<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
  <head>
    <meta content="text/html; charset=ISO-8859-1"
      http-equiv="Content-Type">
  </head>
  <body bgcolor="#ffffff" text="#000000">
    Weird, weird, weird. This really worked! How can a Centroids
    X-position differ between two numbers that are _vertically_
    mirrored?<br>
    <br>
    Here is what I get:<br>
    <br>
    0:&nbsp; Size: 18 - Elongation: 1.44449 - Flatness: 1.44449 - Centroid:
    5,7 - dif: 2<br>
    1:&nbsp; Size: 10 - Elongation: 3.65503 - Flatness: 3.65503 - Centroid:
    20,7 - dif: 1<br>
    2:&nbsp; Size: 15 - Elongation: 2.06489 - Flatness: 2.06489 - Centroid:
    35,7 - dif: 2<br>
    3:&nbsp; Size: 15 - Elongation: 1.86812 - Flatness: 1.86812 - Centroid:
    50,7 - dif: 2<br>
    4:&nbsp; Size: 16 - Elongation: 1.63288 - Flatness: 1.63288 - Centroid:
    65,7 - dif: 2<br>
    5:&nbsp; Size: 17 - Elongation: 1.7793 - Flatness: 1.7793 - Centroid:
    81,7 - dif: 2<br>
    6:&nbsp; Size: 19 - Elongation: 1.54404 - Flatness: 1.54404 - Centroid:
    95,7 - dif: 1<br>
    7:&nbsp; Size: 12 - Elongation: 2.50164 - Flatness: 2.50164 - Centroid:
    108,8 - dif: 1<br>
    8:&nbsp; Size: 19 - Elongation: 1.53748 - Flatness: 1.53748 - Centroid:
    123,7 - dif: 2<br>
    9:&nbsp; Size: 19 - Elongation: 1.54404 - Flatness: 1.54404 - Centroid:
    138,7 - dif: 2<br>
    <br>
    ...but I don't get it. :) Anyway, thanks alot Sergio and Juan!<br>
    <br>
    <br>
    Best regards,<br>
    Christian<br>
    <br>
    <br>
    Am 27.09.2010 17:25, schrieb Juan Vidal:
    <blockquote
      cite="mid:AANLkTinE4FHPpUqqqLw9Sv+GNJp6qLfvaKOPpM93PrHo@mail.gmail.com"
      type="cite">What about using Sergio's idea in a "per-number"
      basis?&nbsp;
      <div><br>
      </div>
      <div>You could substract the x-positoin of the leftmost pixel of
        the number to its CentroidX, and maybe that result will help you
        to differenciate both numbers.</div>
      <div><br>
      </div>
      <div>Good luck!</div>
      <div><br>
      </div>
      <div><br>
        <br>
        <div class="gmail_quote">2010/9/27 Christian Werner <span
            dir="ltr">&lt;<a moz-do-not-send="true"
              href="mailto:christian.werner@rwth-aachen.de">christian.werner@rwth-aachen.de</a>&gt;</span><br>
          <blockquote class="gmail_quote" style="margin: 0pt 0pt 0pt
            0.8ex; border-left: 1px solid rgb(204, 204, 204);
            padding-left: 1ex;">
            <div bgcolor="#ffffff" text="#000000"> Thanks for the
              response, but the CentroidX already changes with the
              position of the number. So if I am analyzing a sequence of
              numbers "146998" the CentroidX will be different than in
              the sequence "6138549".<br>
              <br>
              This is really funny, who is responsible for that? Why use
              9 symbols to represent 10 digits? :)<br>
              <br>
              <br>
              &nbsp;<br>
              Am 27.09.2010 17:02, schrieb Sergio Vera:
              <div>
                <div class="h5">
                  <blockquote type="cite">
                    <div>Hello</div>
                    <div>Maybe it's a crazy/stupid/already tried idea
                      but,&nbsp;what about adding CentroidX to the list of
                      features? I would say that '6' should have the
                      centroid more to the left and '9' more to the
                      right...</div>
                    <div><br>
                    </div>
                    <div>HTH</div>
                    <div><br>
                    </div>
                    <div>Sergio</div>
                    <br>
                    <div class="gmail_quote">On Mon, Sep 27, 2010 at
                      4:44 PM, Christian Werner <span dir="ltr">&lt;<a
                          moz-do-not-send="true"
                          href="mailto:christian.werner@rwth-aachen.de"
                          target="_blank">christian.werner@rwth-aachen.de</a>&gt;</span>
                      wrote:<br>
                      <blockquote class="gmail_quote" style="margin: 0pt
                        0pt 0pt 0.8ex; border-left: 1px solid rgb(204,
                        204, 204); padding-left: 1ex;">&nbsp;Hi!<br>
                        <br>
                        I am doing a nice little OCR for some numbers.
