<html xmlns="http://www.w3.org/1999/xhtml">
<head>
<meta charset="utf-8">
<meta name="generator" content="pdf2htmlEX">
<meta http-equiv="X-UA-Compatible" content="IE=edge,chrome=1">
<link rel="stylesheet" href="https://static.pudn.com/base/css/base.min.css">
<link rel="stylesheet" href="https://static.pudn.com/base/css/fancy.min.css">
<link rel="stylesheet" href="https://static.pudn.com/prod/directory_preview_static/626618ed4f8811599ea3a33d/raw.css">
<script src="https://static.pudn.com/base/js/compatibility.min.js"></script>
<script src="https://static.pudn.com/base/js/pdf2htmlEX.min.js"></script>
<script>
try{
pdf2htmlEX.defaultViewer = new pdf2htmlEX.Viewer({});
}catch(e){}
</script>
<title></title>
</head>
<body>
<div id="sidebar" style="display: none">
<div id="outline">
</div>
</div>
<div id="pf1" class="pf w0 h0" data-page-no="1"><div class="pc pc1 w0 h0"><img class="bi x0 y0 w1 h1" alt="" src="https://static.pudn.com/prod/directory_preview_static/626618ed4f8811599ea3a33d/bg1.jpg"><div class="c x0 y1 w2 h2"><div class="t m0 x1 h3 y2 ff1 fs0 fc0 sc0 ls0 ws0">Matlab<span class="_ _0"> </span><span class="ff2">程序设计任务书</span></div><div class="t m0 x2 h4 y3 ff2 fs1 fc0 sc0 ls0 ws0">分院</div><div class="t m0 x3 h4 y4 ff2 fs1 fc0 sc0 ls0 ws0">（系）</div><div class="t m0 x4 h5 y5 ff2 fs2 fc0 sc0 ls0 ws0">信息科学与工程</div><div class="t m0 x5 h4 y6 ff2 fs1 fc0 sc0 ls0 ws0">专业</div><div class="t m0 x6 h4 y7 ff2 fs1 fc0 sc0 ls0 ws0">学生姓名<span class="_ _1"> </span>学号</div><div class="t m0 x6 h4 y8 ff2 fs1 fc0 sc0 ls0 ws0">设计题目</div><div class="t m0 x7 h5 y9 ff2 fs2 fc0 sc0 ls0 ws0">车牌识别系统设计</div><div class="t m0 x8 h4 ya ff2 fs1 fc0 sc0 ls0 ws0">内容及要求：</div><div class="t m0 x9 h5 yb ff1 fs2 fc0 sc0 ls0 ws0"> <span class="ff2">车牌定位系统的目的在于正确获取整个图像中车牌的区域，</span></div><div class="t m0 xa h5 yc ff1 fs2 fc0 sc0 ls0 ws0"> <span class="ff2">并识别出车牌号。通过设计实现车牌识别系统，能够提高学生</span></div><div class="t m0 xb h5 yd ff2 fs2 fc0 sc0 ls0 ws0">分析问题和解决问题的能力，还能培养一定的科研能力。</div><div class="t m0 xb h5 ye ff1 fs2 fc0 sc0 ls0 ws0">1.<span class="ff2">牌照识别系统应包括车辆检测、图像采集、牌照识别等几</span></div><div class="t m0 xc h5 yf ff2 fs2 fc0 sc0 ls0 ws0">部分。</div><div class="t m0 xb h5 y10 ff1 fs2 fc0 sc0 ls0 ws0">2.<span class="ff2">当车辆检测部分检测到车辆到达时</span>,<span class="ff2">触发图像采集单元，采</span></div><div class="t m0 xc h5 y11 ff2 fs2 fc0 sc0 ls0 ws0">集当前的视频图像。</div><div class="t m0 xb h5 y12 ff1 fs2 fc0 sc0 ls0 ws0">3.<span class="ff2">牌照识别单元对图像进行处理，定位出牌照位置，再将牌</span></div><div class="t m0 xc h5 y13 ff2 fs2 fc0 sc0 ls0 ws0">照中的字符分割出来进行识别，然后组成牌照号码输出。