git.gag.com Git - fw/stlink/blob - exampleF4/CMSIS/Documentation/DSP_Lib/html/group___corr.html

   1 <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
   2 <html xmlns="http://www.w3.org/1999/xhtml">
   3 <head>
   4 <meta http-equiv="Content-Type" content="text/xhtml;charset=UTF-8"/>
   5 <title>CMSIS DSP Software Library: Correlation</title>
   6 <link href="tabs.css" rel="stylesheet" type="text/css"/>
   7 <link href="search/search.css" rel="stylesheet" type="text/css"/>
   8 <script type="text/javaScript" src="search/search.js"></script>
   9 <link href="doxygen.css" rel="stylesheet" type="text/css"/>
  10 </head>
  11 <body onload='searchBox.OnSelectItem(0);'>
  12 <!-- Generated by Doxygen 1.7.2 -->
  13 <script type="text/javascript"><!--
  14 var searchBox = new SearchBox("searchBox", "search",false,'Search');
  15 --></script>
  16 <div class="navigation" id="top">
  17   <div class="tabs">
  18     <ul class="tablist">
  19       <li><a href="index.html"><span>Main&#160;Page</span></a></li>
  20       <li><a href="modules.html"><span>Modules</span></a></li>
  21       <li><a href="annotated.html"><span>Data&#160;Structures</span></a></li>
  22       <li><a href="files.html"><span>Files</span></a></li>
  23       <li><a href="examples.html"><span>Examples</span></a></li>
  24       <li id="searchli">
  25         <div id="MSearchBox" class="MSearchBoxInactive">
  26         <span class="left">
  27           <img id="MSearchSelect" src="search/mag_sel.png"
  28                onmouseover="return searchBox.OnSearchSelectShow()"
  29                onmouseout="return searchBox.OnSearchSelectHide()"
  30                alt=""/>
  31           <input type="text" id="MSearchField" value="Search" accesskey="S"
  32                onfocus="searchBox.OnSearchFieldFocus(true)"
  33                onblur="searchBox.OnSearchFieldFocus(false)"
  34                onkeyup="searchBox.OnSearchFieldChange(event)"/>
  35           </span><span class="right">
  36             <a id="MSearchClose" href="javascript:searchBox.CloseResultsWindow()"><img id="MSearchCloseImg" border="0" src="search/close.png" alt=""/></a>
  37           </span>
  38         </div>
  39       </li>
  40     </ul>
  41   </div>
  42 </div>
  43 <div class="header">
  44   <div class="summary">
  45 <a href="#func-members">Functions</a>  </div>
  46   <div class="headertitle">
  47 <h1>Correlation<br/>
  48 <small>
  49 [<a class="el" href="group__group_filters.html">Filtering Functions</a>]</small>
  50 </h1>  </div>
  51 </div>
  52 <div class="contents">
  53 <table class="memberdecls">
  54 <tr><td colspan="2"><h2><a name="func-members"></a>
  55 Functions</h2></td></tr>
  56 <tr><td class="memItemLeft" align="right" valign="top">void&#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="group___corr.html#ga22021e4222773f01e9960358a531cfb8">arm_correlate_f32</a> (<a class="el" href="arm__math_8h.html#a4611b605e45ab401f02cab15c5e38715">float32_t</a> *pSrcA, uint32_t <a class="el" href="arm__convolution__example__f32_8c.html#ace48ed566e2cd6a680f0681192e6af28">srcALen</a>, <a class="el" href="arm__math_8h.html#a4611b605e45ab401f02cab15c5e38715">float32_t</a> *pSrcB, uint32_t <a class="el" href="arm__convolution__example__f32_8c.html#aea71286f498978c5ed3775609b974fc8">srcBLen</a>, <a class="el" href="arm__math_8h.html#a4611b605e45ab401f02cab15c5e38715">float32_t</a> *pDst)</td></tr>
  57 <tr><td class="memItemLeft" align="right" valign="top">void&#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="group___corr.html#ga1367dc6c80476406c951e68d7fac4e8c">arm_correlate_q31</a> (<a class="el" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0">q31_t</a> *pSrcA, uint32_t <a class="el" href="arm__convolution__example__f32_8c.html#ace48ed566e2cd6a680f0681192e6af28">srcALen</a>, <a class="el" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0">q31_t</a> *pSrcB, uint32_t <a class="el" href="arm__convolution__example__f32_8c.html#aea71286f498978c5ed3775609b974fc8">srcBLen</a>, <a class="el" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0">q31_t</a> *pDst)</td></tr>
