<?xml version="1.0" encoding="UTF-8"?>
<rss xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#" xmlns:taxo="http://purl.org/rss/1.0/modules/taxonomy/" version="2.0">
  <channel>
    <title>topic WarpAffine in Intel® Integrated Performance Primitives</title>
    <link>https://community.intel.com/t5/Intel-Integrated-Performance/WarpAffine/m-p/785072#M1921</link>
    <description>Hi Sid,&lt;BR /&gt;&lt;BR /&gt;Let affine transform is &lt;BR /&gt; xD = c00*xS + c01*yS + c02&lt;BR /&gt;&lt;DIV class="Nth msg"&gt; yD = c10*xS + c11*yS + c12&lt;/DIV&gt;&lt;BR /&gt;The coordinates of centers are (xScenter, yScenter) and (xDcenter, yDcenter).&lt;BR /&gt;So we should use:&lt;BR /&gt;&lt;DIV class="Nth msg"&gt; xD = c00*(xS-xScenter) + c01*(yS-yScenter) + xDcenter&lt;/DIV&gt;&lt;DIV class="Nth msg"&gt; yD = c10*(xS-xScenter) + c11*(yS-yScenter) + yDcenter&lt;/DIV&gt;&lt;BR /&gt;Then the desired coefficients are&lt;BR /&gt;&lt;DIV class="Nth msg"&gt; 02 = xDcenter - c00*xScenter - c01*yScenter&lt;/DIV&gt;&lt;DIV class="Nth msg"&gt; 12 = yDcenter - c10*xScenter - c11*yScenter&lt;/DIV&gt;&lt;BR /&gt;Thanks,&lt;BR /&gt; Beg</description>
    <pubDate>Fri, 11 Mar 2011 07:50:24 GMT</pubDate>
    <dc:creator>Yuri_Tikhomirov__Int</dc:creator>
    <dc:date>2011-03-11T07:50:24Z</dc:date>
    <item>
      <title>WarpAffine</title>
      <link>https://community.intel.com/t5/Intel-Integrated-Performance/WarpAffine/m-p/785070#M1919</link>
      <description>I am looking for some help to do the following: I have a 2D affine&lt;BR /&gt;transform obtained from an external source, and I am trying to use&lt;BR /&gt;this to resample an Image using WarpAffine. The transformation parameters&lt;BR /&gt;were calculated with reference to a center different from the origin. Since &lt;BR /&gt;WarpAffine has no parameter to specify thre center of the transform, I wanted&lt;BR /&gt;to know what sequence of steps I should pass my coeff[2][3] through, to ensure &lt;BR /&gt;that the transformation applied to the image is around the same center with &lt;BR /&gt;reference to which it was derived.&lt;BR /&gt;&lt;BR /&gt;Thanks,&lt;BR /&gt;Sid.</description>
      <pubDate>Thu, 10 Mar 2011 01:38:51 GMT</pubDate>
      <guid>https://community.intel.com/t5/Intel-Integrated-Performance/WarpAffine/m-p/785070#M1919</guid>
      <dc:creator>siddy</dc:creator>
      <dc:date>2011-03-10T01:38:51Z</dc:date>
    </item>
    <item>
      <title>WarpAffine</title>
      <link>https://community.intel.com/t5/Intel-Integrated-Performance/WarpAffine/m-p/785071#M1920</link>
      <description>Hi again,&lt;BR /&gt; I though a bit more about the problem, and I figured out&lt;BR /&gt;that I will have to recenter the coordinates, i.e X-C, prior to&lt;BR /&gt;applying the resampling on the grid. To restate,&lt;BR /&gt;If the usual approach for resampling is X' = AX, where A is the&lt;BR /&gt;transform and X is the image grid (giving f(X') actually), how do&lt;BR /&gt;I go about implementing X' = A(X-C) using available ip primitives, &lt;BR /&gt;where C is the center of the transform. &lt;BR /&gt;Thanks again,&lt;BR /&gt;Sid.</description>
      <pubDate>Thu, 10 Mar 2011 19:17:40 GMT</pubDate>
      <guid>https://community.intel.com/t5/Intel-Integrated-Performance/WarpAffine/m-p/785071#M1920</guid>
      <dc:creator>siddy</dc:creator>
      <dc:date>2011-03-10T19:17:40Z</dc:date>
    </item>
    <item>
      <title>WarpAffine</title>
      <link>https://community.intel.com/t5/Intel-Integrated-Performance/WarpAffine/m-p/785072#M1921</link>
      <description>Hi Sid,&lt;BR /&gt;&lt;BR /&gt;Let affine transform is &lt;BR /&gt; xD = c00*xS + c01*yS + c02&lt;BR /&gt;&lt;DIV class="Nth msg"&gt; yD = c10*xS + c11*yS + c12&lt;/DIV&gt;&lt;BR /&gt;The coordinates of centers are (xScenter, yScenter) and (xDcenter, yDcenter).&lt;BR /&gt;So we should use:&lt;BR /&gt;&lt;DIV class="Nth msg"&gt; xD = c00*(xS-xScenter) + c01*(yS-yScenter) + xDcenter&lt;/DIV&gt;&lt;DIV class="Nth msg"&gt; yD = c10*(xS-xScenter) + c11*(yS-yScenter) + yDcenter&lt;/DIV&gt;&lt;BR /&gt;Then the desired coefficients are&lt;BR /&gt;&lt;DIV class="Nth msg"&gt; 02 = xDcenter - c00*xScenter - c01*yScenter&lt;/DIV&gt;&lt;DIV class="Nth msg"&gt; 12 = yDcenter - c10*xScenter - c11*yScenter&lt;/DIV&gt;&lt;BR /&gt;Thanks,&lt;BR /&gt; Beg</description>
      <pubDate>Fri, 11 Mar 2011 07:50:24 GMT</pubDate>
      <guid>https://community.intel.com/t5/Intel-Integrated-Performance/WarpAffine/m-p/785072#M1921</guid>
      <dc:creator>Yuri_Tikhomirov__Int</dc:creator>
      <dc:date>2011-03-11T07:50:24Z</dc:date>
    </item>
    <item>
      <title>WarpAffine</title>
      <link>https://community.intel.com/t5/Intel-Integrated-Performance/WarpAffine/m-p/785073#M1922</link>
      <description>Hi Beg,&lt;BR /&gt; Let me rejuvenate this thread, with reference to my latest thread regarding getAffineTransform for a quadilateral:&lt;BR /&gt;&lt;BR /&gt;if the initial quad of Q = {(xi,yi), i=0,1,2,3}, and the final quad is Q'= {(x'i,y'i), i = 0,1,2,3}, and coeffs[2][3]&lt;BR /&gt;are obtained by ippiGetAffineTransform between Q and Q', to what center are this coeffs referred to? Am i right&lt;BR /&gt;in thinking : (x2-x1)/2 and (y2-y1)/2 ?&lt;BR /&gt;&lt;BR /&gt;Thanks&lt;BR /&gt;Sid.&lt;BR /&gt; &lt;BR /&gt;</description>
      <pubDate>Wed, 04 Jan 2012 01:38:45 GMT</pubDate>
      <guid>https://community.intel.com/t5/Intel-Integrated-Performance/WarpAffine/m-p/785073#M1922</guid>
      <dc:creator>siddy</dc:creator>
      <dc:date>2012-01-04T01:38:45Z</dc:date>
    </item>
  </channel>
</rss>