</div><div class="t m0 x8 h4 y14 ff2 fs1 fc0 sc0 ls0 ws0">进度安排：</div><div class="t m0 xd h5 y15 ff1 fs2 fc0 sc0 ls0 ws0">19<span class="_ _2"> </span><span class="ff2">周：</span>Matlab<span class="_ _2"> </span><span class="ff2">环境熟悉与基础知识学习</span></div><div class="t m0 xd h5 y16 ff1 fs2 fc0 sc0 ls0 ws0">19<span class="_ _2"> </span><span class="ff2">周：课程设计选题与题目分析</span></div><div class="t m0 xd h5 y17 ff1 fs2 fc0 sc0 ls0 ws0">20<span class="_ _2"> </span><span class="ff2">周：程序设计编程实现</span></div><div class="t m0 xd h5 y18 ff1 fs2 fc0 sc0 ls0 ws0">20<span class="_ _2"> </span><span class="ff2">周：课程设计验收与答辩</span></div><div class="t m0 x8 h4 y19 ff2 fs1 fc0 sc0 ls0 ws0">指导教师（签字）：<span class="_ _3"> </span>学院院长（签字）：</div></div><div class="t m0 xe h6 y1a ff3 fs3 fc0 sc0 ls0 ws0">1</div></div><div class="pi" data-data='{"ctm":[1.611850,0.000000,0.000000,1.611850,0.000000,0.000000]}'></div></div>
</body>
</html>
<div id="pf2" class="pf w0 h0" data-page-no="2"><div class="pc pc2 w0 h0"><img class="bi x0 y0 w1 h1" alt="" src="https://static.pudn.com/prod/directory_preview_static/626618ed4f8811599ea3a33d/bg2.jpg"><div class="c x0 y1 w2 h2"><div class="t m0 xf h4 y1b ff2 fs1 fc0 sc0 ls0 ws0">年 月 日<span class="_ _4"> </span>年 月 日</div><div class="t m0 x10 h7 y1c ff2 fs4 fc0 sc1 ls0 ws0">目<span class="ff4 sc0"> </span>录</div><div class="t m0 xd h4 y1d ff2 fs1 fc0 sc0 ls0 ws0">一．课程设计目的……………………………………………<span class="ff1">3</span></div><div class="t m0 x11 h4 y1e ff2 fs1 fc0 sc0 ls0 ws0">二．设计原理…………………………………………………<span class="ff1">3</span></div><div class="t m0 x11 h4 y1f ff2 fs1 fc0 sc0 ls0 ws0">三．详细设计步骤……………………………………………<span class="ff1">3</span></div><div class="t m0 x11 h4 y20 ff2 fs1 fc0 sc0 ls0 ws0">四<span class="ff1">. </span>设计结果及分析…………………………………………<span class="ff1">18</span></div><div class="t m0 x11 h4 y21 ff2 fs1 fc0 sc0 ls0 ws0">五<span class="ff1">. </span>总结………………………………………………………<span class="ff1">19</span></div><div class="t m0 x12 h8 y22 ff1 fs1 fc0 sc0 ls0 ws0"> </div><div class="t m0 x11 h4 y23 ff2 fs1 fc0 sc0 ls0 ws0">六<span class="ff1">. </span>设计体会…………………………………………………<span class="ff1">20</span></div><div class="t m0 x12 h8 y24 ff1 fs1 fc0 sc0 ls0 ws0"> </div><div class="t m0 x11 h4 y25 ff2 fs1 fc0 sc0 ls0 ws0">七<span class="ff1">. </span>参考文献…………………………………………………<span class="ff1">21</span></div></div><div class="t m0 xe h6 y1a ff3 fs3 fc0 sc0 ls0 ws0">2</div></div><div class="pi" data-data='{"ctm":[1.611850,0.000000,0.000000,1.611850,0.000000,0.000000]}'></div></div>
