用2023a版matlab写一段将四副256*256的灰度图像I1、I2、I3、I4进行图像归一化处理，图像归一化的原理是：首先计算并确定原始图像对各种仿射变换具有较强不变性的图像矩，然后根据得到的图像矩确定变换函数的具体参数，最后利用变换函数将原始图像转换为具有标准形式的图像，其归一化过程主要包括4个步骤：平移归一化、X-shear归一化、Y-shear归一化和缩放归一化

首先，需要定义一个函数来计算图像矩。代码如下：

function [M, N] = image_moments(I)
% 计算图像矩
    [rows, cols] = size(I);
    x = repmat(1:cols, rows, 1);
    y = repmat((1:rows)', 1, cols);
    M = sum(sum(double(I).*x));
    N = sum(sum(double(I).*y));
end

接下来，定义一个函数来进行图像归一化处理。代码如下：

function [J1, J2, J3, J4] = normalize_images(I1, I2, I3, I4)
% 图像归一化处理
    % 平移归一化
    [M1, N1] = image_moments(I1);
    [M2, N2] = image_moments(I2);
    [M3, N3] = image_moments(I3);
    [M4, N4] = image_moments(I4);
    T1 = [1 0 -M1; 0 1 -N1; 0 0 1];
    T2 = [1 0 -M2; 0 1 -N2; 0 0 1];
    T3 = [1 0 -M3; 0 1 -N3; 0 0 1];
    T4 = [1 0 -M4; 0 1 -N4; 0 0 1];
    I1 = imwarp(I1, affine2d(T1));
    I2 = imwarp(I2, affine2d(T2));
    I3 = imwarp(I3, affine2d(T3));
    I4 = imwarp(I4, affine2d(T4));
    
    % X-shear归一化
    [M1, N1] = image_moments(I1);
    [M2, N2] = image_moments(I2);
    [M3, N3] = image_moments(I3);
    [M4, N4] = image_moments(I4);
    a1 = 0;
    a2 = 0;
    a3 = -(N3-N1)/(M3-M1);
    a4 = -(N4-N2)/(M4-M2);
    T1 = [1 a1 0; 0 1 0; 0 0 1];
    T2 = [1 a2 0; 0 1 0; 0 0 1];
    T3 = [1 a3 0; 0 1 0; 0 0 1];
    T4 = [1 a4 0; 0 1 0; 0 0 1];
    I1 = imwarp(I1, affine2d(T1));
    I2 = imwarp(I2, affine2d(T2));
    I3 = imwarp(I3, affine2d(T3));
    I4 = imwarp(I4, affine2d(T4));
    
    % Y-shear归一化
    [M1, N1] = image_moments(I1);
    [M2, N2] = image_moments(I2);
    [M3, N3] = image_moments(I3);
    [M4, N4] = image_moments(I4);
    b1 = 0;
    b2 = -(M2-M1)/(N2-N1);
    b3 = 0;
    b4 = -(M4-M3)/(N4-N3);
    T1 = [1 0 0; b1 1 0; 0 0 1];
    T2 = [1 0 0; b2 1 0; 0 0 1];
    T3 = [1 0 0; b3 1 0; 0 0 1];
    T4 = [1 0 0; b4 1 0; 0 0 1];
    I1 = imwarp(I1, affine2d(T1));
    I2 = imwarp(I2, affine2d(T2));
    I3 = imwarp(I3, affine2d(T3));
    I4 = imwarp(I4, affine2d(T4));
    
    % 缩放归一化
    [M1, N1] = image_moments(I1);
    [M2, N2] = image_moments(I2);
    [M3, N3] = image_moments(I3);
    [M4, N4] = image_moments(I4);
    s1 = sqrt((M1^2+N1^2)/2);
    s2 = sqrt((M2^2+N2^2)/2);
    s3 = sqrt((M3^2+N3^2)/2);
    s4 = sqrt((M4^2+N4^2)/2);
    T1 = [1 0 0; 0 1 0; -M1 -N1 1]/s1;
    T2 = [1 0 0; 0 1 0; -M2 -N2 1]/s2;
    T3 = [1 0 0; 0 1 0; -M3 -N3 1]/s3;
    T4 = [1 0 0; 0 1 0; -M4 -N4 1]/s4;
    J1 = imwarp(I1, affine2d(T1));
    J2 = imwarp(I2, affine2d(T2));
    J3 = imwarp(I3, affine2d(T3));
    J4 = imwarp(I4, affine2d(T4));
end

最后，我们可以调用该函数来进行图像归一化处理：

I1 = imread('image1.jpg');
I2 = imread('image2.jpg');
I3 = imread('image3.jpg');
I4 = imread('image4.jpg');
[J1, J2, J3, J4] = normalize_images(I1, I2, I3, I4);