法对图像进行压缩，得到一组压缩编码，然后解码这组编码，得到一幅解压缩图像．对解码图像与原始图像求差值，得到一差值图像，然后对该差值图像进行适当的编码．对差值图像的编码与初级编码共同构成对原始图像的编码．这种方法需要选择合适的初级编码方法与差值编码方法，使得这两者相结合，可以得到一种综合性能较好的编码方法。
其定义如图。
%%%%%%%%%%%%%%
clear
tic
%Image1=imread('pic\cameraman.tif');
xianshi;
number=input(' input the number:');
Image1=suoxiao('pic\cameraman.tif',number);
[imagem imagen]=size(Image1);
Sr=4;Sd=8;
Rnum=(imagem/Sr)*(imagen/Sr);
Dnum=(imagem/Sd)*(imagen/Sd);
Image2=zeros(Dnum,Sr,Sr);
Image2=blkproc(Image1,[Sd/Sr,Sd/Sr],'mean(mean(x))');
%压缩image1为原来1/2
% there are no eight tranformation for simpleness
RBlocks=zeros(Rnum,Sr,Sr);
DBlocks=zeros(Dnum,Sd,Sd);
DBlocksReduce=zeros(Dnum*8,Sr,Sr);
%%取R块，K记标号----------------------------------
for i=1:imagem/Sr
for j=1:imagen/Sr
k=(i-1)*imagen/Sr+j;
RBlocks(k,:,:)=Image1((i-1)*Sr+1:i*Sr,(j-1)*Sr+1:j*Sr);
end
end
%取R块，K记标号----------------------------------
for i=1:imagem/Sd
for j=1:imagen/Sd
k=(i-1)*imagen/Sd+j;
m=Sr;n=Sr;
DBlocksReduce(k,:,:)=Image2((i-1)*Sr+1:i*Sr,(j-1)*Sr+1:j*Sr);
DBlocksReduce(k+Dnum,:,:)=DBlocksReduce(k,m:-1:1,:); % 行上下翻转===(x轴对称)
DBlocksReduce(k+2*Dnum,:,:)=DBlocksReduce(k,:,n:-1:1); % 列左右翻转 ==== y轴对称
DBlocksReduce(k+3*Dnum,:,:)=DBlocksReduce(k,m:-1:1,n:-1:1); % 先行翻，再列翻 旋转180度
DBlocksReduce(k+4*Dnum,:,:)=reshape(DBlocksReduce(k,:,:),Sr,Sr)'; % 关于y=-x对称
A=reshape( DBlocksReduce(k+3*Dnum,:,:),Sr,Sr)';
DBlocksReduce(k+5*Dnum,:,:)=A(:,n:-1:1); % 关于y=x对称
DBlocksReduce(k+6*Dnum,:,:)=imrotate(reshape(DBlocksReduce(k,:,:),Sr,Sr),90); % 逆时针旋转90度
DBlocksReduce(k+7*Dnum,:,:)=imrotate(reshape(DBlocksReduce(k,:,:),Sr,Sr),270); % 逆时针旋转270度
DBlocks(k,:,:)=Image1((i-1)*Sd+1:i*Sd,(j-1)*Sd+1:j*Sd);
end
end
RandDbest=zeros(Rnum,1)+256^3;
RandDbests=zeros(Rnum,1);
RandDbesto=zeros(Rnum,1);
RandDbestj=zeros(Rnum,1);
for i=1:Rnum
x=reshape(RBlocks(i,:,:),Sr*Sr,1);
meanx=mean(x);
for j=1:Dnum*8
y=reshape(DBlocksReduce(j,:,:),Sr*Sr,1);
meany=mean(y);
s=(x-meanx)'*(y-meany)/((y-meany)'*(y-meany));%计算s
o=(meanx-s*meany);%计算o
c=(x-s*y-o)'*(x-s*y-o);%距离
if (RandDbest(i)>c)&(abs(s)<1)
RandDbest(i)=c;
RandDbests(i)=s;
RandDbesto(i)=o;
RandDbestj(i)=j;%可以找到对应变换和D块
end
end
end
%iteration limit
toc
tic
m=8;%解码迭代次数
e=mean(mean(Image1));
Image3=e*ones(imagem,imagen);%解码初始图象
for L=1:m
Image4=blkproc(Image3,[Sd/Sr,Sd/Sr],'mean(mean(x))');
for i=1:imagem/Sr
for j=1:imagen/Sr
m=Sr;n=Sr;
k=(i-1)*imagen/Sr+j;
l=RandDbestj(k);
k1=mod(l-1,Dnum)+1;%第几个D
l1=(l-k1)/Dnum+1;%变换号