  58 <tr><td class="memItemLeft" align="right" valign="top">void&#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="group___corr.html#gabecd3d7b077dbbef43f93e9e037815ed">arm_correlate_fast_q31</a> (<a class="el" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0">q31_t</a> *pSrcA, uint32_t <a class="el" href="arm__convolution__example__f32_8c.html#ace48ed566e2cd6a680f0681192e6af28">srcALen</a>, <a class="el" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0">q31_t</a> *pSrcB, uint32_t <a class="el" href="arm__convolution__example__f32_8c.html#aea71286f498978c5ed3775609b974fc8">srcBLen</a>, <a class="el" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0">q31_t</a> *pDst)</td></tr>
  59 <tr><td class="memItemLeft" align="right" valign="top">void&#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="group___corr.html#ga5ec96b8e420d68b0e626df0812274d46">arm_correlate_q15</a> (<a class="el" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea">q15_t</a> *pSrcA, uint32_t <a class="el" href="arm__convolution__example__f32_8c.html#ace48ed566e2cd6a680f0681192e6af28">srcALen</a>, <a class="el" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea">q15_t</a> *pSrcB, uint32_t <a class="el" href="arm__convolution__example__f32_8c.html#aea71286f498978c5ed3775609b974fc8">srcBLen</a>, <a class="el" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea">q15_t</a> *pDst)</td></tr>
  60 <tr><td class="memItemLeft" align="right" valign="top">void&#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="group___corr.html#gac8de3da44f58e86c2c86156276ca154f">arm_correlate_fast_q15</a> (<a class="el" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea">q15_t</a> *pSrcA, uint32_t <a class="el" href="arm__convolution__example__f32_8c.html#ace48ed566e2cd6a680f0681192e6af28">srcALen</a>, <a class="el" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea">q15_t</a> *pSrcB, uint32_t <a class="el" href="arm__convolution__example__f32_8c.html#aea71286f498978c5ed3775609b974fc8">srcBLen</a>, <a class="el" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea">q15_t</a> *pDst)</td></tr>
  61 <tr><td class="memItemLeft" align="right" valign="top">void&#160;</td><td class="memItemRight" valign="bottom"><a class="el" href="group___corr.html#ga284ddcc49e4ac532d52a70d0383c5992">arm_correlate_q7</a> (<a class="el" href="arm__math_8h.html#ae541b6f232c305361e9b416fc9eed263">q7_t</a> *pSrcA, uint32_t <a class="el" href="arm__convolution__example__f32_8c.html#ace48ed566e2cd6a680f0681192e6af28">srcALen</a>, <a class="el" href="arm__math_8h.html#ae541b6f232c305361e9b416fc9eed263">q7_t</a> *pSrcB, uint32_t <a class="el" href="arm__convolution__example__f32_8c.html#aea71286f498978c5ed3775609b974fc8">srcBLen</a>, <a class="el" href="arm__math_8h.html#ae541b6f232c305361e9b416fc9eed263">q7_t</a> *pDst)</td></tr>
  62 </table>
  63 <hr/><a name="_details"></a><h2>Detailed Description</h2>
  64 <p>Correlation is a mathematical operation that is similar to convolution. As with convolution, correlation uses two signals to produce a third signal. The underlying algorithms in correlation and convolution are identical except that one of the inputs is flipped in convolution. Correlation is commonly used to measure the similarity between two signals. It has applications in pattern recognition, cryptanalysis, and searching. The CMSIS library provides correlation functions for Q7, Q15, Q31 and floating-point data types. Fast versions of the Q15 and Q31 functions are also provided.</p>
  65 <dl class="user"><dt><b>Algorithm </b></dt><dd>Let <code>a[n]</code> and <code>b[n]</code> be sequences of length <code>srcALen</code> and <code>srcBLen</code> samples respectively. The convolution of the two signals is denoted by <pre>
  66                    c[n] = a[n] * b[n]
  67  </pre> In correlation, one of the signals is flipped in time <pre>
  68                    c[n] = a[n] * b[-n]
  69  </pre></dd></dl>
  70 <dl class="user"><dt><b></b></dt><dd>and this is mathematically defined as <div align="center">
  71 <img src="CorrelateEquation.gif" alt="CorrelateEquation.gif"/>
  72 </div>
  73  </dd></dl>
  74 <dl class="user"><dt><b></b></dt><dd>The <code>pSrcA</code> points to the first input vector of length <code>srcALen</code> and <code>pSrcB</code> points to the second input vector of length <code>srcBLen</code>. The result <code>c[n]</code> is of length <code>2 * max(srcALen, srcBLen) - 1</code> and is defined over the interval <code>n=0, 1, 2, ..., (2 * max(srcALen, srcBLen) - 2)</code>. The output result is written to <code>pDst</code> and the calling function must allocate <code>2 * max(srcALen, srcBLen) - 1</code> words for the result.</dd></dl>
