Infrared-normalization.rar_红外_infrared_模式识别(视觉/语音等)

<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/622b578715da9b288bc328e7/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/622b578715da9b288bc328e7/bg1.jpg"><div class="t m0 x1 h2 y1 ff1 fs0 fc0 sc0 ls0 ws0">13 </div><div class="t m0 x2 h3 y2 ff1 fs1 fc0 sc0 ls1 ws1">Blood Perfusion Models for </div><div class="t m0 x3 h3 y3 ff1 fs1 fc0 sc0 ls1 ws0">Infrared Face Recognition </div><div class="t m0 x4 h4 y4 ff2 fs2 fc0 sc0 ls2 ws2">Shiqian Wu, Zhi-Jun Fang, Zhi-Hua Xie and Wei Liang </div><div class="t m0 x5 h5 y5 ff3 fs3 fc0 sc0 ls3 ws3">School of information technology, <span class="_ _0"></span>Jiangx<span class="ls4 ws4">i University of Finance and Economics, </span></div><div class="t m0 x6 h5 y6 ff3 fs3 fc0 sc0 ls5 ws0">China </div><div class="t m0 x7 h6 y7 ff1 fs3 fc0 sc0 ls6 ws5">1. Introduction </div><div class="t m0 x7 h7 y8 ff2 fs4 fc0 sc0 ls7 ws6">Infrared (IR) technology has traditionally been<span class="ls0 ws7"> applied to military u<span class="_ _0"></span>se and remote sensing. </span></div><div class="t m0 x7 h7 y9 ff2 fs4 fc0 sc0 ls7 ws8">During the last two decades, the cost of IR <span class="ls8 ws9">cameras (especially un<span class="_ _0"></span>cooled imagers) has been </span></div><div class="t m0 x7 h7 ya ff2 fs4 fc0 sc0 ls9 wsa">significantly reduced with the developmen<span class="_ _0"></span><span class="lsa wsb">t of CCD technology, and therefore civil </span></div><div class="t m0 x7 h7 yb ff2 fs4 fc0 sc0 ls0 ws7">applications have increased constantly due to <span class="wsc">its un<span class="_ _0"></span>ique features. One of such applications </span></div><div class="t m0 x7 h7 yc ff2 fs4 fc0 sc0 lsb wsd">is IR face recognition (Prokoski et al., <span class="lsa wse">19<span class="_ _0"></span>92, Prokoski, 2000, Kong et al., 2005). The </span></div><div class="t m0 x7 h7 yd ff2 fs4 fc0 sc0 lsc wsf">fundamentals behind it are, as indicated by Ko<span class="lsd ws10">ng et al (Kong et al., 2005) that IR images are </span></div><div class="t m0 x7 h7 ye ff2 fs4 fc0 sc0 lsc ws11">independent of external illumination. While<span class="ls9"> vi<span class="_ _0"></span>sible images represent the reflectance </span></div><div class="t m0 x7 h7 yf ff2 fs4 fc0 sc0 lsd ws12">information of the face surface, IR face imag<span class="lse ws13">es contain more fundamental information about </span></div><div class="t m0 x7 h7 y10 ff2 fs4 fc0 sc0 ls7 ws14">faces themselves, such <span class="lsa ws15">as anatomical information (Prokosk</span><span class="ws16">i, et al., 1992, <span class="_ _0"></span>Prokoski, 2000); the </span></div><div class="t m0 x7 h7 y11 ff2 fs4 fc0 sc0 ls7 ws17">thermal characteristics of faces with variat<span class="_ _0"></span>io<span class="lse ws18">ns in facial expression and make-up remain </span></div><div class="t m0 x7 h7 y12 ff2 fs4 fc0 sc0 lsf ws19">nearly invariant (Socolinsky & Selinger, 2002) an<span class="ls8 ws1a">d the tasks of face detection, lo<span class="_ _0"></span>calization, </span></div><div class="t m0 x7 h7 y13 ff2 fs4 fc0 sc0 lsd ws1b">and segmentation are relatively easier and more<span class="ls7 ws1c"> reliable than those in visible <span class="_ _0"></span>images (Kong </span></div><div class="t m0 x7 h7 y14 ff2 fs4 fc0 sc0 ls9 ws1d">et al., 2005). It has been pointed by Prokoski <span class="lsf ws1e">et al. (<span class="_ _0"></span>Prokoski et al., 1992) that humans are </span></div><div class="t m0 x7 h7 y15 ff2 fs4 fc0 sc0 ls10 ws1f">homoiotherm and hence capable of maintaining constant temperature under different </div><div class="t m0 x7 h7 y16 ff2 fs4 fc0 sc0 lsc ws20">surroundings. The thermal imag<span class="_ _0"></span><span class="ws21">es collected over 20 years have demonstrated that t<span class="_ _0"></span>he </span></div><div class="t m0 x7 h7 y17 ff2 fs4 fc0 sc0 lsa ws22">thermal measurements of individuals are high<span class="ls11 ws23">ly repeatable under the same conditions. </span></div><div class="t m0 x7 h7 y18 ff2 fs4 fc0 sc0 ls12 ws24">Furthermore, as discuss<span class="_ _0"></span>ed by<span class="ls13 ws25"> Prokoski (Prokoski, 2000), a facia<span class="_ _0"></span>l thermal pattern is </span></div><div class="t m0 x7 h7 y19 ff2 fs4 fc0 sc0 lsb ws26">determined by the vascular structure of each<span class="_ _0"></span><span class="ls11 ws27"> face, which is <span class="_ _0"></span>irreproducible and un<span class="_ _0"></span>ique. </span></div><div class="t m0 x7 h7 y1a ff2 fs4 fc0 sc0 lsb ws28">Based on the assumption that facial therma<span class="ls14 ws29">l patterns are determined by blood vessels </span></div><div class="t m0 x7 h7 y1b ff2 fs4 fc0 sc0 lsd ws2a">transporting warm blood, Prokos<span class="ls9 ws2b">ki tried to extract the blood <span class="ls15 ws2c">vessel minutiae (Prokoski<span class="_ _1"></span>, </span></span></div><div class="t m0 x7 h7 y1c ff2 fs4 fc0 sc0 ls10 ws2d">2001) or vascular network (Buddharaju et al., <span class="lsf ws2e">2004, Buddharaju et al., 2005) as the facial </span></div><div class="t m0 x7 h7 y1d ff2 fs4 fc0 sc0 ls14 ws2f">features for recognition. The basic <span class="_ _0"></span>idea is to extract su<span class="_ _0"></span>ch features using image segm<span class="_ _0"></span>entation. </div><div class="t m0 x7 h7 y1e ff2 fs4 fc0 sc0 lsf ws30">It has been indicated by Guyt<span class="ls13 ws31">on & Hall (Guyton & Hall, 1996) th<span class="lsa ws32">at the average diameter of </span></span></div><div class="t m0 x7 h7 y1f ff2 fs4 fc0 sc0 lse ws33">blood vessels is around 10~15<span class="ff4 ls10 ws0">μ<span class="ff2 ls16">m<span class="_ _2"></span>,<span class="_ _2"></span> w<span class="_ _2"></span>h<span class="_ _2"></span>i<span class="_ _3"></span>c<span class="_ _2"></span>h<span class="_ _2"></span> i<span class="_ _2"></span>s<span class="_ _3"></span> t<span class="_ _2"></span>o<span class="_ _2"></span>o<span class="_ _2"></span> s<span class="_ _2"></span>m<span class="_ _3"></span>a<span class="_ _2"></span>l<span class="_ _2"></span>l<span class="_ _2"></span> t<span class="_ _2"></span>o<span class="_ _3"></span> b<span class="_ _2"></span>e<span class="_ _2"></span> d<span class="_ _2"></span>e<span class="_ _2"></span>t<span class="_ _3"></span>e<span class="_ _2"></span>c<span class="_ _2"></span>t<span class="_ _2"></span>e<span class="_ _3"></span>d<span class="_ _2"></span> b<span class="_ _2"></span>y<span class="_ _2"></span> c<span class="_ _2"></span>u<span class="_ _2"></span>r<span class="_ _2"></span>r<span class="_ _3"></span>e<span class="_ _2"></span>n<span class="_ _2"></span>t<span class="_ _2"></span> I<span class="_ _3"></span>R<span class="_ _2"></span> c<span class="_ _2"></span>a<span class="_ _2"></span>m<span class="_ _3"></span>e<span class="_ _2"></span>r<span class="_ _2"></span>a<span class="_ _2"></span>s<span class="_ _3"></span> </span></span></div><div class="t m0 x7 h7 y20 ff2 fs4 fc0 sc0 ls14 ws34">(limited by the spatial resolut<span class="_ _0"></span>ion); the skin di<span class="ls9 ws35">rectly above a blood vessel is <span class="_ _0"></span>on average 0.1°C </span></div><div class="t m0 x7 h7 y21 ff2 fs4 fc0 sc0 ls17 ws36">warmer than the adjacent skin, which is beyond<span class="ls13 ws37"> the thermal accuracy of current IR cameras. </span></div><div class="t m0 x7 h7 y22 ff2 fs4 fc0 sc0 lse ws38">The methods using image segmentation in <span class="lsb ws39">(Prokoski, 2001, Buddharaju et al., 2004, </span></div><div class="t m0 x7 h7 y23 ff2 fs4 fc0 sc0 ls9 ws3a">Buddharaju et al., 2005) are heuristic, and it <span class="lse ws3b">still remains a big