Why am I unable to read readings of an Image using ocr?

I am unable to read numbers from any image like this. First I used the code:

   i=imread('speed.jpg');
    i= imresize(i, 1.4);
    f= ocr(i,'CharacterSet','.0123456789');
    recognizedText = f.Words;

the output i get is:

        ' '
      ' 1  . 5 '
      '91 11 915 021 '
      ' 011 '
      '106 9059'

Then I tried using a better code by binarizing, dilating, and using regionprops for a better solution. This time, I don't get the output at all. The above code works perfectly for other images with numbers in linear fashion. Please help.