Finding neighbours in a matrix

Hi,

If I have an m x n matrix, how can I print a list of all the possible neighbours(horizontally, vertically and diagonally)? For example matrix(1, 1) should only give 3 values, while an element somewhere in the centre of the matrix should give 8 values. I know there are questions on this website resembling mine, but since I'm really new to matlab, I wondered if there are any easier solutions(I don't mind if they're longer than they'd need to be) Thank you very much!

[Steven Lord]

I'm guessing you're not trying to find the neighbors simply to have a list of the neighbors. If you explain how you plan to use this information, what you want to obtain at the end, we may be able to offer more targeted suggestions for how to achieve your end goal.

For instance, since you said you're new to MATLAB, one reason you might want to do this is you've been given a homework assignment to program Conway's Game of Life. If so, after you've written up your version you might be interested in reading the chapter on the Game of Life in Cleve Moler's Experiments with MATLAB.

Answer 1

[Image Analyst]

You can use blockproc(). It can do that. See attached test.m demo. If you don't want to have it "jump" by 3 pixels every time, but want it to move over by every time, you have to overlap the windows like I do in the other attached demo. See if you can adapt it, if you need that (being a smart engineer, I'm sure you can).

[Walter Roberson]

Naomi Pudney comments

I downloaded these and now MATLAB is giving me a huge error message and won't work properly.

[Image Analyst]

I just downloaded test.m and it ran fine:

Same for blockproc_overlapping.

What error did you get? You forgot to say. Also, attach your m-file.

Answer 2

[KALYAN ACHARJYA]

You can do it one by one, I don't think there is any direct command for that.

Suppose the present matrix position is i,j, matrix represent the given matrix.

%Here i,j represents the rows and colm value of the matrix
el_1=matrix(i,j) % Give the value of present element having position i,j
el_2=matrix(i,j-1) % Left element
el_3=matrix(i,j+1) % Right element
el_4=matrix(i-1,j) % for upper element 
el_5=matrix(i+1,j)%Upper element
el_6=matrix(i-1,j-1) % West-North Diagonal Element....
.........

You can do for the rest from the following figure

Maybe indexing helps to reduce the lines, hope some expert will show the path in that direction.

[AelinAG]

What if my (x,y) is in a corner, how do I prevent Matlab telling me indices don't exist?

[Image Analyst]

blockproc() won't do that.

[Kishan Patel]

var = bad answer

[Image Analyst]

@Kishan Patel, what does this mean? What or who is "var"? No one named "var" gave an answer, "bad" or otherwise.

[Ethan]

How do you get i and j to represent a point on the matrix in the first place?

[Image Analyst]

@Ethan you assign them, like

i = 2;
j = 4;
value = matrix(i, j)

To learn other fundamental concepts, invest 2 hours of your time here:

MATLAB Academy - Free 2 hour training

[Ethan]

@Image Analyst thankyou, I'll check that out

Answer 3

[Andrei Bobrov]

Let a - your double array.

m = size(a)+2;
b = nan(m);
b(2:end-1,2:end-1) = a;
iii = reshape(1:numel(b),m);
ii = reshape(iii(~isnan(b)),1,1,[]);
out = reshape(b(ones(3).*ii + (-m(1):m(1):m(1)) + (-1:1)'),9,[]);

[Image Analyst]

So if

a =
    11    81    21    26    63    52    45
    28    44    95    15    50    51    10
    30    56    17    66    13    72    45
    72     6    29    60    86    67    37
    32    44    90    30    44     2    45
    26    34    39    35    13    73    35

Then the first output should be [81,44,28] (the values of the 3 neighbors) and the second should be (if you move over one) [21,95,44,28,11] (the values of the 5 neighbors). Your code gives

out =
  Columns 1 through 20
   NaN   NaN   NaN   NaN   NaN   NaN   NaN    11    28    30    72    32   NaN    81    44    56     6    44   NaN    21
   NaN   NaN   NaN   NaN   NaN   NaN    11    28    30    72    32    26    81    44    56     6    44    34    21    95
   NaN   NaN   NaN   NaN   NaN   NaN    28    30    72    32    26   NaN    44    56     6    44    34   NaN    95    17
   NaN    11    28    30    72    32   NaN    81    44    56     6    44   NaN    21    95    17    29    90   NaN    26
    11    28    30    72    32    26    81    44    56     6    44    34    21    95    17    29    90    39    26    15
    28    30    72    32    26   NaN    44    56     6    44    34   NaN    95    17    29    90    39   NaN    15    66
   NaN    81    44    56     6    44   NaN    21    95    17    29    90   NaN    26    15    66    60    30   NaN    63
    81    44    56     6    44    34    21    95    17    29    90    39    26    15    66    60    30    35    63    50
    44    56     6    44    34   NaN    95    17    29    90    39   NaN    15    66    60    30    35   NaN    50    13
  Columns 21 through 40
    95    17    29    90   NaN    26    15    66    60    30   NaN    63    50    13    86    44   NaN    52    51    72
    17    29    90    39    26    15    66    60    30    35    63    50    13    86    44    13    52    51    72    67
    29    90    39   NaN    15    66    60    30    35   NaN    50    13    86    44    13   NaN    51    72    67     2
    15    66    60    30   NaN    63    50    13    86    44   NaN    52    51    72    67     2   NaN    45    10    45
    66    60    30    35    63    50    13    86    44    13    52    51    72    67     2    73    45    10    45    37
    60    30    35   NaN    50    13    86    44    13   NaN    51    72    67     2    73   NaN    10    45    37    45
    50    13    86    44   NaN    52    51    72    67     2   NaN    45    10    45    37    45   NaN   NaN   NaN   NaN
    13    86    44    13    52    51    72    67     2    73    45    10    45    37    45    35   NaN   NaN   NaN   NaN
    86    44    13   NaN    51    72    67     2    73   NaN    10    45    37    45    35   NaN   NaN   NaN   NaN   NaN
  Columns 41 through 42
    67     2
     2    73
    73   NaN
    37    45
    45    35
    35   NaN
   NaN   NaN
   NaN   NaN
   NaN   NaN

Not sure I understand how to interpret that.

[Andrei Bobrov]

interpretation of my result out:

We consider the matrix b-matrix a with boundary from nan elements.

The number of columns of the matrix out is equal to the number of elements of the matrix a.

Each column of the matrix out is a reshape of small region (3 by 3) of the matrix b.

b(1:3,1:3) -> out(:,1)

   NaN   NaN   NaN              NaN
   NaN    11    81              NaN
   NaN    28    44     ->       NaN
                                NaN
                                11
                                28
                                NaN
                                81
                                44

b(2:4,1:3) -> out(:,2)

   NaN    11    81              NaN
   NaN    28    44              NaN
   NaN    30    56    ->        NaN
                                11
                                28
                                30
                                81
                                44
                                56

and so on.