Reconstructing the image from specific number of coefficicents

Hi there!

Alright, so I've been working on this little piece of code to construct an image using specific coefficicents and there's this error I'm getting:

Subscripted assignment dimension mismatch.

So the code goes this way:

j0=128;
M = 2^(-j0);
m=sqrt(M);
I = imread('lena.png');
J = perform_wavelet_transf(I,j0,+1);
n = size(I,1); 
J1 = zeros(n);
J1(1:n/4,1:n/4) = J(1:m,1:m);
result = perform_wavelet_transf(J1,j0,-1);
imshow(result,[]);

I'm getting the error at this line: J1(1:n/4,1:n/4) = J(1:m,1:m);

What do you think?

And yes, Thanks alot!!! for your help in advance :)

[Amit]

attach the image please

[Amit]

I meant, can you upload lena.png here. That way I can test it.

Answer 1

[Amit]

One thing I see first is that M = 2^(-128) = 2.93e-39 and thus m = 5.42e-20 m is not an integer, thus J(1:m,1:m) would not work and give you atleast one error.

[Ayesha]

I tried setting j0 to zero but its still not giving the proper result.

[Amit]

To make this work, m should be equal to n/4.

when you are J1(1:n/4,1:n/4) = J(1:m,1:m), you are copying an mxm matrix to a n/4xn/4 matrix. That would be only possible if m = n/4.

[Ayesha]

you mean the size because I've been wondering if there's any point of equalizing both the terms and then copy one's matrix onto the other. What do you think?

[Amit]

I dont have the image lena on my matlab.

The problem is when you define j0 = 0, m will become 1. That means J(1:m,1:m) is an scalar.

Given n/4 is an positive integer (you need to make sure that), J1(1:n/4,1:n/4) = J(1:m,1:m); will store J(1:m,1:m) scalar in all n/4xn/4 matrix in J1. I am quite certain this is not what you're looking for. I think somewhere in your algorithm (code), something is missing.

[Amit]

Also attach the perform_wavelet_transf function. This is not a inbuilt function in matlab.

[Amit]

perform_wavelet_trasf has many dependencies which you did not posted.

I am sorry, this is just frustrating.

[Amit]

its alright. In a nut shell, I think there is a problem at M = 2^(-j0) which you use to define m = sqrt(M).

Please double check the algorithm that you must be following that this is the correct way to do this.

[Ayesha]

I've been doing the same since long. Anyways, Thanks.