[block coding] simulation of image block coding based on MATLAB

1. Software version

matlab2013b

2. System Overview

      Block coding is also called block truncation coding (block truncation coding, abbreviation BTC) Is a kind of still image coding , The principle is to divide an image into sizes N×N The child is like a block ( Short for sub block ), Because the adjacent pixels in the small block have the correlation of similar brightness , Therefore, only two appropriate brightness values are selected to approximately represent the original brightness of each pixel in the small block , Then it indicates which brightness each pixel in the sub block belongs to .

        Still image coding is to study how to make use of the inherent statistical characteristics of images ( Source characteristics ), And the physiology of vision 、 Psychological characteristics ( Sink characteristics ), Or recording equipment ( Such as ordinary photographic negative ) And display devices ( Like TV 、 Monitor ) The characteristics of etc , After compression coding, effective information is extracted from the original image information , Try to get rid of the useless redundant information , In order to efficiently carry out digital transmission or digital storage of images ; However, the restored image similar to the original image can still be obtained during restoration , That is, keep the effective information in the image information .

3. Part of the source code

function code(tp,I)


fd=fopen('btc.bin','w+');
[m,n]=size(I);
fwrite(fd,m,'long');fwrite(fd,n,'long');
I=double(I);

A=zeros(tp);B=zeros(tp);c(1)=0;
for i=1:tp:m
    for j=1:tp:n
        A=I(i:i+tp-1,j:j+tp-1);
        sum_col_A=sum(A,1);   % Subblocks are summed by columns 
        sum_A=sum(sum_col_A,2);   % Sum of subblock matrices 
        mean_A=mean2(A);    % The mean value of the sub block 
        for L=1:tp
            for k=1:tp
                if A(L,k)>=mean_A
                   B(L,k)=1;
                else
                   B(L,k)=0;
                end
            end
        end
        nozeronum=nnz(B);  % Non in image matrix 0 Number of elements 
        AmulB=times(A,B);
        sum_AmulB=sum(sum(AmulB(:,:),1),2);
        if nozeronum==0
            c(1)=0;
            c(2)=0;
        else
            c(1)=round(sum_AmulB/nozeronum);  % Reconstruction level in sub block matrix 
            if nozeronum==tp*tp
                c(2)=c(1);
            else
                c(2)=round((sum_A-sum_AmulB)/(tp*tp-nozeronum));
            end
        end
        fwrite(fd,B,'ubit1');
        fwrite(fd,c,'ubit8');
    end
end
fclose(fd);
fd=fopen('btc.bin','r');
m=fread(fd,1,'long');n=fread(fd,1,'long');
A=zeros(m,n);
for i=1:tp:m
    for j=1:tp:n
        B=fread(fd,[tp,tp],'ubit1');
        c(1)=fread(fd,1,'ubit8');
        c(2)=fread(fd,1,'ubit8');
        for k=1:tp
            for L=1:tp
                if B(k,L)==0
                    A(i+(k-1),j+(L-1))=c(2);
                else
                    A(i+(k-1),j+(L-1))=c(1);
                end
            end
        end
    end
end
A=uint8(A);
imshow(A);title(' After coding with blocks ')
imwrite(A,'lenabianma.bmp');
fclose(fd);

                

4. Simulation results

2*2

4*4

8*8

A09-30