari]

I do not want to change the values. I just want to deform the image with the same pixel values. Is there any other way?

[KALYAN ACHARJYA]

After image deformation, there may be some pixels are changed.

Suppose following is the image (please note image size 2x5)

0 1 0 2 0
3 2 0 1 3

Now what you want?

[Golnoush Asaeikheybari]

I want each pixels goes ahead 2 times with the same value like this:

NaN NaN 0 1 0

NaN Nan 3 2 0

Or any other transformation but with exactly the same value.

[KALYAN ACHARJYA]

>> Image=[0 1 0 1 0;1 0 3 2 0]
Image =
     0     1     0     1     0
     1     0     3     2     0
>> Image(:,1:2)=NaN
Image =
   NaN   NaN     0     1     0
   NaN   NaN     3     2     0
>> 

or if the image is very large, then go for interpolation (nearest method) as suggested by @StarStrider.