                        See the attached image. I can identify each
                        number as an object. When I look at each
                        object's attributes, I see a fatal similarity
                        between the number '6' and '9':<br>
                        <br>
                        0: Size: 18 - Elongation: 1.44449 - Flatness:
                        1.44449 - CentroidY: 6<br>
                        1: Size: 10 - Elongation: 3.65503 - Flatness:
                        3.65503 - CentroidY: 6<br>
                        2: Size: 15 - Elongation: 2.06489 - Flatness:
                        2.06489 - CentroidY: 6<br>
                        3: Size: 15 - Elongation: 1.86812 - Flatness:
                        1.86812 - CentroidY: 6<br>
                        4: Size: 16 - Elongation: 1.63288 - Flatness:
                        1.63288 - CentroidY: 6<br>
                        5: Size: 17 - Elongation: 1.7793 - Flatness:
                        1.7793 - CentroidY: 6<br>
                        6: Size: 19 - Elongation: 1.54404 - Flatness:
                        1.54404 - CentroidY: 6<br>
                        7: Size: 12 - Elongation: 2.50164 - Flatness:
                        2.50164 - CentroidY: 7<br>
                        8: Size: 19 - Elongation: 1.53748 - Flatness:
                        1.53748 - CentroidY: 6<br>
                        9: Size: 19 - Elongation: 1.54404 - Flatness:
                        1.54404 - CentroidY: 6<br>
                        <br>
                        Which means, that I can't distinguish '6' from a
                        '9'. Does anybody have any idea how I could do
                        that?<br>
                        <br>
                        Best regards,<br>
                        <font color="#888888"> Christian<br>
                        </font><br>
                        _____________________________________<br>
                        Powered by <a moz-do-not-send="true"
                          href="http://www.kitware.com" target="_blank">www.kitware.com</a><br>
                        <br>
                        Visit other Kitware open-source projects at<br>
                        <a moz-do-not-send="true"
                          href="http://www.kitware.com/opensource/opensource.html"
                          target="_blank">http://www.kitware.com/opensource/opensource.html</a><br>
                        <br>
                        Kitware offers ITK Training Courses, for more
                        information visit:<br>
                        <a moz-do-not-send="true"
                          href="http://www.kitware.com/products/protraining.html"
                          target="_blank">http://www.kitware.com/products/protraining.html</a><br>
                        <br>
                        Please keep messages on-topic and check the ITK
                        FAQ at:<br>
                        <a moz-do-not-send="true"
                          href="http://www.itk.org/Wiki/ITK_FAQ"
                          target="_blank">http://www.itk.org/Wiki/ITK_FAQ</a><br>
                        <br>
                        Follow this link to subscribe/unsubscribe:<br>
                        <a moz-do-not-send="true"
                          href="http://www.itk.org/mailman/listinfo/insight-users"
                          target="_blank">http://www.itk.org/mailman/listinfo/insight-users</a><br>
                        <br>
                      </blockquote>
                    </div>
                    <br>
                    <br clear="all">
                    <br>
                    -- <br>
                    Sergio Vera<br>
                    <br>
                    &nbsp;Alma IT Systems<br>
                    &nbsp;C/ Vilana, 4B, 4&ordm; 1&ordf;<br>
                    &nbsp;08022 Barcelona<br>
                    &nbsp;T. (+34) 932 380 592<br>
                    &nbsp;<a moz-do-not-send="true"
                      href="http://www.alma3d.com" target="_blank">www.alma3d.com</a><br>
                  </blockquote>
                  <br>
                </div>
              </div>
            </div>
            <br>
            _____________________________________<br>
            Powered by <a moz-do-not-send="true"
              href="http://www.kitware.com" target="_blank">www.kitware.com</a><br>
            <br>
            Visit other Kitware open-source projects at<br>
            <a moz-do-not-send="true"
              href="http://www.kitware.com/opensource/opensource.html"
              target="_blank">http://www.kitware.com/opensource/opensource.html</a><br>
            <br>
            Kitware offers ITK Training Courses, for more information
            visit:<br>
            <a moz-do-not-send="true"
              href="http://www.kitware.com/products/protraining.html"
              target="_blank">http://www.kitware.com/products/protraining.html</a><br>
            <br>
            Please keep messages on-topic and check the ITK FAQ at:<br>
            <a moz-do-not-send="true"
              href="http://www.itk.org/Wiki/ITK_FAQ" target="_blank">http://www.itk.org/Wiki/ITK_FAQ</a><br>
            <br>
            Follow this link to subscribe/unsubscribe:<br>
            <a moz-do-not-send="true"
              href="http://www.itk.org/mailman/listinfo/insight-users"
              target="_blank">http://www.itk.org/mailman/listinfo/insight-users</a><br>
            <br>
          </blockquote>
        </div>
        <br>
        <br clear="all">
        <br>
        -- <br>
        Juan Vidal Allende<br>
      </div>
    </blockquote>
    <br>
  </body>
</html>