<div id="pf3" class="pf w0 h0" data-page-no="3"><div class="pc pc3 w0 h0"><img class="bi x0 y0 w1 h1" alt="" src="https://static.pudn.com/prod/directory_preview_static/626618ed4f8811599ea3a33d/bg3.jpg"><div class="c x0 y1 w2 h2"><div class="t m0 x13 h3 y26 ff2 fs0 fc0 sc0 ls0 ws0">一、课程设计目的</div><div class="t m0 x14 h5 y27 ff2 fs2 fc0 sc0 ls0 ws0">车牌<span class="_ _5"></span>定位<span class="_ _5"></span>系统<span class="_ _5"></span>的目<span class="_ _5"></span>的在<span class="_ _5"></span>于正<span class="_ _5"></span>确获<span class="_ _5"></span>取整<span class="_ _5"></span>个图<span class="_ _5"></span>像中<span class="_ _5"></span>车牌<span class="_ _5"></span>的区<span class="_ _5"></span>域，<span class="_ _5"></span>并识<span class="_ _5"></span>别出<span class="_ _5"></span>车牌<span class="_ _5"></span>号。<span class="_ _5"></span>通</div><div class="t m0 x15 h5 y28 ff2 fs2 fc0 sc0 ls0 ws0">过设<span class="_ _5"></span>计实<span class="_ _5"></span>现车<span class="_ _5"></span>牌识<span class="_ _5"></span>别系<span class="_ _5"></span>统，<span class="_ _5"></span>能够<span class="_ _5"></span>提高<span class="_ _5"></span>学生<span class="_ _5"></span>分析<span class="_ _5"></span>问题<span class="_ _5"></span>和解<span class="_ _5"></span>决问<span class="_ _5"></span>题的<span class="_ _5"></span>能力<span class="_ _5"></span>，还<span class="_ _5"></span>能培<span class="_ _5"></span>养一<span class="_ _5"></span>定</div><div class="t m0 x15 h5 y29 ff2 fs2 fc0 sc0 ls0 ws0">的科研能力。</div><div class="t m0 x14 h9 y2a ff1 fs2 fc0 sc0 ls0 ws0"> <span class="_ _5"></span> </div><div class="t m0 x13 h3 y2b ff2 fs0 fc0 sc0 ls0 ws0">二、设计原理<span class="ff1">:</span></div><div class="t m0 x16 h5 y2c ff2 fs2 fc0 sc0 ls0 ws0">牌照自动识别是一项利用车辆的动态视频或静态图像进行牌照号码、牌照颜色自动</div><div class="t m0 x13 h5 y2d ff2 fs2 fc0 sc0 ls0 ws0">识别的模式识别技术。其硬件基础一般包括触发设备、摄像设备、照明设备、图像采集</div><div class="t m0 x13 h5 y2e ff2 fs2 fc0 sc0 ls0 ws0">设备、识别车牌号码的处理机等，其软件核心包括车牌定位算法、车牌字符分割算法和</div><div class="t m0 x13 h5 y2f ff2 fs2 fc0 sc0 ls0 ws0">光学字符识别算法等。某些牌照识别系统还具有通过视频图像判断车辆驶入视野的功能</div><div class="t m0 x13 h5 y30 ff2 fs2 fc0 sc0 ls0 ws0">称之为视频车辆检测。一个完整的牌照识别系统应包括车辆检测、图像采集、牌照识别</div><div class="t m0 x13 h5 y31 ff2 fs2 fc0 sc0 ls0 ws0">等几部分。当车辆检测部分检测到车辆到达时触发图像采集单元，采集当前的视频图像。</div><div class="t m0 x13 h5 y32 ff2 fs2 fc0 sc0 ls0 ws0">牌照识别单元对图像进行处理，定位出牌照位置，再将牌照中的字符分割出来进行识别，</div><div class="t m0 x13 h5 y33 ff2 fs2 fc0 sc0 ls0 ws0">然后组成牌照号码输出。</div><div class="t m0 x16 h9 y34 ff1 fs2 fc0 sc0 ls0 ws0"> <span class="_ _5"></span> </div><div class="t m0 x13 h3 y35 ff2 fs0 fc0 sc0 ls0 ws0">三、详细设计步骤<span class="ff1">:</span></div></div><div class="t m0 xe h6 y1a ff3 fs3 fc0 sc0 ls0 ws0">3</div></div><div class="pi" data-data='{"ctm":[1.611850,0.000000,0.000000,1.611850,0.000000,0.000000]}'></div></div>
