# Optimize a code

1 view (last 30 days)
mortain Antonio on 18 Apr 2011
Hello,
I attach here a piece of code which I think works quite slow because some variables are called, written and recalled...this just wastes lot of time. I'm new, but I think that someone might help.
As you can read, the quantities hue, inten and sat are used, but saved as matrix, whereas I need them as column vector to be used, so I need to convert the matrix to column vector by writing on a file, then read from the file those values and re-write on a file the same quantity plus the temperature.
Please, do not hesitate to ask me question if you ahve any or to point out some manner to mak the code working faster and nicer.
Kind regards, Antonio
%input image###.bmp, cal.txt (calibration file)
%output out_hue2temp.dat (tecplot file of temperature)
clear all
close all
X0 =1
Y0 =1
H = 492
W = 658
%W = 290
%H = 56
%X0 = 162
%Y0 = 81
%save as matrix variables the hue, inten and sat of the picture
name = '28.60';
ifile = strcat( name,'.bmp');
x = imread( ifile, 'BMP' );
x = x(Y0:Y0+H-1,X0:X0+W-1,:);
y = RGB2HSV(x);
hue = y(:,:,1)* 255;
inten = y(:,:,3)* 255;
sat = y(:,:,2)*255;
figure;imshow(x)
%%%%%%write initial HUE or INTENSITY %%%%%%%
%This writing to file is useful so that I transoform hue, inten and sat from matrix to vector of one column
ofile = strcat( 'totHUE.dat');
fid = fopen(ofile,'w');
for ii = 1:H
for j = 1:W
fprintf(fid, '%6.2f %6.2f %3.1f %3.1f %3.1f \n',ii,j,hue(ii,j),inten(ii,j),sat(ii,j));
end
end
fclose(fid);
datawh=[W,H]
save data.dat datawh -ascii
%I take the columns of inten, sat and hue from the file
hue=totHUE;
x=totHUE(:,1);
y=totHUE(:,2);
z=totHUE(:,3);
v=totHUE(:,4);
w=totHUE(:,5);
%plot(cal(:,1),cal(:,2))
temp=x;
temp(:)=0;
for ii = 1:length(x(:,1))
temp(ii)=interp1(calHUE(:,1), calHUE(:,2), z(ii),'nearest');
end
% Resave on a file the values of before and temperature
W=data(1);
H=data(2);
ofile2 = strcat('out_hue2temp.dat');
fid2 = fopen(ofile2,'w');
fprintf(fid2,'VARIABLE="X","Y","TEMP","HUE","INTEN","SAT",',W,H)
fprintf(fid2,'\n zone i=%d j=%d\n',W,H)
for ii = 1:length(x(:,1))
fprintf(fid2,'%6.2f %6.2f %3.1f %3.1f %3.1f %3.1f \n',x(ii),y(ii),temp(ii),z(ii),v(ii),w(ii));
end

Jarrod Rivituso on 18 Apr 2011
1. I would use the Profiler to see where your code is going slow: http://www.mathworks.com/help/techdoc/ref/profile.html
2. You should not need to write out to a file just to reshape a matrix to a column vector. Try the reshape function: http://www.mathworks.com/help/techdoc/ref/reshape.html
mortain Antonio on 18 Apr 2011
1) Thank you, I've done it and most of the time is for the calculations.
2) the result now is:
hueR = y(:,:,1)* 255;
intenR = y(:,:,3)* 255;
satR = y(:,:,2)*255;
hueT=transpose(hueR);
intenT=transpose(intenR);
satT=transpose(satR);
hue=reshape(hueT,[],1);
inten=reshape(intenT,[],1);
sat=reshape(satT,[],1);
which is, first I write the matrix, then I transpose it and then I convert into a vector. IT is a bit faster and even nicer to see, since I took out some part of the code. But, the fprintf does not work anymore as before.....
clear all
close all
W = 290
H = 56
X0 = 81
Y0 = 162
name = '28.60';
ifile = strcat( name,'.bmp');
x = imread( ifile, 'BMP' );
x = x(X0:X0+H-1,Y0:Y0+W-1,:);
y = RGB2HSV(x);
hueR = y(:,:,1)* 255;
intenR = y(:,:,3)* 255;
satR = y(:,:,2)*255;
hueT=transpose(hueR);
intenT=transpose(intenR);
satT=transpose(satR);
hue=reshape(hueT,[],1);
inten=reshape(intenT,[],1);
sat=reshape(satT,[],1);
figure;imshow(x)
temp=hue;
temp(:)=0;
for ii = 1:length(x(:,1))
temp(ii)=interp1(calHUE(:,1), calHUE(:,2), hue(ii),'nearest');
end
W=data(1);
H=data(2);
ofile2 = strcat('out_hue2temp.dat');
fid2 = fopen(ofile2,'w');
fprintf(fid2,'VARIABLE="X","Y","TEMP","HUE","INTEN","SAT",',W,H);
fprintf(fid2,'\n zone i=%d j=%d\n',W,H);
for ii = 1:H
for j = 1 : W
fprintf(fid2,'%6.2f %6.2f %3.1f %3.1f %3.1f %3.1f \n',ii,j,temp(:),hue(:),inten(:),sat(:))
end
end
have you got any suggestion?