%R对应D在Image4的起始点
j1=mod(k1-1,imagen/Sd)+1;
i1=(k1-j1)/(imagen/Sd)+1;
%变换------------------------------------------------------------------------
DBlocksReduce(k1,:,:)=Image4((i1-1)*Sr+1:i1*Sr,(j1-1)*Sr+1:j1*Sr);
switch l1-1
case 0
DBlocksReduce(l,:,:)=Image4((i1-1)*Sr+1:i1*Sr,(j1-1)*Sr+1:j1*Sr);
case 1
DBlocksReduce(l,:,:)=DBlocksReduce(k1,m:-1:1,:);
case 2
DBlocksReduce(l,:,:)=DBlocksReduce(k1,:,n:-1:1);
case 3
DBlocksReduce(l,:,:)=DBlocksReduce(k1,m:-1:1,n:-1:1);
case 4
DBlocksReduce(l,:,:)=reshape(DBlocksReduce(k1,:,:),Sr,Sr)';
case 5
DBlocksReduce(k1+3*Dnum,:,:)=DBlocksReduce(k1,m:-1:1,n:-1:1);
A=reshape( DBlocksReduce(k1+3*Dnum,:,:),Sr,Sr)';
DBlocksReduce(l,:,:)=A(:,n:-1:1);
case 6
DBlocksReduce(l,:,:)=imrotate(reshape(DBlocksReduce(k1,:,:),Sr,Sr),90);
case 7
DBlocksReduce(l,:,:)=imrotate(reshape(DBlocksReduce(k1,:,:),Sr,Sr),270);
end
%变换结束--------------------------------------------------------------------
RBlocks(k,:,:)=RandDbests(k)*DBlocksReduce(l,:,:)+RandDbesto(k);
%生成R---------------------------
Image3((i-1)*Sr+1:i*Sr,(j-1)*Sr+1:j*Sr)=reshape(RBlocks(k,:,:),Sr,Sr);%更新迭代图象
end
end
wucha=double(Image1)-Image3;%误差图
Ps1(L)=20*log10(255/(sqrt(mean(mean(wucha.^2)))))
PSNR=psnr(wucha)
figure
imshow(uint8(Image3))
end
toc
figure
wucha=uint8(wucha);
imshow(wucha)
figure
imshow(uint8(Image1)),title('原图');
save('sa.mat')
zhifangtu(wucha);%%%%分形主函数
%%%%%%%%%%%%%%%%%%%%%%%%%%%%
子函数:1：
function b=suoxiao(filename,bili)
a=imread(filename);
a=double(a);
[m,n]=size(a);
i=1;
while i=m/bili
j=1;
while j=n/bili
k=mean(mean(a(bili*(i-1)+1:bili*(i-1)+bili,bili*(j-1)+1:bili*(j-1)+bili)));
b(i,j)=k;
j=j+1;
end
i=i+1;
end
%b=uint8(b);
size(b)
%imshow(b)
子函数2：
%clc
function zhifangtu(a)
J=a;
%计算灰度图象的直方图数据，a为如象数组
L=256; %灰度级
Ps = zeros(L,1); %统计直方图结果数据
nk=zeros(L,1);
[row,col]=size(a);
n=row*col; %总像素个数
for i = 1:row
for j = 1:col
num = double(a(i,j))+1; %获取像素点灰度级
nk(num) = nk(num)+1; %统计nk
end
end
%计算直方图概率估计
for i=1:L
Ps(i)=nk(i)/n;
end
figure;
subplot(3,1,1);imshow(J),title('误差图');
subplot(3,1,2),plot(nk),title('直方图（nk）');
subplot(3,1,3),plot(Ps),title('直方图（Ps）');
子函数3：
function PSNR=psnr(a)
[m,n]=size(a);
a=uint8(a);
a=double(a);
imagesize=m*n;
MSE=sum(dot(a,a))/ imagesize;
PSNR=10*log10(255^2/MSE);
%%%%%%%%%%%%%%%%%%%%%%
说明：
1、因为本程序时间长，FX中先选择图片的大小
2、编码与解码
3、做误差图和只方图
4：画出每次迭代的解码图象
我是不会，看到楼上的哥们那么辛苦的回答， 希望能对你有用，祝福。
内容全部来源于网络收集，如有侵权，请联系网站删除：QQ：24596024