<div id="pf4" class="pf w0 h0" data-page-no="4"><div class="pc pc4 w0 h0"><img class="bi x0 y0 w1 h1" alt="" src="https://static.pudn.com/prod/directory_preview_static/626618ed4f8811599ea3a33d/bg4.jpg"><div class="c x0 y1 w2 h2"><div class="t m0 x13 h4 y36 ff1 fs1 fc0 sc0 ls0 ws0">1. <span class="ff2">提出总体设计方案</span>:</div><div class="t m0 x13 h5 y37 ff2 fs2 fc0 sc0 ls0 ws0">牌照号码、颜色识别 </div><div class="t m0 x13 h5 y38 ff2 fs2 fc0 sc0 ls0 ws0">为了进行牌照识别，需要以下几个基本的步骤：</div><div class="t m0 x13 h5 y39 ff1 fs2 fc0 sc0 ls0 ws0">a.<span class="ff2">牌照定位，定位图片中的牌照位置；</span></div><div class="t m0 x13 h5 y3a ff1 fs2 fc0 sc0 ls0 ws0">b.<span class="ff2">牌照字符分割，把牌照中的字符分割出来；</span></div><div class="t m0 x13 h5 y3b ff1 fs2 fc0 sc0 ls0 ws0">c.<span class="ff2">牌照字符识别，把分割好的字符进行识别，最终组成牌照号码。</span></div><div class="t m0 x13 h5 y3c ff2 fs2 fc0 sc0 ls0 ws0">　　牌照识别过程中，牌照颜色的识别依据算法不同，可能在上述不同步骤实现，通常</div><div class="t m0 x13 h5 y3d ff2 fs2 fc0 sc0 ls0 ws0">与牌照识别互相配合、互相验证。</div><div class="t m0 x13 h5 y3e ff1 fs2 fc0 sc0 ls0 ws0">(1<span class="ff2">）牌照定位</span>:</div><div class="t m0 x16 h5 y3f ff2 fs2 fc0 sc0 ls0 ws0">自然环境下，汽车图像背景复杂、光照不均匀，如何在自然背景中准确地确定牌照</div><div class="t m0 x13 h5 y40 ff2 fs2 fc0 sc0 ls0 ws0">区域是整个识别过程的关键。首先对采集到的视频图像进行<span class="ff5">大范围</span>相关<span class="ff5">搜索</span>，<span class="ff5">找</span>到符合</div><div class="t m0 x13 h5 y41 ff2 fs2 fc0 sc0 ls0 ws0">汽车牌照<span class="ff5">特征</span>的<span class="ff5">若干</span>区域<span class="ff5">作</span>为<span class="ff5">候</span>选区，然后对<span class="ff5">这</span>些<span class="ff5">侯</span>选区域<span class="ff5">做</span>进一步分析、<span class="ff5">评</span>判，最后</div><div class="t m0 x13 h5 y42 ff2 fs2 fc0 sc0 ls0 ws0">选定一个最<span class="ff5">佳</span>的区域<span class="ff5">作</span>为牌照区域，并将其<span class="ff5">从</span>图<span class="ff5">象</span>中分割出来。</div><div class="t m0 x16 h5 y43 ff5 fs2 fc0 sc0 ls0 ws0">流<span class="ff2">程图<span class="ff1">:</span></span></div><div class="t m0 x17 ha y44 ff1 fs5 fc0 sc0 ls0 ws0"><span class="fc1 sc0"> </span></div></div><div class="t m0 xe h6 y1a ff3 fs3 fc0 sc0 ls0 ws0">4</div><div class="t m0 x18 h5 y45 ff2 fs2 fc0 sc0 ls0 ws0">导入原<span class="ff5">始</span></div><div class="t m0 x19 h5 y46 ff2 fs2 fc0 sc0 ls0 ws0">图像</div><div class="t m0 x1a h5 y45 ff2 fs2 fc0 sc0 ls0 ws0">图像<span class="ff5">预</span>处理<span class="ff5">增</span></div><div class="t m0 x1b h5 y46 ff5 fs2 fc0 sc0 ls0 ws0">强效<span class="ff2">果图像</span></div><div class="t m0 x1c h5 y45 ff5 fs2 fc0 sc0 ls0 ws0">边缘<span class="ff2">提取<span class="_ _6"> </span>车牌定位<span class="_ _7"></span>对图像<span class="ff5">开</span></span></div><div class="t m0 x1d h5 y46 ff5 fs2 fc0 sc0 ls0 ws0">闭运<span class="ff2">算</span></div></div><div class="pi" data-data='{"ctm":[1.611850,0.000000,0.000000,1.611850,0.000000,0.000000]}'></div></div>