  75 <p><b>Note</b> </p>
  76 <dl class="user"><dt><b></b></dt><dd>The <code>pDst</code> should be initialized to all zeros before being used.</dd></dl>
  77 <p><b>Fixed-Point Behavior</b> </p>
  78 <dl class="user"><dt><b></b></dt><dd>Correlation requires summing up a large number of intermediate products. As such, the Q7, Q15, and Q31 functions run a risk of overflow and saturation. Refer to the function specific documentation below for further details of the particular algorithm used. </dd></dl>
  79 <hr/><h2>Function Documentation</h2>
  80 <a class="anchor" id="ga22021e4222773f01e9960358a531cfb8"></a><!-- doxytag: member="arm_correlate_f32.c::arm_correlate_f32" ref="ga22021e4222773f01e9960358a531cfb8" args="(float32_t *pSrcA, uint32_t srcALen, float32_t *pSrcB, uint32_t srcBLen, float32_t *pDst)" -->
  81 <div class="memitem">
  82 <div class="memproto">
  83       <table class="memname">
  84         <tr>
  85           <td class="memname">void arm_correlate_f32 </td>
  86           <td>(</td>
  87           <td class="paramtype"><a class="el" href="arm__math_8h.html#a4611b605e45ab401f02cab15c5e38715">float32_t</a> *&#160;</td>
  88           <td class="paramname"> <em>pSrcA</em>, </td>
  89         </tr>
  90         <tr>
  91           <td class="paramkey"></td>
  92           <td></td>
  93           <td class="paramtype">uint32_t&#160;</td>
  94           <td class="paramname"> <em>srcALen</em>, </td>
  95         </tr>
  96         <tr>
  97           <td class="paramkey"></td>
  98           <td></td>
  99           <td class="paramtype"><a class="el" href="arm__math_8h.html#a4611b605e45ab401f02cab15c5e38715">float32_t</a> *&#160;</td>
 100           <td class="paramname"> <em>pSrcB</em>, </td>
 101         </tr>
 102         <tr>
 103           <td class="paramkey"></td>
 104           <td></td>
 105           <td class="paramtype">uint32_t&#160;</td>
 106           <td class="paramname"> <em>srcBLen</em>, </td>
 107         </tr>
 108         <tr>
 109           <td class="paramkey"></td>
 110           <td></td>
 111           <td class="paramtype"><a class="el" href="arm__math_8h.html#a4611b605e45ab401f02cab15c5e38715">float32_t</a> *&#160;</td>
 112           <td class="paramname"> <em>pDst</em>&#160;</td>
 113         </tr>
 114         <tr>
 115           <td></td>
 116           <td>)</td>
 117           <td></td><td></td>
 118         </tr>
 119       </table>
 120 </div>
 121 <div class="memdoc">
 122
 123 <p>Correlation of floating-point sequences. </p>
 124 <dl><dt><b>Parameters:</b></dt><dd>
 125   <table class="params">
 126     <tr><td class="paramdir">[in]</td><td class="paramname">*pSrcA</td><td>points to the first input sequence. </td></tr>
 127     <tr><td class="paramdir">[in]</td><td class="paramname">srcALen</td><td>length of the first input sequence. </td></tr>
 128     <tr><td class="paramdir">[in]</td><td class="paramname">*pSrcB</td><td>points to the second input sequence. </td></tr>
 129     <tr><td class="paramdir">[in]</td><td class="paramname">srcBLen</td><td>length of the second input sequence. </td></tr>
 130     <tr><td class="paramdir">[out]</td><td class="paramname">*pDst</td><td>points to the location where the output result is written. Length 2 * max(srcALen, srcBLen) - 1. </td></tr>
 131   </table>
 132   </dd>
 133 </dl>
 134 <dl class="return"><dt><b>Returns:</b></dt><dd>none. </dd></dl>
 135
 136 <p>Definition at line <a class="el" href="arm__correlate__f32_8c_source.html#l00095">95</a> of file <a class="el" href="arm__correlate__f32_8c_source.html">arm_correlate_f32.c</a>.</p>
 137
 138 </div>
 139 </div>
 140 <a class="anchor" id="ga1367dc6c80476406c951e68d7fac4e8c"></a><!-- doxytag: member="arm_correlate_q31.c::arm_correlate_q31" ref="ga1367dc6c80476406c951e68d7fac4e8c" args="(q31_t *pSrcA, uint32_t srcALen, q31_t *pSrcB, uint32_t srcBLen, q31_t *pDst)" -->
 141 <div class="memitem">
 142 <div class="memproto">
 143       <table class="memname">
 144         <tr>
 145           <td class="memname">void arm_correlate_q31 </td>
 146           <td>(</td>