challenge to capture the </span></div><div class="t m0 x7 h7 y24 ff2 fs4 fc0 sc0 ls12 ws3c">pattern of blood vessels<span class="_ _0"></span> on each face. <span class="_ _0"></span> </div><div class="t m0 x7 h7 y25 ff2 fs4 fc0 sc0 lsd ws3d">On the other hand, the phenomenon of <span class="ls10 ws3e">“homoiotherm” due to human temperature </span></div><div class="t m0 x7 h7 y26 ff2 fs4 fc0 sc0 ls9 ws3f">regulation has led to the direct use of thermograms for recognition (Wilder et al., 1996, </div><div class="c x8 y27 w2 h8"><div class="t m1 x9 h9 y28 ff5 fs5 fc0 sc0 ls10 ws0">Open Access Database www.intechweb.org</div></div><div class="c xa y29 w3 ha"><div class="t m2 xb hb y2a ff6 fs6 fc0 sc0 ls10 ws0">Source: Recent Advances in Face Recognition, Book edited by: Kresimir Delac, Mislav Grgic and Marian Stewart Bartlett, </div><div class="t m2 xc hb y2b ff6 fs6 fc0 sc0 ls10 ws0">ISBN 978-953-7619-34-3, pp. 236, December 2008, I-Tech, Vienna, Austria</div></div></div><div class="pi" data-data='{"ctm":[1.991701,0.000000,0.000000,1.991701,0.000000,0.000000]}'></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/622b578715da9b288bc328e7/bg2.jpg"><div class="t m0 xd hc y2c ff7 fs7 fc0 sc0 ls18 ws40"> <span class="_ _4"> </span>Recent Advances in Fa<span class="_ _1"></span>ce Recogniti<span class="_ _1"></span>on </div><div class="t m0 x7 hd y2d ff8 fs8 fc0 sc0 ls10 ws0"> </div><div class="t m0 x7 he y2e ff7 fs4 fc0 sc0 ls19 ws0">184 </div><div class="t m0 x7 h7 y2f ff2 fs4 fc0 sc0 lsb ws41">Socolinsky & Selinger, 2002, Wu et al., 2003, Chen <span class="ls0 ws42">et al., 2005). Wilder et <span class="ls13 ws43">al. (Wilder et al., </span></span></div><div class="t m0 x7 h7 y30 ff2 fs4 fc0 sc0 lsd ws44">1996) used three different feature-extraction <span class="ls10 ws45">and decision-making algorithms for test. The </span></div><div class="t m0 x7 h7 y31 ff2 fs4 fc0 sc0 ls14 ws46">recognition results revealed th<span class="_ _0"></span>at both visible and IR ima<span class="_ _0"></span>geries perform similarly across </div><div class="t m0 x7 h7 y32 ff2 fs4 fc0 sc0 ls8 ws47">algorithms. The real-time IR<span class="_ _0"></span> face recognition sy<span class="_ _0"></span>stem developed by Wu et a<span class="_ _0"></span>l. (Wu et al., 2003) </div><div class="t m0 x7 h7 y33 ff2 fs4 fc0 sc0 ls13 ws48">achieves good performance. Especially, Soco<span class="_ _0"></span>li<span class="ls1a ws49">nsky & Selinger (Socolinsky & Selinger, 2002)<span class="_ _1"></span> </span></div><div class="t m0 x7 h7 y34 ff2 fs4 fc0 sc0 ls9 ws4a">simultaneously registered the IR and visible <span class="ls7 ws4b">im<span class="_ _0"></span>ages of each candidate under controll<span class="_ _0"></span>ed </span></div><div class="t m0 x7 h7 y35 ff2 fs4 fc0 sc0 lse ws4c">conditions. It has been concluded from their ex<span class="ls9 ws4d">perimental results that (<span class="_ _0"></span>1) variations of IR </span></div><div class="t m0 x7 h7 y36 ff2 fs4 fc0 sc0 ls9 ws4e">images are less than those of <span class="ls7 ws4f">visible images; (2) IR images <span class="ls1b ws50">are less sensitive to facial </span></span></div><div class="t m0 x7 h7 y37 ff2 fs4 fc0 sc0 ls7 ws51">expression changes. The experiments condu<span class="_ _0"></span>c<span class="lsa ws52">ted with the common methods like principle </span></div><div class="t m0 x7 h7 y38 ff2 fs4 fc0 sc0 ls0 ws53">component analysis (PCA), linear discrimin<span class="_ _0"></span>ant <span class="ls13 ws54">analysis (LDA), local feature analysis (LFA) </span></div><div class="t m0 x7 h7 y39 ff2 fs4 fc0 sc0 lse ws55">and independent component anal<span class="ls1c ws56">ysis (ICA) demonstrated that using thermal infrared </span></div><div class="t m0 x7 h7 y3a ff2 fs4 fc0 sc0 ls1a ws57">imagery yields higher perfo<span class="_ _1"></span>rmance than usin<span class="ls1b ws58">g visible images under many circumstances </span></div><div class="t m0 x7 h7 y3b ff2 fs4 fc0 sc0 lsd ws59">(Socolinsky & Selinger, 2002). </div><div class="t m0 x7 h7 y3c ff2 fs4 fc0 sc0 lsb ws5a">It is noted that the aforementioned database ma<span class="ls11 ws5b">inly involved same-session data (i.e.