conversion of floating point to integer

My code to match a image with the images in database is :

img=handles.y;

mat=sum(img,1);

mat=mat(:);

k=zeros(30,1);

for i=1:30

      str=strcat(int2str(i),'.pgm');  
      %concatenates two strings that form the name of the image
      eval('img1=imread(str);');
      mat1=sum(img1,1);
      mat1=mat1(:);
      k(i)=max(abs(normxcorr2(mat1,mat)));
  end

%take maximum from matrix & store its value & position      

vmax=max(k)

uval=int8(fix(vmax))

if uval>=1

      figure(11);
      op=imread('match.jpg');
      imshow(op);

else

      figure(12);
      dp=imread('mismatch.jpg');
      imshow(dp);

why int8(fix(x)) or int16(fix(x)) or int32(fix(x)) or int64(fix(x)) gives value 1 for vmax=1.0000 for few input images and gives value 0 for vmax=1.0000 for few other input images??

Answer 1

[Walter Roberson]

Because vmax is not exactly 1 in those other images.

http://matlab.wikia.com/wiki/FAQ#Why_is_0.3_-_0.2_-_0.1_.28or_similar.29_not_equal_to_zero.3F

[prateek]

Actually there are two test images which I have to check for matching with the images in my database.

When I display the value of vmax on command window for first image,it shows 1.0000 and int8(fix(vmax)) converts this value to 1. Till here its okay.

But for the second image when I display the value of vmax on command window for first image,it shows 1.0000 and int8(fix(vmax)) converts this value to 0.

Why is it so?

I know 1.0000 is slightly greater than 1 but int8(fix(x)) should convert all values of x>1 to 1.But this is not happening in my case.

[Walter Roberson]

1.0000 is not necessarily slightly greater than 1. When you use "format short" then MATLAB displays all numbers within 5/10000 of 1 as 1.0000 including exactly 1 itself.

Display

vmax - 1

to see the difference between vmax and 1 .

[prateek]

Thanks a lot...it helped..