帮我写一段NL-means去噪MATLAB算法

好的，下面是一个简单的NL-means去噪MATLAB算法：

function [denoisedImg] = nlMeansDenoising(noisyImg, patchSize, windowSize, h)
% Input: 
%       noisyImg - 带有噪声的图像
%       patchSize - 每个块的大小
%       windowSize - 用于搜索相似块的窗口大小
%       h - 控制去噪强度的参数
% Output: 
%       denoisedImg - 去噪后的图像

% 转换为灰度图像
if size(noisyImg, 3) == 3
    noisyImg = rgb2gray(noisyImg);
end

% 复制一份噪声图像作为去噪后的图像的初始值
denoisedImg = double(noisyImg);

% 计算图像中每个像素周围的块
halfPatchSize = floor(patchSize/2);
halfWindowSize = floor(windowSize/2);
[rows, cols] = size(noisyImg);
paddedImg = padarray(noisyImg, [halfWindowSize halfWindowSize], 'symmetric');
for i = 1+halfWindowSize:rows+halfWindowSize
    for j = 1+halfWindowSize:cols+halfWindowSize
        % 获取以(i,j)为中心的块
        i1 = i - halfPatchSize;
        i2 = i + halfPatchSize;
        j1 = j - halfPatchSize;
        j2 = j + halfPatchSize;
        patch = paddedImg(i1:i2, j1:j2);
        
        % 在以(i,j)为中心的窗口内搜索相似块
        w1 = i - halfWindowSize;
        w2 = i + halfWindowSize;
        h1 = j - halfWindowSize;
        h2 = j + halfWindowSize;
        searchWindow = paddedImg(w1:w2, h1:h2);
        [searchRows, searchCols] = size(searchWindow);
        minMSE = Inf;
        for m = 1:searchRows-halfPatchSize*2
            for n = 1:searchCols-halfPatchSize*2
                % 获取搜索窗口中的相似块
                m1 = m + halfWindowSize;
                m2 = m + halfWindowSize + patchSize - 1;
                n1 = n + halfWindowSize;
                n2 = n + halfWindowSize + patchSize - 1;
                similarPatch = searchWindow(m1:m2, n1:n2);
                
                % 计算均方误差
                mse = sum(sum((patch-similarPatch).^2))/(patchSize^2);
                
                % 记录最小的均方误差和相似块的位置
                if mse < minMSE
                    minMSE = mse;
                    minM = m;
                    minN = n;
                end
            end
        end
        
        % 计算权重并进行去噪
        weightSum = 0;
        weightedPatchSum = 0;
        for m = minM:minM+patchSize-1
            for n = minN:minN+patchSize-1
                % 获取搜索窗口中的相似块
                m1 = m + halfWindowSize;
                m2 = m + halfWindowSize + patchSize - 1;
                n1 = n + halfWindowSize;
                n2 = n + halfWindowSize + patchSize - 1;
                similarPatch = searchWindow(m1:m2, n1:n2);
                
                % 计算相似块与目标块的权重
                weight = exp(-minMSE/(h^2));
                weightedPatchSum = weightedPatchSum + weight*similarPatch;
                weightSum = weightSum + weight;
            end
        end
        denoisedImg(i-halfWindowSize, j-halfWindowSize) = weightedPatchSum/weightSum;
    end
end

% 转换图像类型为uint8
denoisedImg = uint8(denoisedImg);
end

其中，`noisyImg`为带有噪声的输入图像，`patchSize`为每个块的大小，`windowSize`为用于搜索相似块的窗口大小，`h`为控制去噪强度的参数。函数返回去噪后的图像`denoisedImg`。