<span class="_ _0"></span>, nearly-</span></div><div class="t m0 x7 h7 y3d ff2 fs4 fc0 sc0 ls13 ws5c">simultaneous acquisition of tr<span class="_ _0"></span>aining and testing <span class="ls1d ws5d">data). Besi<span class="_ _0"></span>des the same-session test<span class="_ _0"></span>, Chen et </span></div><div class="t m0 x7 h7 y3e ff2 fs4 fc0 sc0 lsd ws5e">al. (Chen et al., 2005) paid more attention to te<span class="lse ws5f">st of time-lapse data (i.e., training data and </span></div><div class="t m0 x7 h7 y3f ff2 fs4 fc0 sc0 ls14 ws60">testing data being collected in different time <span class="ls9 ws61">sessions). Th<span class="_ _0"></span>e intervals among training data </span></div><div class="t m0 x7 h7 y40 ff2 fs4 fc0 sc0 ls0 ws62">and testing data are several weeks, <span class="_ _0"></span>severa<span class="lsf ws63">l months or even one year respectively. The </span></div><div class="t m0 x7 h7 y41 ff2 fs4 fc0 sc0 ls0 ws64">large-scale studies invol<span class="_ _0"></span>ving both same-session<span class="ls17 ws65"> and time-lapse <span class="_ _0"></span>data indicated that in a </span></div><div class="t m0 x7 h7 y42 ff2 fs4 fc0 sc0 ls14 ws66">same-session scenario, neither modality is sig<span class="_ _0"></span>n<span class="lsf ws67">ificantly better than the other using the PCA-</span></div><div class="t m0 x7 h7 y43 ff2 fs4 fc0 sc0 ls0 ws68">based recognition; however<span class="_ _0"></span>, using visible im<span class="_ _0"></span>agery outperforms that using<span class="_ _0"></span> IR imagery for </div><div class="t m0 x7 h7 y44 ff2 fs4 fc0 sc0 ls7 ws69">time-lapse data. </div><div class="t m0 x7 h7 y45 ff2 fs4 fc0 sc0 ls1e ws6a">When we mention that humans are homoi<span class="_ _1"></span>otherma<span class="ls14 ws6b">l, it should be highlighted that the so </span></div><div class="t m0 x7 h7 y46 ff2 fs4 fc0 sc0 ls13 ws54">called “homoiotherm” only refers to the approx<span class="ls10 ws6c">imately constant temperature in deep body </span></div><div class="t m0 x7 h7 y47 ff2 fs4 fc0 sc0 ls1d ws6d">(i.e., the core temperatu<span class="_ _0"></span>re), wh<span class="ls7 ws6e">ereas the skin temperature dist<span class="lsf ws6f">ribution fluctuates with the </span></span></div><div class="t m0 x7 h7 y48 ff2 fs4 fc0 sc0 ls17 ws70">ambient temperature, changes from person to <span class="ls10 ws71">person, and from time to time, as shown in </span></div><div class="t m0 x7 h7 y49 ff2 fs4 fc0 sc0 ls9 ws61">(Houdas & Ring, 1982, Guyton & Hall, 1996, Jo<span class="ls10 ws45">nes <span class="_ _0"></span>& Plassmann, 2000). It should also be </span></div><div class="t m0 x7 h7 y4a ff2 fs4 fc0 sc0 ls10 ws72">noted that an IR camera can only capture the apparent temperature instead of deep </div><div class="t m0 x7 h7 y4b ff2 fs4 fc0 sc0 lsc ws73">temperature. As indicated by Housdas & Ring (Houdas & Ring, 1982), the variations in </div><div class="t m0 x7 h7 y4c ff2 fs4 fc0 sc0 lsc ws74">facial thermograms result from not only ex<span class="_ _0"></span><span class="ws75">ternal conditions, such as environmental </span></div><div class="t m0 x7 h7 y4d ff2 fs4 fc0 sc0 ls0 ws76">temperature, imaging conditions, bu<span class="_ _0"></span>t also variou<span class="ls11 ws77">s internal conditions, such as phy<span class="_ _0"></span>siological </span></div><div class="t m0 x7 h7 y4e ff2 fs4 fc0 sc0 ls9 ws78">or psychological conditions. Socolinsky & Selinger (Socolinsky & Selinger, 2004A, </div><div class="t m0 