partioning the image

I want to divide the image into 6 blocks ,i used the code below,now i want to display those divided blocks,can  tell how to display it please,I am performing CLD,PLEASE tell hoe to calculate the centroid from "out"

img=imread('dock.jpg')
[img_x,img_y]=size(img);

block_size=8;
slide_len=1;

for ix=block_size/2:slide_len:img_x-block_size/2
    for jy=block_size/2:slide_len:img_y-block_size/2
        current_block=img((ix-block_size/2+1):(ix+block_size/2),(jy-block_size/2+1):(jy+block_size/2));
        dct_coeff=reshape(dct2(current_block),1,block_size^2);
        out=zigzag(dct_coeff)

      end
  end

Answer 1

[Walter Roberson]

Your code is overwriting your "out" matrix in each iteration, so at the end of the loop it does not have the necessary information to display all of the blocks.

You have not told us enough about zigzag() for us to predict what operations could be performed on "out".

[Pat]

ActuallY i want to perform CLD,on each frame,i have 12 frames ,and i want to calcualte centroid for each frame

[Walter Roberson]

Please clarify how "frame" relates to your above description. Is a "frame" an individual image?

You did not tell us about zigzag()

It is not clear to me exactly what it is that the centroid is being taken of, but perhaps that would make more sense if I knew about zigzag().

[Pat]

walter in my previous post u only told to use zigzag for performing CLD..yes walter my frame is an idividual image,

is else let me assume this way

i want to perform CLD FOR 12 IMAGES and calculate its centroid

[Walter Roberson]

zigzag() is your _implementation_ of an idea; we can see what you are passing in as input, but not what you do with it.

If you are doing CLD then why are you not dividing your image in to 64 blocks (instead of 6), and why are you not applying your algorithm to a color image (your size() call assumes 2D which can only happen with grayscale for .jpg images), and where are the other two DCT matrices and zigzag outputs that would be built for CLD stage 4 and 5? http://en.wikipedia.org/wiki/Color_layout_descriptor

[Pat]

Walter please dont get angry,in my previous post i asked how to partition into 64 blocks and color representation ,u gave an link to refer,that coding was that,if i am wrong can u please tell how to partition image into 64 blocks and do so

[Walter Roberson]

In your Question, above, you wrote, "I want to divide the image into 6 blocks". Did you miss out the "4" of "64" ?

JPEG files can only store grayscale images (2D) or true-color images (3D) and JPEG cannot store indexed images (2D). If you are reading a color image, 3D, then [img_x,img_y]=size(img); is going to get the right number of rows stored in img_x, but the value stored in img_y is going to be the number of columns multiplied by the number of bands (3) . See the last syntax case in the size() documentation.

The DCT stage for CLD produces 3 DCT coefficient matrices; you appear to only be producing one.

[Pat]

sorry walter for missing 4,and per wikipedia u sent ,i need output image as those,please guide

[Walter Roberson]

Looks to me as if you would have one DCT coefficient matrix per color plane. But first you are going to have to find an image that is simultaneously 3D (truecolor) and not 3D (in order to work with your calls to size())

[Pat]

Walter the size of my image is 256x256x3

[Walter Roberson]

Can't be. Your line

[img_x,img_y]=size(img);

is only appropriate for 2D images.

[Pat]

walter when i give

A=size(img),the output is 256x256x3

[img_x,img_y]=size(img);

i get 256x1450

[Walter Roberson]

And if it is not appropriate for your code to think that img_y is 1450 (or, more likely, 768) then you will need to change that line of code.