 147           <td class="paramtype"><a class="el" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0">q31_t</a> *&#160;</td>
 148           <td class="paramname"> <em>pSrcA</em>, </td>
 149         </tr>
 150         <tr>
 151           <td class="paramkey"></td>
 152           <td></td>
 153           <td class="paramtype">uint32_t&#160;</td>
 154           <td class="paramname"> <em>srcALen</em>, </td>
 155         </tr>
 156         <tr>
 157           <td class="paramkey"></td>
 158           <td></td>
 159           <td class="paramtype"><a class="el" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0">q31_t</a> *&#160;</td>
 160           <td class="paramname"> <em>pSrcB</em>, </td>
 161         </tr>
 162         <tr>
 163           <td class="paramkey"></td>
 164           <td></td>
 165           <td class="paramtype">uint32_t&#160;</td>
 166           <td class="paramname"> <em>srcBLen</em>, </td>
 167         </tr>
 168         <tr>
 169           <td class="paramkey"></td>
 170           <td></td>
 171           <td class="paramtype"><a class="el" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0">q31_t</a> *&#160;</td>
 172           <td class="paramname"> <em>pDst</em>&#160;</td>
 173         </tr>
 174         <tr>
 175           <td></td>
 176           <td>)</td>
 177           <td></td><td></td>
 178         </tr>
 179       </table>
 180 </div>
 181 <div class="memdoc">
 182
 183 <p>Correlation of Q31 sequences. </p>
 184 <dl><dt><b>Parameters:</b></dt><dd>
 185   <table class="params">
 186     <tr><td class="paramdir">[in]</td><td class="paramname">*pSrcA</td><td>points to the first input sequence. </td></tr>
 187     <tr><td class="paramdir">[in]</td><td class="paramname">srcALen</td><td>length of the first input sequence. </td></tr>
 188     <tr><td class="paramdir">[in]</td><td class="paramname">*pSrcB</td><td>points to the second input sequence. </td></tr>
 189     <tr><td class="paramdir">[in]</td><td class="paramname">srcBLen</td><td>length of the second input sequence. </td></tr>
 190     <tr><td class="paramdir">[out]</td><td class="paramname">*pDst</td><td>points to the location where the output result is written. Length 2 * max(srcALen, srcBLen) - 1. </td></tr>
 191   </table>
 192   </dd>
 193 </dl>
 194 <dl class="return"><dt><b>Returns:</b></dt><dd>none.</dd></dl>
 195 <p><b>Scaling and Overflow Behavior:</b></p>
 196 <dl class="user"><dt><b></b></dt><dd>The function is implemented using an internal 64-bit accumulator. The accumulator has a 2.62 format and maintains full precision of the intermediate multiplication results but provides only a single guard bit. There is no saturation on intermediate additions. Thus, if the accumulator overflows it wraps around and distorts the result. The input signals should be scaled down to avoid intermediate overflows. Scale down one of the inputs by 1/min(srcALen, srcBLen)to avoid overflows since a maximum of min(srcALen, srcBLen) number of additions is carried internally. The 2.62 accumulator is right shifted by 31 bits and saturated to 1.31 format to yield the final result.</dd></dl>
 197 <dl class="user"><dt><b></b></dt><dd>See <code><a class="el" href="group___corr.html#gabecd3d7b077dbbef43f93e9e037815ed" title="Correlation of Q31 sequences (fast version) for Cortex-M3 and Cortex-M4.">arm_correlate_fast_q31()</a></code> for a faster but less precise implementation of this function for Cortex-M3 and Cortex-M4. </dd></dl>
 198
 199 <p>Definition at line <a class="el" href="arm__correlate__q31_8c_source.html#l00071">71</a> of file <a class="el" href="arm__correlate__q31_8c_source.html">arm_correlate_q31.c</a>.</p>
 200
 201 </div>
 202 </div>
 203 <a class="anchor" id="gabecd3d7b077dbbef43f93e9e037815ed"></a><!-- doxytag: member="arm_correlate_fast_q31.c::arm_correlate_fast_q31" ref="gabecd3d7b077dbbef43f93e9e037815ed" args="(q31_t *pSrcA, uint32_t srcALen, q31_t *pSrcB, uint32_t srcBLen, q31_t *pDst)" -->
 204 <div class="memitem">
 205 <div class="memproto">
 206       <table class="memname">
 207         <tr>
 208           <td class="memname">void arm_correlate_fast_q31 </td>
 209           <td>(</td>
 210           <td class="paramtype"><a class="el" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0">q31_t</a> *&#160;</td>
 211           <td class="paramname"> <em>pSrcA</em>, </td>