x7 h7 y4f ff2 fs4 fc0 sc0 lsb ws26">Socolinsky & Selinger, 2004B) also<span class="ls1f ws79"> explored such variations. </span></div><div class="t m0 x7 h7 y50 ff2 fs4 fc0 sc0 ls14 ws34">To improve the performance of IR face recognition for time-lapse session,<span class="_ _0"></span> more efforts have </div><div class="t m0 x7 h7 y51 ff2 fs4 fc0 sc0 lsc ws7a">been put on classifier desi<span class="lsa ws7b">gn (Socolinsky & Selinger, 2004<span class="_ _0"></span><span class="lsc ws7a">A, Socolinsky & Selinger, 2004B, </span></span></div><div class="t m0 x7 h7 y52 ff2 fs4 fc0 sc0 lsc ws5c">Srivastava & Liu, 2003). Meanwhile, some resear<span class="lsa ws7c">chers focus on feature extraction. Yoshitomi </span></div><div class="t m0 x7 h7 y53 ff2 fs4 fc0 sc0 ls7 ws6e">et al. (Yoshitomi et al., 1997) used both ther<span class="ls10 ws7d">mal information and geometrical features for </span></div><div class="t m0 x7 h7 y54 ff2 fs4 fc0 sc0 lsc ws7e">recognition. Wu et al. (Wu et al., 2005A) proposed a model to convert th<span class="_ _0"></span>e thermograms to </div><div class="t m0 x7 h7 y55 ff2 fs4 fc0 sc0 lsd ws7f">blood perfusion data and the performance on ti<span class="ls11 ws80">me-lapse data is <span class="_ _0"></span>significantly improved.<span class="_ _0"></span> The </span></div><div class="t m0 x7 h7 y56 ff2 fs4 fc0 sc0 ls9 ws81">modified blood perfusion model by Wu et al. <span class="ls1f ws82">(Wu et al., 2007) further improves the time-</span></div><div class="t m0 x7 h7 y57 ff2 fs4 fc0 sc0 lsf ws83">lapse performance. </div><div class="t m0 x7 h7 y58 ff2 fs4 fc0 sc0 lsc ws84">In this chapter, we will provide a comprehensive study on the proposed blood perfu<span class="_ _0"></span>sion </div><div class="t m0 x7 h7 y59 ff2 fs4 fc0 sc0 ls9 ws85">models. It is revealed that the transforms by<span class="lsb ws86"> the blood perf<span class="_ _0"></span>usion models reduce the within-</span></div><div class="t m0 x7 h7 y5a ff2 fs4 fc0 sc0 ls13 ws87">class scatter of thermogr<span class="_ _0"></span>ams and obtains more consistent featu<span class="_ _0"></span>res to represent human faces. </div><div class="t m0 x7 h7 y5b ff2 fs4 fc0 sc0 ls13 ws88">In the following Section, the thermal pattern variations are analyzed. The blood perfusion </div><div class="t m0 x7 h7 y5c ff2 fs4 fc0 sc0 ls9 ws89">models are presented and analyzed in Section 3. <span class="_ _0"></span>A variety of experiments on blood </div><div class="t m0 x7 h7 y5d ff2 fs4 fc0 sc0 lsf ws1e">perfusion and thermal data are performed in <span class="ls9 ws1d">Section 4, and the conc<span class="ls13 ws8a">lusions are drawn in </span></span></div><div class="t m0 x7 h7 y5e ff2 fs4 fc0 sc0 ls12 ws3c">Section 5. </div></div><div class="pi" data-data='{"ctm":[1.991701,0.000000,0.000000,1.991701,0.000000,0.000000]}'></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/622b578715da9b288bc328e7/bg3.jpg"><div class="t m0 x7 hc y2c ff7 fs7 fc0 sc0 ls10 ws8b">Blood Perfusion Models for Infrared Fa<span class="_ _1"></span>ce Recognition </div><div class="t m0 xe hd y2d ff8 fs8 fc0 sc0 ls10 ws0"> </div><div class="t m0 xf he y2e ff7 fs4 fc0 sc0 ls19 ws0">185 </div><div class="t m0 x7 h6 y5f ff1 fs3 fc0 sc0 ls20 ws8c">2. Thermal pattern variations and analy<span class="_ _0"></span>sis </div><div class="t m0 x7 h7 y60 ff2 fs4 fc0 sc0 ls21 ws8d">Although IR images are ind<span class="_ _1"></span>ependent of illuminati<span class="ls22 ws8e">on, fluctuations in thermal appe<span class="_ _1"></span>arance occur </span></div><div class="t m0 x7 h7 y61 ff2 fs4 fc0 sc0 ls23 ws8">in relation to ambient conditi<span class="_ _1"></span>ons, subject’s meta<span class="ls22 ws8f">bolism and so on. It is necessary to learn how </span></div><div class="t m0 x7 h7 y62 ff2 fs4 fc0 sc0 ls23 ws90">the thermal patterns vary in <span class="_ _1"></span>different situatio<span class="ls24 ws91">ns bef<span class="_ _1"></span>ore presenting the propos<span class="_ _1"></span>ed methods. </span></div><div class="t m0 x7 h7 y63 ff2 fs4 fc0 sc0 ls25 ws92">Some of the factors affecting thermal distributi<span class="ls26 ws93">on are presente<span class="_ _1"></span>d in the following subsections.