ピクセルごとに非線形​最小二乗法を行い、そ​の画像を出力する方法​についてご教示くださ​い

各画像のピクセルごとに最小二乗法を行い、最終的にその最適化された値を画像として読み込みたいです。モデル関数は既知で、tとyも既知です。64*64などのmatrixになると関数をどのように設定すればいいか分からず、また、エラーが起こりピクセルごとの計算がうまくいきません。

-------------------------------------------------------------------------------------------------------------------- x1 = double(imread('1.png')); x2 = double(imread('2.png')); x3 = double(imread('3.png'));

t = [zeros(size(x1)) repmat(20,size(x1)) repmat(30,size(x1))];

y = [ones(size(x1)) x2./x1 x3./x1];

f = (exp(-x(1)*t) + exp(-(x(1)+x(2))*t)) - y;

%モデル関数とした場合に、

プログラムで利用できる最大変数サイズを超えました。

エラー lsqncommon (line 139) Jstr = sparse(ones(Jrows,Jcols));

エラー lsqnonlin (line 237) [xCurrent,Resnorm,FVAL,EXITFLAG,OUTPUT,LAMBDA,JACOB] = ...

エラー lm (line 31) [x,resnorm,residual,exitflag,output] = lsqnonlin(@myfun1,x0); % Invoke optimizer

------------------------------------------------------------------------------------------------------------------ 「各ピクセルに対するlsqnonlin関数の使い方」と「各ピクセルごとに連続的に計算を行う」という点が分かっていません。何卒ご教示ください。

[Steven Lord]

After translation via Google Translate and code formatting:

To perform the least square method for each pixel of each image, you want to read finally its optimized value as an image. Model function is known, t and y are also known. When it comes to the matrix, such as 64 * 64 do not know how should I how to set the function, also, does not work is the calculation of each error occurs pixels.

-------------------------------------------------- -------------------------------------------------- ----------------

 x1 = double (imread ('1.png'));
 x2 = double (imread ('2.png'));
 x3 = double (imread ( '3.png'));

 t = [zeros (size (x1)) repmat (20, size (x1)) repmat (30, size (x1))];

 y = [.. ones (size (x1)) x2 / x1 x3 / x1];

 f = (exp (-x (1) * t) + exp (- (x (1) + x (2)) * t)) - y;

% In the case of the model function,

 You have exceeded the maximum variable size that are available in the program.

 Error lsqncommon (line 139)
 Jstr = sparse (ones (Jrows, Jcols));

 Error lsqnonlin (line 237)
 [xCurrent, Resnorm, FVAL, EXITFLAG, OUTPUT, LAMBDA, JACOB] = ...

 Error lm (line 31)
 [x, resnorm, residual, exitflag, output] = lsqnonlin (@ myfun1, x0);% Invoke optimizer

-------------------------------------------------- -------------------------------------------------- --------------

do not know is that the "use of lsqnonlin function for each pixel," "make a continuously calculated for each pixel." Please be taught kindly.

[Tohru Kikawada]

Thank you for translating, Steven.

[Tohru Kikawada]

エラーメッセージが出るプログラムの全体を見せていただけないでしょうか。

Could you let me know the whole code that emits the error.