 212         </tr>
 213         <tr>
 214           <td class="paramkey"></td>
 215           <td></td>
 216           <td class="paramtype">uint32_t&#160;</td>
 217           <td class="paramname"> <em>srcALen</em>, </td>
 218         </tr>
 219         <tr>
 220           <td class="paramkey"></td>
 221           <td></td>
 222           <td class="paramtype"><a class="el" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0">q31_t</a> *&#160;</td>
 223           <td class="paramname"> <em>pSrcB</em>, </td>
 224         </tr>
 225         <tr>
 226           <td class="paramkey"></td>
 227           <td></td>
 228           <td class="paramtype">uint32_t&#160;</td>
 229           <td class="paramname"> <em>srcBLen</em>, </td>
 230         </tr>
 231         <tr>
 232           <td class="paramkey"></td>
 233           <td></td>
 234           <td class="paramtype"><a class="el" href="arm__math_8h.html#adc89a3547f5324b7b3b95adec3806bc0">q31_t</a> *&#160;</td>
 235           <td class="paramname"> <em>pDst</em>&#160;</td>
 236         </tr>
 237         <tr>
 238           <td></td>
 239           <td>)</td>
 240           <td></td><td></td>
 241         </tr>
 242       </table>
 243 </div>
 244 <div class="memdoc">
 245
 246 <p>Correlation of Q31 sequences (fast version) for Cortex-M3 and Cortex-M4. </p>
 247 <dl><dt><b>Parameters:</b></dt><dd>
 248   <table class="params">
 249     <tr><td class="paramdir">[in]</td><td class="paramname">*pSrcA</td><td>points to the first input sequence. </td></tr>
 250     <tr><td class="paramdir">[in]</td><td class="paramname">srcALen</td><td>length of the first input sequence. </td></tr>
 251     <tr><td class="paramdir">[in]</td><td class="paramname">*pSrcB</td><td>points to the second input sequence. </td></tr>
 252     <tr><td class="paramdir">[in]</td><td class="paramname">srcBLen</td><td>length of the second input sequence. </td></tr>
 253     <tr><td class="paramdir">[out]</td><td class="paramname">*pDst</td><td>points to the location where the output result is written. Length 2 * max(srcALen, srcBLen) - 1. </td></tr>
 254   </table>
 255   </dd>
 256 </dl>
 257 <dl class="return"><dt><b>Returns:</b></dt><dd>none.</dd></dl>
 258 <p><b>Scaling and Overflow Behavior:</b></p>
 259 <dl class="user"><dt><b></b></dt><dd>This function is optimized for speed at the expense of fixed-point precision and overflow protection. The result of each 1.31 x 1.31 multiplication is truncated to 2.30 format. These intermediate results are accumulated in a 32-bit register in 2.30 format. Finally, the accumulator is saturated and converted to a 1.31 result.</dd></dl>
 260 <dl class="user"><dt><b></b></dt><dd>The fast version has the same overflow behavior as the standard version but provides less precision since it discards the low 32 bits of each multiplication result. In order to avoid overflows completely the input signals must be scaled down. The input signals should be scaled down to avoid intermediate overflows. Scale down one of the inputs by 1/min(srcALen, srcBLen)to avoid overflows since a maximum of min(srcALen, srcBLen) number of additions is carried internally.</dd></dl>
 261 <dl class="user"><dt><b></b></dt><dd>See <code><a class="el" href="group___corr.html#ga1367dc6c80476406c951e68d7fac4e8c" title="Correlation of Q31 sequences.">arm_correlate_q31()</a></code> for a slower implementation of this function which uses 64-bit accumulation to provide higher precision. </dd></dl>
 262
 263 <p>Definition at line <a class="el" href="arm__correlate__fast__q31_8c_source.html#l00070">70</a> of file <a class="el" href="arm__correlate__fast__q31_8c_source.html">arm_correlate_fast_q31.c</a>.</p>
 264
 265 </div>
 266 </div>
 267 <a class="anchor" id="ga5ec96b8e420d68b0e626df0812274d46"></a><!-- doxytag: member="arm_correlate_q15.c::arm_correlate_q15" ref="ga5ec96b8e420d68b0e626df0812274d46" args="(q15_t *pSrcA, uint32_t srcALen, q15_t *pSrcB, uint32_t srcBLen, q15_t *pDst)" -->
 268 <div class="memitem">
 269 <div class="memproto">
 270       <table class="memname">
 271         <tr>
 272           <td class="memname">void arm_correlate_q15 </td>
 273           <td>(</td>
 274           <td class="paramtype"><a class="el" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea">q15_t</a> *&#160;</td>