<span class="_ _1"></span> </span></div><div class="t m0 x7 hf y64 ff1 fs4 fc0 sc0 ls14 ws94">2.1 Deep body tempera<span class="_ _0"></span>ture vs skin temperature </div><div class="t m0 x7 h7 y65 ff2 fs4 fc0 sc0 ls27 ws43">In 1958, Aschoff and Weve<span class="_ _1"></span>r introduced the term<span class="ls28 ws95">s <span class="_ _1"></span>“thermal core”, the temperatures of <span class="_ _1"></span>which </span></div><div class="t m0 x7 h7 y66 ff2 fs4 fc0 sc0 ls24 ws7a">remain almost exactl<span class="_ _1"></span>y constant, within +/- 0.<span class="_ _1"></span>6°C, day in an<span class="_ _1"></span>d day out except when a pers<span class="_ _1"></span>on </div><div class="t m0 x7 h7 y67 ff2 fs4 fc0 sc0 ls29 ws96">develops a febrile illness (Guyton & Hall, 1996). Bl<span class="ws97">atteis (Blatteis, 1998) indi<span class="_ _1"></span>cates that even if </span></div><div class="t m0 x7 h7 y68 ff2 fs4 fc0 sc0 ls27 ws98">ambient temperature varies wi<span class="_ _1"></span>dely, core temperat<span class="ls23 ws99">ure does not change as a f<span class="_ _1"></span>unction of ambient </span></div><div class="t m0 x7 h7 y69 ff2 fs4 fc0 sc0 ls21 ws77">temperature. This is due to the presence of a cl<span class="ls2a ws0">os<span class="_ _3"></span>ed<span class="_ _3"></span> co<span class="_ _3"></span>nt<span class="_ _3"></span>ro<span class="_ _3"></span>l l<span class="_ _0"></span>o<span class="_ _3"></span>op<span class="_ _3"></span> wi<span class="_ _3"></span>th<span class="_ _3"></span> ne<span class="_ _0"></span>g<span class="_ _3"></span>at<span class="_ _0"></span>i<span class="_ _0"></span>v<span class="_ _3"></span>e f<span class="_ _3"></span>ee<span class="_ _3"></span>db<span class="_ _0"></span>a<span class="_ _3"></span>ck<span class="_ _0"></span> i<span class="_ _3"></span>n t<span class="_ _0"></span>h<span class="_ _0"></span>e<span class="_ _0"></span> </span></div><div class="t m0 x7 h7 y6a ff2 fs4 fc0 sc0 ls2b ws9a">body system which prevents mea<span class="_ _1"></span>n body temper<span class="ls2c ws9b">ature f<span class="_ _1"></span>rom deviating extensively f<span class="_ _1"></span>rom this </span></div><div class="t m0 x7 h7 y6b ff2 fs4 fc0 sc0 ls2d ws9c">value taken under thermoneutral conditions (Blatteis<span class="ls21 ws9d">, 1<span class="_ _1"></span>998). In fact, a rise in core te<span class="_ _1"></span>mperature </span></div><div class="t m0 x7 h7 y6c ff2 fs4 fc0 sc0 ls2e ws0">o<span class="_ _2"></span>f<span class="_ _5"></span> o<span class="_ _5"></span>n<span class="_ _2"></span>l<span class="_ _2"></span>y<span class="_ _5"></span> 0<span class="_ _5"></span>.<span class="_ _2"></span>5<span class="_ _2"></span> º<span class="_ _5"></span>C<span class="_ _5"></span> <span class="_ _1"></span>c<span class="_ _2"></span>a<span class="_ _5"></span>u<span class="_ _2"></span>s<span class="_ _5"></span>e<span class="_ _2"></span>s<span class="_ _5"></span> e<span class="_ _2"></span>x<span class="_ _5"></span>t<span class="_ _2"></span>r<span class="_ _5"></span>e<span class="_ _2"></span>m<span class="_ _5"></span>e<span class="_ _2"></span> p<span class="_ _5"></span>e<span class="_ _2"></span>r<span class="_ _5"></span>i<span class="_ _2"></span>p<span class="_ _2"></span>h<span class="_ _5"></span>e<span class="_ _2"></span>r<span class="_ _5"></span>a<span class="_ _2"></span>l<span class="_ _5"></span> v<span class="_ _2"></span>a<span class="_ _5"></span>s<span class="_ _2"></span>o<span class="_ _5"></span>d<span class="_ _2"></span>i<span class="_ _5"></span>l<span class="_ _2"></span>a<span class="_ _5"></span>t<span class="_ _2"></span>i<span class="_ _5"></span>o<span class="_ _2"></span>n<span class="_ _2"></span> (<span class="_ _5"></span>f<span class="_ _2"></span>l<span class="_ _5"></span>u<span class="_ _2"></span>s<span class="_ _5"></span>h<span class="_ _2"></span>i<span class="_ _5"></span>n<span class="_ _2"></span>g<span class="_ _5"></span> o<span class="_ _2"></span>f<span class="_ _5"></span> t<span class="_ _2"></span>h<span class="_ _5"></span>e<span class="_ _2"></span> s<span class="_ _5"></span>k<span class="_ _2"></span>i<span class="_ _5"></span>n<span class="_ _2"></span> i<span class="_ _2"></span>n<span class="_ _5"></span> h<span class="_ _5"></span>u<span class="_ _2"></span>m<span class="_ _5"></span>a<span class="_ _2"></span>n<span class="_ _5"></span>s<span class="_ _2"></span>)<span class="_ _2"></span>.<span class="_ _5"></span> T<span class="_ _2"></span>h<span class="_ _5"></span>i<span class="_ _2"></span>s<span class="_ _5"></span> </div><div class="t m0 x7 h7 y6d ff2 fs4 fc0 sc0 ls2d ws9e">stability implies that the heat produced in the body and that lo<span class="ls21 ws9f">st fro<span class="_ _1"></span>m it stay in relative </span></div><div class="t m0 x7 h7 y6e ff2 fs4 fc0 sc0 ls21 wsa0">balance, despite the large vari<span class="_ _1"></span>ations in ambient temperat<span class="_ _1"></span>ure (Blatteis, 1998). </div><div class="t m0 x7 h7 y6f ff2 fs4 fc0 sc0 ls2f wsa1">The skin temperature, i<span class="_ _1"></span>n <span class="ls30 wsa2">contrast to th<span class="_ _0"></span>e core temp<span class="_ _0"></span>erature, fluctua<span class="_ _0"></span>tes with th<span class="_ _0"></span>e tempera<span class="_ _0"></span>ture of the </span></div><div class="t m0 x7 h7 y70 ff2 fs4 fc0 sc0 ls31 wsa3">surroundings (G<span class="_ _1"></span>uyton & Hall, 1996, Blat<span class="_ _1"></span>teis, 1998). One ma<span class="_ _1"></span>y infer, therefore, t<span class="_ _1"></span>hat in order to </div><div class="t m0 x7 h7 y71 ff2 fs4 fc0 sc0 ls2a ws0">m<span class="_ _3"></span>ai<span class="_ _3"></span>nt<span class="_ _3"></span>a<span class="_ _0"></span>i<span class="_ _0"></span>n<span class="_ _3"></span> co<span class="_ _3"></span>re<span class="_ _3"></span> t<span class="_ _0"></span>e<span class="_ _0"></span>m<span class="_ _3"></span>pe<span class="_ _3"></span>ra<span class="_ _3"></span>t<span class="_ _0"></span>u<span class="_ _0"></span>r<span class="_ _3"></span>e s<span class="_ _3"></span>t<span class="_ _0"></span>a<span class="_ _0"></span>b<span class="_ _3"></span>le<span class="_ _3"></span>, t<span class="_ _3"></span>h<span class="_ _0"></span>e<span class="_ _3"></span> <span class="_ _1"></span>r<span class="_ _0"></span>a<span class="_ _3"></span>te<span class="_ _3"></span> of<span class="_ _3"></span> h<span class="_ _0"></span>e<span class="_ _3"></span>at<span class="_ _3"></span> f<span class="_ _0"></span>l<span class="_ _0"></span>o<span class="_ _3"></span>w f<span class="_ _3"></span>ro<span class="_ _3"></span>m<span class="_ _0"></span> c<span class="_ _0"></span>o<span class="_ _3"></span>re<span class="_ _3"></span> t<span class="_ _0"></span>o<span class="_ _0"></span> s<span class="_ _3"></span>ki<span class="_ _3"></span>n i<span class="_ _3"></span>s<span class="_ _0"></span> d<span class="_ _0"></span>j<span class="_ _3"></span>us<span class="_ _3"></span>t<span class="_ _0"></span>e<span class="_ _0"></span>d<span class="_ _3"></span> ac<span class="_ _3"></span>co<span class="_ _3"></span>r<span class="_ _0"></span>d<span class="_ _3"></span>in<span class="_ _3"></span>g t<span class="_ _3"></span>o </div><div class="t