<div id="pf5" class="pf w0 h0" data-page-no="5"><div class="pc pc5 w0 h0"><img class="bi x0 y0 w1 h1" alt="" src="https://static.pudn.com/prod/directory_preview_static/626618ed4f8811599ea3a33d/bg5.jpg"><div class="c x0 y1 w2 h2"><div class="t m0 x13 h5 y47 ff2 fs2 fc0 sc0 ls0 ws0">（<span class="ff1">2</span>）牌照字符分割 <span class="ff1">:</span></div><div class="t m0 x16 h5 y48 ff2 fs2 fc0 sc0 ls0 ws0">完成牌照区域的定位后，再将牌照区域分割成单个字符，然后进行识别。字符分割</div><div class="t m0 x13 h5 y49 ff2 fs2 fc0 sc0 ls0 ws0">一般采用<span class="ff5">垂直投影</span>法。<span class="ff5">由</span>于字符在<span class="ff5">垂直</span>方<span class="ff5">向</span>上的<span class="ff5">投影必</span>然在字符<span class="ff5">间</span>或字符内的<span class="ff5">间隙</span>处取</div><div class="t m0 x13 h5 y4a ff5 fs2 fc0 sc0 ls0 ws0">得局<span class="ff2">部最</span>小值<span class="ff2">的</span>附近<span class="ff2">，并</span>且这<span class="ff2">个位置应</span>满足<span class="ff2">牌照的字符书</span>写格<span class="ff2">式、字符、</span>尺寸限制<span class="ff2">和一</span></div><div class="t m0 x13 h5 y4b ff2 fs2 fc0 sc0 ls0 ws0">些其<span class="ff5">他条</span>件。利用<span class="ff5">垂直投影</span>法对复杂环境下的汽车图像中的字符分割有<span class="ff5">较</span>好的<span class="ff5">效</span>果。</div><div class="t m0 x17 ha y4c ff1 fs5 fc0 sc0 ls0 ws0"><span class="fc1 sc0"> </span></div><div class="t m0 x13 h5 y4d ff2 fs2 fc0 sc0 ls0 ws0">（<span class="ff1">3</span>）牌照字符识别<span class="sc1"> </span><span class="ff6">:</span></div><div class="t m0 x1e h5 y4e ff2 fs2 fc0 sc0 ls0 ws0">字符识别方法目前<span class="ff5">主</span>要有基于模<span class="ff5">板匹</span>配算法和基于<span class="ff5">人</span>工<span class="ff5">神经网络</span>算法。基于模<span class="ff5">板匹</span></div><div class="t m0 x13 h5 y4f ff2 fs2 fc0 sc0 ls0 ws0">配算法首先将分割后的字符二<span class="ff5">值化<span class="ff1">,</span></span>并将其<span class="ff5">尺寸大小缩放</span>为字符<span class="ff5">数</span>据<span class="ff5">库</span>中模<span class="ff5">板</span>的<span class="ff5">大小</span>，</div><div class="t m0 x13 h5 y50 ff2 fs2 fc0 sc0 ls0 ws0">然后与<span class="ff5">所</span>有的模<span class="ff5">板</span>进行<span class="ff5">匹</span>配，最后选最<span class="ff5">佳匹</span>配<span class="ff5">作</span>为结果。