MATLAB 实现 256x256 矩阵窗口方差加权

以下是 MATLAB 代码实现：

% 生成两个 256*256 的矩阵 A 和 B
A = randi([0 255], 256, 256);
B = randi([0 255], 256, 256);

% 定义窗口大小和加权系数
window_size = 3;
weight = 0.6;

% 计算窗口方差加权
for i = 1:256
    for j = 1:256
        % 窗口左上角和右下角的坐标
        row_start = max(1, i - floor(window_size/2));
        row_end = min(256, i + floor(window_size/2));
        col_start = max(1, j - floor(window_size/2));
        col_end = min(256, j + floor(window_size/2));
        
        % 提取窗口内的数据
        window_A = A(row_start:row_end, col_start:col_end);
        window_B = B(row_start:row_end, col_start:col_end);
        
        % 计算窗口方差和加权
        variance_A = var(window_A(:));
        variance_B = var(window_B(:));
        weight_A = weight^(variance_A / (variance_A + variance_B));
        weight_B = 1 - weight_A;
        
        % 更新矩阵 A 和 B
        A(i,j) = weight_A * A(i,j) + weight_B * B(i,j);
        B(i,j) = weight_B * B(i,j) + weight_A * A(i,j);
    end
end

这里使用了双向加权，即对于每个像素，都计算它在 A 矩阵和 B 矩阵中的方差，并根据方差大小计算加权系数。最终更新 A 和 B 矩阵时使用了加权平均。