 275           <td class="paramname"> <em>pSrcA</em>, </td>
 276         </tr>
 277         <tr>
 278           <td class="paramkey"></td>
 279           <td></td>
 280           <td class="paramtype">uint32_t&#160;</td>
 281           <td class="paramname"> <em>srcALen</em>, </td>
 282         </tr>
 283         <tr>
 284           <td class="paramkey"></td>
 285           <td></td>
 286           <td class="paramtype"><a class="el" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea">q15_t</a> *&#160;</td>
 287           <td class="paramname"> <em>pSrcB</em>, </td>
 288         </tr>
 289         <tr>
 290           <td class="paramkey"></td>
 291           <td></td>
 292           <td class="paramtype">uint32_t&#160;</td>
 293           <td class="paramname"> <em>srcBLen</em>, </td>
 294         </tr>
 295         <tr>
 296           <td class="paramkey"></td>
 297           <td></td>
 298           <td class="paramtype"><a class="el" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea">q15_t</a> *&#160;</td>
 299           <td class="paramname"> <em>pDst</em>&#160;</td>
 300         </tr>
 301         <tr>
 302           <td></td>
 303           <td>)</td>
 304           <td></td><td></td>
 305         </tr>
 306       </table>
 307 </div>
 308 <div class="memdoc">
 309
 310 <p>Correlation of Q15 sequences. </p>
 311 <dl><dt><b>Parameters:</b></dt><dd>
 312   <table class="params">
 313     <tr><td class="paramdir">[in]</td><td class="paramname">*pSrcA</td><td>points to the first input sequence. </td></tr>
 314     <tr><td class="paramdir">[in]</td><td class="paramname">srcALen</td><td>length of the first input sequence. </td></tr>
 315     <tr><td class="paramdir">[in]</td><td class="paramname">*pSrcB</td><td>points to the second input sequence. </td></tr>
 316     <tr><td class="paramdir">[in]</td><td class="paramname">srcBLen</td><td>length of the second input sequence. </td></tr>
 317     <tr><td class="paramdir">[out]</td><td class="paramname">*pDst</td><td>points to the location where the output result is written. Length 2 * max(srcALen, srcBLen) - 1. </td></tr>
 318   </table>
 319   </dd>
 320 </dl>
 321 <dl class="return"><dt><b>Returns:</b></dt><dd>none.</dd></dl>
 322 <p><b>Scaling and Overflow Behavior:</b></p>
 323 <dl class="user"><dt><b></b></dt><dd>The function is implemented using a 64-bit internal accumulator. Both inputs are in 1.15 format and multiplications yield a 2.30 result. The 2.30 intermediate results are accumulated in a 64-bit accumulator in 34.30 format. This approach provides 33 guard bits and there is no risk of overflow. The 34.30 result is then truncated to 34.15 format by discarding the low 15 bits and then saturated to 1.15 format.</dd></dl>
 324 <dl class="user"><dt><b></b></dt><dd>Refer to <code><a class="el" href="group___corr.html#gac8de3da44f58e86c2c86156276ca154f" title="Correlation of Q15 sequences (fast version) for Cortex-M3 and Cortex-M4.">arm_correlate_fast_q15()</a></code> for a faster but less precise version of this function for Cortex-M3 and Cortex-M4. </dd></dl>
 325
 326 <p>Definition at line <a class="el" href="arm__correlate__q15_8c_source.html#l00068">68</a> of file <a class="el" href="arm__correlate__q15_8c_source.html">arm_correlate_q15.c</a>.</p>
 327
 328 </div>
 329 </div>
 330 <a class="anchor" id="gac8de3da44f58e86c2c86156276ca154f"></a><!-- doxytag: member="arm_correlate_fast_q15.c::arm_correlate_fast_q15" ref="gac8de3da44f58e86c2c86156276ca154f" args="(q15_t *pSrcA, uint32_t srcALen, q15_t *pSrcB, uint32_t srcBLen, q15_t *pDst)" -->
 331 <div class="memitem">
 332 <div class="memproto">
 333       <table class="memname">
 334         <tr>
 335           <td class="memname">void arm_correlate_fast_q15 </td>
 336           <td>(</td>
 337           <td class="paramtype"><a class="el" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea">q15_t</a> *&#160;</td>
 338           <td class="paramname"> <em>pSrcA</em>, </td>
 339         </tr>
 340         <tr>
 341           <td class="paramkey"></td>
 342           <td></td>
 343           <td class="paramtype">uint32_t&#160;</td>
 344           <td class="paramname"> <em>srcALen</em>, </td>
 345         </tr>
 346         <tr>
 347           <td class="paramkey"></td>
 348           <td></td>