基于<span class="ff5">人</span>工<span class="ff5">神经</span>元<span class="ff5">网络</span>的算法有</div><div class="t m0 x13 h5 y51 ff5 fs2 fc0 sc0 ls0 ws0">两种<span class="ff2">：一</span>种<span class="ff2">是先对</span>待<span class="ff2">识别字符进行</span>特征<span class="ff2">提取，然后用</span>所<span class="ff2">获</span>得特征<span class="ff2">来</span>训练神经网络<span class="ff2">分配</span>器<span class="ff2">；</span></div><div class="t m0 x13 h5 y52 ff5 fs2 fc0 sc0 ls0 ws0">另<span class="ff2">一</span>种<span class="ff2">方法是</span>直接<span class="ff2">把</span>待<span class="ff2">处理图像输入</span>网络<span class="ff2">，</span>由网络<span class="ff2">自动实现</span>特征<span class="ff2">提取</span>直至<span class="ff2">识别出结果。</span></div><div class="t m0 x13 h5 y53 ff2 fs2 fc0 sc0 ls0 ws0">实<span class="ff5">际</span>应用中，牌照识别系统的识别<span class="ff5">率</span>与牌照<span class="ff5">质量</span>和<span class="ff5">拍</span>摄<span class="ff5">质量密切</span>相关。牌照<span class="ff5">质量</span>会<span class="ff5">受</span>到</div><div class="t m0 x13 h5 y54 ff5 fs2 fc0 sc0 ls0 ws0">各种因素<span class="ff2">的</span>影响<span class="ff2">，如生</span>锈<span class="ff2">、</span>污损<span class="ff2">、</span>油漆剥落<span class="ff2">、字体</span>褪<span class="ff2">色、牌照</span>被遮挡<span class="ff2">、牌照</span>倾斜<span class="ff2">、高</span>亮</div><div class="t m0 x13 h5 y55 ff5 fs2 fc0 sc0 ls0 ws0">反<span class="ff2">光、</span>多<span class="ff2">牌照、</span>假<span class="ff2">牌照等等；实</span>际拍<span class="ff2">摄过程</span>也<span class="ff2">会</span>受<span class="ff2">到环境</span>亮<span class="ff2">度、</span>拍<span class="ff2">摄</span>亮<span class="ff2">度、车辆</span>速<span class="ff2">度等</span></div><div class="t m0 x13 h5 y56 ff2 fs2 fc0 sc0 ls0 ws0">等<span class="ff5">因素</span>的<span class="ff5">影响</span>。<span class="ff5">这</span>些<span class="ff5">影响因素</span>不同程度上<span class="ff5">降低</span>了牌照识别的识别<span class="ff5">率</span>，<span class="ff5">也</span>正是牌照识别系</div><div class="t m0 x13 h5 y57 ff2 fs2 fc0 sc0 ls0 ws0">统的<span class="ff5">困难</span>和<span class="ff5">挑战所</span>在。为了提高识别<span class="ff5">率</span>，<span class="ff5">除</span>了不断的完<span class="ff5">善</span>识别算法，还应<span class="ff5">该想办</span>法<span class="ff5">克服</span></div><div class="t m0 x13 h5 y58 ff5 fs2 fc0 sc0 ls0 ws0">各种<span class="ff2">光照</span>条<span class="ff2">件，</span>使<span class="ff2">采集到的图像最利于识别。</span></div><div class="t m0 x13 h4 y59 ff1 fs1 fc0 sc0 ls0 ws0"><span class="fc1 sc0">2.</span><span class="fc1 sc0"> </span><span class="ff5"><span class="fc1 sc0">各</span><span class="ff2"><span class="fc1 sc0">模</span></span><span class="fc1 sc0">块</span><span class="ff2"><span class="fc1 sc0">的</span>实现</span></span>:</div><div class="t m0 x13 h5 y5a ff1 fs2 fc0 sc0 ls0 ws0">2.1<span class="_ _2"> </span><span class="ff2">输入<span class="ff5">待</span>处理的原<span