How can i input each pixel value intensity information into an array?

I am trying to do my final year project and am not so great in MATLAB please help, How can i input each pixel value intensity information into an array? What mistake have i made on the code below? 
Also is there a way to find the difference of intensity between a chosen pixel and its neighboring pixels - this information would be used to help me find an edge around a missing pixel and know which direction to interpolate in as not to get a fuzzy result.

function intensity()

lena=imread('lena.bmp');
size(lena)
[m n p]=size(lena)
a=rgb2gray(lena);
imshow(a);
title('Original Grayscale Image');

k=zeros(m,n);

for i=1:512;
    for j=1:512;
        intensityValue{i}=k(m,n);
    end
   intensityValue{j}=k(m,n);
end

figure, imhist(a)

Answer 1

[Image Analyst]

You already have the intensity. It's lena or a.

To find the difference between a chosen pixel and it's neighbors, you can use imfilter or conv2():

kernel = [-1 -1 -1;-1 8 -1; -1 -1 -1]/8;
differenceImage = conv2(double(a), kernel, 'same');

[SSZ]

I am quite lost at the moment i had a supervisor who was working on this project but he left unexpected and now im left without any support or someone who really understands what im trying to do and that i can ask questions to so these may seem very trivial:

The kernal is that similar to sobel?

The code i have done below the main thing i want to achieve is for 6ax64 bit image with the missing pixels i want to use the conv2 you mentioned above to fix the smallest difference in intensity surrounding the missing pixels and have this smallest difference to know which direction to interpolate. How would i word the for for loop to do this?

Thanks!

    function FYPDCT_Image()

    lena=imread('lena.bmp');
    size(lena)
    [m n p]=size(lena)
    a=rgb2gray(lena);
    imshow(a);
    title('Original Grayscale Image');

    lena_c=zeros(m,n,p);

    lenad=a;

    for i=1:16:512
        for j=1:16:512
            lenad(i:i+7,j:j+7)=0;
        end
    end

    %%DCT

    subplot(2,2,1);
    imshow(lenad)
    [m n] = size(lenad)
    title([num2str(m) ' X ' num2str(n)]);

    subplot(2,2,2);
    lenad_ds1=DS_DCT(lenad,8);
    imshow(uint8(lenad_ds1));    
    [m n] = size(lenad_ds1)
    title([num2str(m) ' X ' num2str(n)]);

    subplot(2,2,3);
    lenad_ds2=DS_DCT(lenad_ds1,4);

    imshow(uint8(255*lenad_ds2/max(max(lenad_ds2))));      
    [m n] = size(lenad_ds2)
    title([num2str(m) ' X ' num2str(n)]);

    subplot(2,2,4);
    lenad_ds3=DS_DCT(lenad_ds2,2);

    imshow(uint8(255*lenad_ds3/max(max(lenad_ds3)))); 
    [m n] = size(lenad_ds3)
    title([num2str(m) ' X ' num2str(n)]);

    kernel = [-1 -1 -1;-1 8 -1; -1 -1 -1]/8;

    differenceImage = conv2(double(lenad_ds3), kernel, 'same');

    differenceImage

    %%Downsample by a factor of square(2) using DCT
    function ImageOut=DS_DCT(ImageIn,MBsize)

    [m n]=size(ImageIn);
    k=1;
    for i=1:MBsize:m-MBsize+1
        l=1;
        for j=1:MBsize:m-MBsize+1
            temp=ImageIn(i:i+MBsize-1,j:j+MBsize-1);
            tempDCT=dct2(temp);
            temp=idct2(tempDCT(1:MBsize/2,1:MBsize/2));
            ImageOut(k:k+MBsize/2-1,l:l+MBsize/2-1)=temp;
            l=l+MBsize/2;
        end
        k=k+MBsize/2;
    end

[Image Analyst]

I don't know what a "6ax64 bit image" is. Is the a really a 4? Is it a 64 row by 64 column image of 8 bit integers? Is it a 64 bit integer image (which MATLAB can't handle)? Can you attach it?

[SSZ]

Sorry mistype yes a 64 row x 64 column image. I have attached an image i am using a pyramid structure to down sample a 512x512 image to 64x64 to get a loss gap of 1 pixel.