Matlab low-level source code realizes the median filtering of the image (used to eliminate some miscellaneous points on the image)

Theoretical basis

       An image is often disturbed by various noises , The noise is often some isolated pixels , It is often random noise superimposed on the image , Make the image polluted like snowflakes , The gray level of the image should be relatively continuous , Generally, it will not suddenly become larger or smaller , This kind of noise can be suppressed by neighborhood .
      N.N Median filtering algorithm , When gray image f China and Israel f(i,j) Centred N.N Screen the window （N It's odd ） The inner median gray scale is a when , Unconditionally will f(i,j)=a Handle ,N Given by the user .

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%N*N median filtering 
%Author:Zhu
% Time ：2022.3
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
clc;
image = imread("E:\\1.jpg");
image_matrix=image(:,:,1);
image_matrix=double(image_matrix);
[height,width,channels]=size(image);

% Enter the value , Set the window size （ The size must be odd ）
maskWidth = 5;
maskHeight = 5;

% It's usually an odd number 
maskCenterX=(maskWidth+1)/2;
maskCenterY=(maskHeight+1)/2;

% Find the median size in the array 
maskValue=zeros(1,maskHeight*maskWidth);

% Sequence number in the middle of the array 
middleNum = maskHeight*maskWidth/2+0.5;

for i=maskCenterY:1:height-maskHeight+maskCenterY
  for j=maskCenterX:1:width-maskWidth+maskCenterX
      % Calculate the pixel value 
      pixelResult = 0;
      % Traverse the template area 
      for k=1:1:maskHeight
          for p = 1:1:maskWidth
              % Put the value into the array 
              maskValue((k-1)*maskHeight+p) = image_matrix(i-(maskHeight+1)/2+k,j-(maskWidth+1)/2+p);  
          end
      end
      
      % Find the value in the array , Bubble sorting method , Arrange the values from large to small 
      for num = 1:maskHeight*maskWidth
          for xx = 1:maskHeight*maskWidth-1
              if(maskValue(xx)<maskValue(xx+1))
                  temp = maskValue(xx);
                  maskValue(xx) = maskValue(xx+1);
                  maskValue(xx+1) = temp;
              end
          end
      end 
      % Update the median to give the current pixel 
      image_matrix(i,j)= maskValue(middleNum);
  end
end

image_out = uint8(image_matrix);
% Show 
subplot(1,2,1);
imshow(image);
subplot(1,2,2);
imshow(image_out);

Realization effect