i need help to compute the algorithe of local AR-2D image by using blocks

hello everyone I want to divide an image into 8 * 8 and for each block I have to do the following steps:

calculate the least squares estimators of the AR-2D model parameter described by:

X (i, j) = teta (1.0) * X (i-1, j) + teta (0.1) * X (i, j-1) + e (i, j)

  after the calculation of teta with the method of least square I have to generate the image with this equation:

X X (i, j) = teta (1.0) * X (i-1, j) + teta (0.1) * X (i, j-1) + e (i, j)

this is my programme

clear all

clc

i=imread('D1.gif');

i= im2double(i);

x = imresize( i,0.25);

[m,n] = size(x);

noise=randn(size(x));

X = mat2cell(x, 8 * ones(1,m/8), 8 * ones(1,n/8));

X=[x(1:40,1:40)]

R = size(X, 1);

C = size(X, 2);

for i=2:R

for j=2:C

Z(i,j)=X(i-1,j)+X(i,j-1);

Z=Z+1;

end

end

teta=inv(z'*z)*z'*X

i have this message

Inner matrix dimensions must agree.

Error in Untitled2 (line 85)

teta=inv(z'*z)*z'*X

who can help me pleas to solve the probleme

Answer 1

[Image Analyst]

See THE FAQ for processing blocks of the image.

The equation

teta=inv(z'*z)*z'*X

is the well known least squares solution, so simply use polyfit() instead of doing it yourself manually.

[ayoub maa]

yes it is least squares solution but how can i use polyfit() for this equation Z(i,j)=X(i-1,j)+X(i,j-1);