class="ff5">始</span>图像</span>:</div><div class="t m0 x13 h9 y5b ff1 fs2 fc0 sc0 ls0 ws0">clear ;</div><div class="t m0 x13 h9 y5c ff1 fs2 fc0 sc0 ls0 ws0">close all;</div><div class="t m0 x13 h5 y5d ff1 fs2 fc0 sc0 ls0 ws0">%Step1 <span class="ff2">获取图像 <span class="ff5">装</span>入<span class="ff5">待</span>处理<span class="ff5">彩</span>色图像并<span class="ff5">显示</span>原<span class="ff5">始</span>图像</span></div><div class="t m0 x13 h5 y5e ff1 fs2 fc0 sc0 ls0 ws0">Scolor = imread('3.jpg');%imread<span class="_ _8"> </span><span class="ff5">函数读<span class="ff2">取图像文件</span></span></div></div><div class="t m0 xe h6 y1a ff3 fs3 fc0 sc0 ls0 ws0">5</div><div class="t m0 x1f h5 y5f ff5 fs2 fc0 sc0 ls0 ws0">按左右宽<span class="ff2">度</span></div><div class="t m0 x1f h5 y60 ff5 fs2 fc0 sc0 ls0 ws0">切<span class="ff2">割出字符</span></div><div class="t m0 x18 h5 y5f ff2 fs2 fc0 sc0 ls0 ws0">计<span class="_ _9"> </span>算<span class="_ _9"> </span><span class="ff5">水<span class="_ _9"> </span>平<span class="_ _9"> </span>投<span class="_ _9"> </span>影<span class="_ _9"> </span></span>进</div><div class="t m0 x18 h5 y60 ff2 fs2 fc0 sc0 ls0 ws0">行车牌<span class="ff5">水平校</span>正</div><div class="t m0 x20 h5 y5f ff5 fs2 fc0 sc0 ls0 ws0">去掉<span class="ff2">车牌</span></div><div class="t m0 x21 h5 y60 ff2 fs2 fc0 sc0 ls0 ws0">的<span class="ff5">框架</span></div><div class="t m0 x22 h5 y5f ff2 fs2 fc0 sc0 ls0 ws0">分析<span class="ff5">垂直投影找</span>到<span class="ff5">每</span></div><div class="t m0 x23 h5 y60 ff2 fs2 fc0 sc0 ls0 ws0">个字符中心位置</div><div class="t m0 x24 h5 y61 ff5 fs2 fc0 sc0 ls0 ws0">切<span class="ff2">割出的字</span></div><div class="t m0 x24 h5 y62 ff2 fs2 fc0 sc0 ls0 ws0">符<span class="ff5">送</span>入<span class="ff5">库</span>中</div><div class="t m0 x25 h5 y61 ff2 fs2 fc0 sc0 ls0 ws0">字符依<span class="ff5">次</span>分析<span class="ff5">显示误差</span></div><div class="t m0 x26 h5 y62 ff2 fs2 fc0 sc0 ls0 ws0">最<span class="ff5">小</span>的图片名字</div><div class="t m0 x27 h5 y61 ff2 fs2 fc0 sc0 ls0 ws0">与<span class="ff5">数</span>据<span class="ff5">库</span>的</div><div class="t m0 x28 h5 y62 ff2 fs2 fc0 sc0 ls0 ws0">图片相<span class="ff5">减</span></div><div class="t m0 x29 h5 y61 ff2 fs2 fc0 sc0 ls0 ws0">分<span class="_ _a"></span>析<span class="_ _a"></span>之<span class="_ _a"></span><span class="ff5">差<span class="_ _a"></span></span>最<span class="_ _a"></span><span class="ff5">小</span></div><div class="t m0 x29 h5 y62 ff2 fs2 fc0 sc0 ls0 ws0">的图片是<span class="ff5">哪张</span></div></div><div class="pi" data-data='{"ctm":[1.611850,0.000000,0.000000,1.611850,0.000000,0.000000]}'></div></div>
