How to use blockproc by location?

Hy all, is blockproc could be used to block by location(x,y) that i give? Ex: I have image with 64x64 pixels, and i want to turn white every pixel around location point(x,y), let's say i want to change [3 3] around the point. could blockproc do it? or there any suggestion for me?

Thanks, :).

Answer 1

[Image Analyst]

Yes, you can even do that without blockproc(). Here's how

yourImage = uint8(255 * ones(yourImage));

that will turn every pixel, at every (x,y) location, white (assuming a uint8 image).

Also, see attached demos for blockproc().

[Elder Winter]

Wow, i thought the overlapping code is most easiest to understand, but i don't how to put my location in those code

[Image Analyst]

Again, using blockproc() to turn every pixel white is unnecessary. Even using it to turn one pixel, or a small set of pixel locations, white is not what blockproc() is supposed to be used for. Just use simple indexing, like

grayImage(y, x) = 255;

so don't try to put it into your code. It's better to just explain what you want to do and we'll tell you how to do it. I doubt turning every pixel white is what you want to do.

[Elder Winter]

Hmmm yeah, it's not all white, but there some gray color, it's depend the gradient color that i got from the other image. So i do as you suggest to me :

        grayImage(locs(i,1)-1,locs(i,2)-1) = T{i}(1,1);
        grayImage(locs(i,1),locs(i,2)-1) = T{i}(1,2);
        grayImage(locs(i,1)+1,locs(i,2)-1) = T{i}(1,3);
        grayImage(locs(i,1)-1,locs(i,2)) = T{i}(2,1);
        grayImage(locs(i,1),locs(i,2)) = T{i}(2,2);
        grayImage(locs(i,1)+1,locs(i,2)) = T{i}(2,3);
        grayImage(locs(i,1)-1,locs(i,2)+1) = T{i}(3,1);
        grayImage(locs(i,1),locs(i,2)+1) = T{i}(3,2);
        grayImage(locs(i,1)+1,locs(i,2)+1) = T{i}(3,3);

Variable T is the gradient color as i said before.

[Steven Lord]

There's no need to set each element individually. Address a matrix of elements on the left side of the equals sign and specify a matrix of the same size on the right.

>> A = zeros(5);
>> cent = [2, 3];
>> A(cent(1)+(-1:1), cent(2)+(-1:1)) = magic(3)

This replaces the submatrix in the first through third rows (2 + (-1:1)) and the second through fourth column (3 + (-1:1)) of A (which is a 3-by-3 submatrix of A) with the 3-by-3 magic matrix. If you don't like the negative numbers in that expression, specify the upper-left corner of the submatrix to be replaced / filled rather than the center.

>> B = zeros(5);
>> ulc = [1 2];
>> B(ulc(1)+(0:2), ulc(2)+(0:2)) = magic(3)

[Image Analyst]

Elder, I don't recall you saying anything about T, the gradient color, or locs variable before your last comment above. Where did you get locs from? Why not just do a for loop over all rows in locs?

for k = 1 : size(locs, 1)
    % Replace this 3x3 block with the values from T{k}:
    grayImage(locs(k,1)-1:locs(k,1)+1, locs(k,2)-1:locs(k,2)+1) = T{k}; 
end

Make sure locs is a list (row, column) coordinates, and not (x,y), or else you won't replace the correct locations!

Answer 2

[KALYAN ACHARJYA]

#Edited

Considering Input Image is Gray, let say image1 and location of pixel is x,y

image1(y-1:y+1,x-1:x+1)=255;

[Image Analyst]

Correction, since images are accessed as image1(row, column), not image1(x, y):

image1(y-1:y+1, x-1:x+1) = 255;

[KALYAN ACHARJYA]

Thank you @ImageAnalyst, Let say I have image size of 20x25, that menas it has 20 rows 25 colm

%

Again let say I want to make white pixel around the spatial coordinats (11,15), neighbour pixels only (3x3) ...............................................................................................^(r,c)

image(11-1:11+1,15-1:15+1) = 255;

Where I am doing wrong? Gratitude!

[Walter Roberson]

spatial coordinates (11,15) would typically be a statement of (x, y) coordinates, which would be column 11 row 15, leading to image1(15-1:15+1,11-1:11+1) = 255;

[KALYAN ACHARJYA]

Thank you @Walter @ImageAnalyst

[Elder Winter]

Ohhh, it like block whole area around the point, and turn it to white. Thank You.

Answer 3

[Walter Roberson]

blockproc is a waste for this kind of task, but it can be done. You would specify a blocksize of 3 x 3 and an overlap of [1 1], and then in the block_struct that is passed to your function, you would test for location == [y-1, x-1] and make the change in that case and otherwise return the input.

But there is just no point to this as you can go directly to that block.

YourImage(y-1:y+1, x-1:x+1, :) = 255;          %assuming uint8

If you do not want to touch the center pixel then

YourImage(y-1:y+1, [x-1 x+1], :) = 255;
YourImage([y-1 y+1], x, :) = 255;

[Elder Winter]

Hmmm, this looks same as above. but i want to change the center too, of course. Thank You