 349           <td class="paramtype"><a class="el" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea">q15_t</a> *&#160;</td>
 350           <td class="paramname"> <em>pSrcB</em>, </td>
 351         </tr>
 352         <tr>
 353           <td class="paramkey"></td>
 354           <td></td>
 355           <td class="paramtype">uint32_t&#160;</td>
 356           <td class="paramname"> <em>srcBLen</em>, </td>
 357         </tr>
 358         <tr>
 359           <td class="paramkey"></td>
 360           <td></td>
 361           <td class="paramtype"><a class="el" href="arm__math_8h.html#ab5a8fb21a5b3b983d5f54f31614052ea">q15_t</a> *&#160;</td>
 362           <td class="paramname"> <em>pDst</em>&#160;</td>
 363         </tr>
 364         <tr>
 365           <td></td>
 366           <td>)</td>
 367           <td></td><td></td>
 368         </tr>
 369       </table>
 370 </div>
 371 <div class="memdoc">
 372
 373 <p>Correlation of Q15 sequences (fast version) for Cortex-M3 and Cortex-M4. </p>
 374 <dl><dt><b>Parameters:</b></dt><dd>
 375   <table class="params">
 376     <tr><td class="paramdir">[in]</td><td class="paramname">*pSrcA</td><td>points to the first input sequence. </td></tr>
 377     <tr><td class="paramdir">[in]</td><td class="paramname">srcALen</td><td>length of the first input sequence. </td></tr>
 378     <tr><td class="paramdir">[in]</td><td class="paramname">*pSrcB</td><td>points to the second input sequence. </td></tr>
 379     <tr><td class="paramdir">[in]</td><td class="paramname">srcBLen</td><td>length of the second input sequence. </td></tr>
 380     <tr><td class="paramdir">[out]</td><td class="paramname">*pDst</td><td>points to the location where the output result is written. Length 2 * max(srcALen, srcBLen) - 1. </td></tr>
 381   </table>
 382   </dd>
 383 </dl>
 384 <dl class="return"><dt><b>Returns:</b></dt><dd>none.</dd></dl>
 385 <p><b>Scaling and Overflow Behavior:</b></p>
 386 <dl class="user"><dt><b></b></dt><dd>This fast version uses a 32-bit accumulator with 2.30 format. The accumulator maintains full precision of the intermediate multiplication results but provides only a single guard bit. There is no saturation on intermediate additions. Thus, if the accumulator overflows it wraps around and distorts the result. The input signals should be scaled down to avoid intermediate overflows. Scale down one of the inputs by 1/min(srcALen, srcBLen) to avoid overflow since a maximum of min(srcALen, srcBLen) number of additions is carried internally. The 2.30 accumulator is right shifted by 15 bits and then saturated to 1.15 format to yield the final result.</dd></dl>
 387 <dl class="user"><dt><b></b></dt><dd>See <code><a class="el" href="group___corr.html#ga5ec96b8e420d68b0e626df0812274d46" title="Correlation of Q15 sequences.">arm_correlate_q15()</a></code> for a slower implementation of this function which uses a 64-bit accumulator to avoid wrap around distortion. </dd></dl>
 388
 389 <p>Definition at line <a class="el" href="arm__correlate__fast__q15_8c_source.html#l00066">66</a> of file <a class="el" href="arm__correlate__fast__q15_8c_source.html">arm_correlate_fast_q15.c</a>.</p>
 390
 391 </div>
 392 </div>
 393 <a class="anchor" id="ga284ddcc49e4ac532d52a70d0383c5992"></a><!-- doxytag: member="arm_correlate_q7.c::arm_correlate_q7" ref="ga284ddcc49e4ac532d52a70d0383c5992" args="(q7_t *pSrcA, uint32_t srcALen, q7_t *pSrcB, uint32_t srcBLen, q7_t *pDst)" -->
 394 <div class="memitem">
 395 <div class="memproto">
 396       <table class="memname">
 397         <tr>
 398           <td class="memname">void arm_correlate_q7 </td>
 399           <td>(</td>
 400           <td class="paramtype"><a class="el" href="arm__math_8h.html#ae541b6f232c305361e9b416fc9eed263">q7_t</a> *&#160;</td>
 401           <td class="paramname"> <em>pSrcA</em>, </td>
 402         </tr>
 403         <tr>
 404           <td class="paramkey"></td>
 405           <td></td>
 406           <td class="paramtype">uint32_t&#160;</td>
 407           <td class="paramname"> <em>srcALen</em>, </td>
 408         </tr>
 409         <tr>
 410           <td class="paramkey"></td>
 411           <td></td>
 412           <td class="paramtype"><a class="el" href="arm__math_8h.html#ae541b6f232c305361e9b416fc9eed263">q7_t</a> *&#160;</td>
 413           <td class="paramname"> <em>pSrcB</em>, </td>
