自适应中值滤波器matlab

在MATLAB中实现自适应中值滤波器可以使用以下代码：

function output = adaptive_median_filter(input, max_window_size)
    % Initialize output and window size
    output = input;
    window_size = 3;
    
    % Loop through every pixel in the image
    for i = 2:size(input,1)-1
        for j = 2:size(input,2)-1
            % Extract window around current pixel
            window = input(i-window_size:i+window_size, j-window_size:j+window_size);
            
            % Compute median and minimum and maximum values in the window
            median_value = median(window(:));
            min_value = min(window(:));
            max_value = max(window(:));
            
            % Check if the current pixel is an impulse noise
            if (input(i,j) == min_value || input(i,j) == max_value)
                window_size = window_size + 2; % Increase window size
                % Check if the new window size exceeds the maximum specified size
                if (window_size > max_window_size)
                    output(i,j) = median_value; % Use median value
                else
                    % Extract new window
                    window = input(i-window_size:i+window_size, j-window_size:j+window_size);
                    % Compute new median and minimum and maximum values in the window
                    median_value = median(window(:));
                    min_value = min(window(:));
                    max_value = max(window(:));
                    % Check if the current pixel is still an impulse noise
                    if (input(i,j) == min_value || input(i,j) == max_value)
                        output(i,j) = median_value; % Use median value
                    else
                        output(i,j) = input(i,j); % Use original value
                    end
                end
            else
                output(i,j) = input(i,j); % Use original value
            end
        end
    end
end

该函数的输入参数包括输入图像和最大窗口大小，输出为经过自适应中值滤波器处理后的图像。在函数内部，首先初始化输出图像和窗口大小，然后循环遍历输入图像中的每个像素。对于每个像素，提取以其为中心的窗口，计算窗口中的中值、最小值和最大值，并检查当前像素是否为脉冲噪声。如果是，则增加窗口大小并重新提取窗口，直到窗口大小达到最大值为止。如果最终窗口中仍然存在最小值或最大值等脉冲噪声，则将该像素替换为窗口中的中值。如果当前像素不是脉冲噪声，则将其保留为原始值。

使用方法：

input = imread('input.jpg');
output = adaptive_median_filter(input, 7);
imshow(output);

其中，input为输入图像，7为最大窗口大小。可以根据需要修改最大窗口大小。