How does Matlab function imresize calculate values on borders of image

Hi, I am wondering how work Matlab function imresize with bicubic interpolation. I understand how it works in the middle of image where it have full access to all data, then usage of bicubic interpolation is quite straight forward.

A small example:

>> A= [3,1,2,6;1,5,2,7;1,3,6,2;4,1,9,2]

A =

     3     1     2     6
     1     5     2     7
     1     3     6     2
     4     1     9     2
>> imresize(A,size(A)*2,'bicubic')

ans =

    3.3109    2.6187    1.1206    0.7166    1.4067    2.8883    5.1615    6.2060
    2.6454    2.4435    2.0052    1.7138    1.5692    2.8062    5.4248    6.6307
    1.2008    2.0609    3.9181    3.8763    1.9355    2.6382    5.9844    7.5302
    0.4985    1.6796    4.2357    4.6921    3.0489    3.2618    5.3307    6.2876
    0.5384    1.2996    2.9579    4.1612    4.9094    4.6769    3.4639    2.9028
    1.5500    1.6768    1.9697    3.5367    6.3776    6.1378    2.8173    1.2798
    3.5334    2.8110    1.2710    2.8185    7.4537    7.6444    3.3907    1.4186
    4.4466    3.3344    0.9529    2.4891    7.9430    8.3334    3.6605    1.4930

But how does Matlab compute values on borders (for example in the upper left corner of the image/Matrix) where there is not enough values for the algorithm? Does it somehow simplified the algorithm? Or add some dull (zeros, ones?) values on the parts where algorithm goes beyond the defined matrix?

Thank you really much for any help and I wish you great weekend :) Maciek

[Paulo Silva]

edit imresize

Answer 1

[Steve Eddins]

When the interpolation computation needs pixel values outside the domain of the image, it assumes they are equal to the nearest boundary pixel. For example, it assumes that the pixel value at the out-of-bounds location row = 5, column = -1 equals the pixel value at row = 5, column = 1.