自动白平衡AWB算法 基于DTM PRA GWA的matlab代码

%Program for white balancing% %Author : Jeny Rajan %usage : W=imbalance(im) %im is the color image array to be corrected. % function W=wbalance(im) %==========================Get file============================= function W=awb_dtm(im) Rorg=im(:,:,1); Gorg=im(:,:,2); Borg=im(:,:,3); [x y z]=size(im); im2=im; im1=rgb2ycbcr(im); %Lu=im1(:,:,1); %Cb=im1(:,:,2); %Cr=im1(:,:,3); Lu= 0.299*double(Rorg) + 0.587*double(Gorg) + 0.114*double(Borg) + 16; Cb=-0.169*double(Rorg) - 0.331*double(Gorg) + 0.500*double(Borg) + 128; Cr= 0.500*double(Rorg) - 0.418*double(Gorg) - 0.082*double(Borg) + 128; %Lu= 0.183*double(Rorg) + 0.624*double(Gorg) + 0.062*double(Borg) +16; %Cb=-0.101*double(Rorg) - 0.338*double(Gorg) + 0.439*double(Borg) +128; %Cr= 0.439*double(Rorg) - 0.399*double(Gorg) - 0.040*double(Borg) +128; Lu_rsp = reshape(Lu, 1, x*y)'; R_rsp = reshape(Rorg, 1, x*y)'; G_rsp = reshape(Gorg, 1, x*y)'; B_rsp = reshape(Borg, 1, x*y)'; RGB_rsp = [R_rsp G_rsp B_rsp]; %======================Calc Mean-Valve============================ tst=zeros(x,y); Mb=sum(sum(Cb)); Mr=sum(sum(Cr)); Mb=Mb/(x*y); Mr=Mr/(x*y); %======================Calc ave asb diff=========================== Db=sum(sum(abs(Cb-Mb)))/(x*y); Dr=sum(sum(abs(Cr-Mr)))/(x*y); %=====================White Region Collect========================= b1=Cb-(Mb+Db*sign(Mb-128)); b2=Cr-(1.5*Mr+Dr*sign(Mr-128))+64; b1=abs(b1); b2=abs(b2); b1_ref=1.5*Db; b2_ref=1.5*Dr; Lu_index=find(b1<b1_ref & b2<b2_ref); Lu_index2=size(Lu_index); Ciny=Lu_rsp(Lu_index,:); cnt=Lu_index2(1); iy=sort(Ciny,'descend'); nn=round(cnt/10); Ciny2(1:nn)=iy(1:nn); mn=min(Ciny2); tst=zeros(x,y); Rtst_rshp = reshape(tst, 1, x*y)'; Gtst_rshp = reshape(tst, 1, x*y)'; Btst_rshp = reshape(tst, 1, x*y)'; Rtst_rshp(Lu_index)=RGB_rsp(Lu_index,1); Gtst_rshp(Lu_index)=RGB_rsp(Lu_index,2); Btst_rshp(Lu_index)=RGB_rsp(Lu_index,3); RGBtst_rshp =[Rtst_rshp Gtst_rshp Btst_rshp]; outR = reshape(Rtst_rshp, x, y); outG = reshape(Gtst_rshp, x, y); outB = reshape(Btst_rshp, x, y); im3(:,:,1) = uint8(outR); im3(:,:,2) = uint8(outG); im3(:,:,3) = uint8(outB); figure,imshow(im3,[]),title('Corrected Image'); Lu_index3=find(Lu_rsp>=mn); Lu_index4=size(Lu_index3 ); tst=zeros(x,y); tst_rshp = reshape(tst, 1, x*y)'; Rtst_rshp = reshape(tst, 1, x*y)'; Gtst_rshp = reshape(tst, 1, x*y)'; Btst_rshp = reshape(tst, 1, x*y)'; Rtst_rshp(Lu_index3)=RGB_rsp(Lu_index3,1); Gtst_rshp(Lu_index3)=RGB_rsp(Lu_index3,2); Btst_rshp(Lu_index3)=RGB_rsp(Lu_index3,3); tst_rshp(Lu_index3)=1; RGBtst_rshp =[Rtst_rshp Gtst_rshp Btst_rshp]; outR = reshape(Rtst_rshp, x, y); outG = reshape(Gtst_rshp, x, y); outB = reshape(Btst_rshp, x, y); im3(:,:,1) = uint8(outR); im3(:,:,2) = uint8(outG); im3(:,:,3) = uint8(outB); figure,imshow(im3,[]),title('Corrected Image'); c=Lu_index4(1); tst=reshape(tst_rshp, x, y); %======================White Point Adjust============================== R=im(:,:,1); G=im(:,:,2); B=im(:,:,3); R=double(R).*tst; G=double(G).*tst; B=double(B).*tst; im3(:,:,1) = R; im3(:,:,2) = G; im3(:,:,3) = B; figure,imshow(im3,[]),title('White Point Image'); %Rav=mean(mean(R)); %Gav=mean(mean(G)); %Bav=mean(mean(B)); Rav=sum(sum(double(R)))/c; Gav=sum(sum(double(G)))/c; Bav=sum(sum(double(B)))/c; %Ymax=double(max(max(Lu)))/15; Ymax=double(max(max(Lu))); %Rgain=Ymax/Rav; %Ggain=Ymax/Gav; %Bgain=Ymax/Bav; Rgain=Gav/Rav; Ggain=1; Bgain=Gav/Bav; im(:,:,1)=im(:,:,1)*Rgain; im(:,:,2)=im(:,:,2)*Ggain; im(:,:,3)=im(:,:,3)*Bgain; W=uint8(im); figure,imshow(im2,[]),title('Original Image'); figure,imshow(im,[]),title('Corrected Image');