 414         </tr>
 415         <tr>
 416           <td class="paramkey"></td>
 417           <td></td>
 418           <td class="paramtype">uint32_t&#160;</td>
 419           <td class="paramname"> <em>srcBLen</em>, </td>
 420         </tr>
 421         <tr>
 422           <td class="paramkey"></td>
 423           <td></td>
 424           <td class="paramtype"><a class="el" href="arm__math_8h.html#ae541b6f232c305361e9b416fc9eed263">q7_t</a> *&#160;</td>
 425           <td class="paramname"> <em>pDst</em>&#160;</td>
 426         </tr>
 427         <tr>
 428           <td></td>
 429           <td>)</td>
 430           <td></td><td></td>
 431         </tr>
 432       </table>
 433 </div>
 434 <div class="memdoc">
 435
 436 <p>Correlation of Q7 sequences. </p>
 437 <dl><dt><b>Parameters:</b></dt><dd>
 438   <table class="params">
 439     <tr><td class="paramdir">[in]</td><td class="paramname">*pSrcA</td><td>points to the first input sequence. </td></tr>
 440     <tr><td class="paramdir">[in]</td><td class="paramname">srcALen</td><td>length of the first input sequence. </td></tr>
 441     <tr><td class="paramdir">[in]</td><td class="paramname">*pSrcB</td><td>points to the second input sequence. </td></tr>
 442     <tr><td class="paramdir">[in]</td><td class="paramname">srcBLen</td><td>length of the second input sequence. </td></tr>
 443     <tr><td class="paramdir">[out]</td><td class="paramname">*pDst</td><td>points to the location where the output result is written. Length 2 * max(srcALen, srcBLen) - 1. </td></tr>
 444   </table>
 445   </dd>
 446 </dl>
 447 <dl class="return"><dt><b>Returns:</b></dt><dd>none.</dd></dl>
 448 <p><b>Scaling and Overflow Behavior:</b></p>
 449 <dl class="user"><dt><b></b></dt><dd>The function is implemented using a 32-bit internal accumulator. Both the inputs are represented in 1.7 format and multiplications yield a 2.14 result. The 2.14 intermediate results are accumulated in a 32-bit accumulator in 18.14 format. This approach provides 17 guard bits and there is no risk of overflow as long as <code>max(srcALen, srcBLen)&lt;131072</code>. The 18.14 result is then truncated to 18.7 format by discarding the low 7 bits and saturated to 1.7 format. </dd></dl>
 450
 451 <p>Definition at line <a class="el" href="arm__correlate__q7_8c_source.html#l00065">65</a> of file <a class="el" href="arm__correlate__q7_8c_source.html">arm_correlate_q7.c</a>.</p>
 452
 453 </div>
 454 </div>
 455 </div>
 456 <!--- window showing the filter options -->
 457 <div id="MSearchSelectWindow"
 458      onmouseover="return searchBox.OnSearchSelectShow()"
 459      onmouseout="return searchBox.OnSearchSelectHide()"
 460      onkeydown="return searchBox.OnSearchSelectKey(event)">
 461 <a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(0)"><span class="SelectionMark">&#160;</span>All</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(1)"><span class="SelectionMark">&#160;</span>Data Structures</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(2)"><span class="SelectionMark">&#160;</span>Files</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(3)"><span class="SelectionMark">&#160;</span>Functions</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(4)"><span class="SelectionMark">&#160;</span>Variables</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(5)"><span class="SelectionMark">&#160;</span>Typedefs</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(6)"><span class="SelectionMark">&#160;</span>Enumerations</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(7)"><span class="SelectionMark">&#160;</span>Enumerator</a><a class="SelectItem" href="javascript:void(0)" onclick="searchBox.OnSelectItem(8)"><span class="SelectionMark">&#160;</span>Defines</a></div>
 462
 463 <!-- iframe showing the search results (closed by default) -->
 464 <div id="MSearchResultsWindow">
 465 <iframe src="" frameborder="0"
 466         name="MSearchResults" id="MSearchResults">
 467 </iframe>
 468 </div>
 469
 470 <hr class="footer"/><address class="footer"><small>Generated on Fri Jul 15 2011 13:16:20 for CMSIS DSP Software Library by&#160;
 471 <a href="http://www.doxygen.org/index.html">
 472 <img class="footer" src="doxygen.png" alt="doxygen"/></a> 1.7.2 </small></address>
 473 </body>
 474 </html>