m0 x7 h7 y72 ff2 fs4 fc0 sc0 ls32 wsa4">the body’s thermal needs <span class="_ _1"></span>and that, as a res<span class="_ _1"></span>ult, sk<span class="ls33 wsa5">in temperat<span class="_ _1"></span>ure varies mo<span class="_ _1"></span>re widely than co<span class="_ _1"></span>re </span></div><div class="t m0 x7 h7 y73 ff2 fs4 fc0 sc0 ls34 ws99">temperature in <span class="_ _1"></span>relation to am<span class="_ _1"></span>bient temperat<span class="_ _1"></span>ure (Bla<span class="ls35 wsa6">tteis, 1998<span class="_ _1"></span>). Under stead<span class="_ _1"></span>y-state cond<span class="_ _1"></span>itions in </span></div><div class="t m0 x7 h7 y74 ff2 fs4 fc0 sc0 ls35 wsa7">a thermoneutral enviro<span class="_ _1"></span>nment, (i.e., one in whi<span class="_ _1"></span>ch neither the <span class="_ _1"></span>mechanism for he<span class="_ _1"></span>at producti<span class="_ _1"></span>on nor </div><div class="t m0 x7 h7 y75 ff2 fs4 fc0 sc0 ls36 wsa8">for heat loss is activ<span class="_ _0"></span>ated and the perce<span class="_ _0"></span>ived thermal comfor<span class="_ _0"></span>t is optimal), core te<span class="_ _0"></span>mperature thus is </div><div class="t m0 x7 h7 y76 ff2 fs4 fc0 sc0 ls37 wsa9">higher than<span class="_ _0"></span> skin temp<span class="_ _0"></span>erature (Blatt<span class="_ _0"></span>eis, 1998). For re<span class="_ _0"></span>sting, naked adul<span class="_ _0"></span>ts, this zone of<span class="_ _0"></span> ambient </div><div class="t m0 x7 h7 y77 ff2 fs4 fc0 sc0 ls34 ws99">temperature lies betwe<span class="_ _1"></span>en 28 and 30 °C (Blatteis,<span class="_ _1"></span> 1998). </div><div class="t m0 x7 h7 y78 ff2 fs4 fc0 sc0 ls17 wsaa">There is no single temperature level that <span class="lsb wsab">can be considered to<span class="_ _0"></span> be normal because </span></div><div class="t m0 x7 h7 y79 ff2 fs4 fc0 sc0 ls38 wsac">measurements on many normal people have shown a <span class="ff3 ls39 ws0">range</span><span class="ls17 wsad"> of normal temperature </span></div><div class="t m0 x7 h7 y7a ff2 fs4 fc0 sc0 lsb ws26">measured orally, from less th<span class="ls10 wsae">an 36.1°C to 37.5°C (Guyton & Ha<span class="lsc wsaf">ll, 1996). Wh<span class="_ _0"></span>en excessive heat </span></span></div><div class="t m0 x7 h7 y7b ff2 fs4 fc0 sc0 ls3a wsb0">is produced in the body by strenuou<span class="_ _0"></span>s exercise<span class="lsf wsb1">, temperature can rise temporarily to as high </span></div><div class="t m0 x7 h7 y7c ff2 fs4 fc0 sc0 lsf wsb2">as 38.33-40.0°C. On the other hand, when the bo<span class="ls14 wsb3">dy is exposed to cold, the temperature can </span></div><div class="t m0 x7 h7 y7d ff2 fs4 fc0 sc0 lsd ws59">often fall to values below 96°F <span class="lsb ws26">(35.56°C) (Guyton & Hall, 199<span class="_ _0"></span>6). </span></div><div class="t m0 x7 hf y7e ff1 fs4 fc0 sc0 ls3a wsb4">2.2 Variation with ambie<span class="_ _0"></span>nt conditions<span class="_ _0"></span> </div><div class="t m0 x7 h7 y7f ff2 fs4 fc0 sc0 ls26 wsb5">The works by Chen et al. Chen et al., 2005),<span class="ws6b"> Socolinsky & Selinger (S<span class="_ _1"></span>ocolinsky & Selinger, </span></div><div class="t m0 x7 h7 y80 ff2 fs4 fc0 sc0 ls21 wsb6">2004A, Socolinsky & Selinger, 2004B), Wu e<span class="_ _1"></span>t al<span class="wsb7">. (Wu et al<span class="_ _1"></span>., 2005A, Wu et al., 2007) ha<span class="_ _1"></span>ve </span></div><div class="t m0 x7 h7 y81 ff2 fs4 fc0 sc0 ls3b wsb8">illustrated that variations in ambient temp<span class="ls25 wsb9">erature significantly <span class="ls3c wsba">change the thermal </span></span></div><div class="t m0 x7 h7 y82 ff2 fs4 fc0 sc0 ls3d wsbb">characteristics of faces, and accordingly affect the performances<span class="ls26 wsbc"> of recognition. Fig. 1 shows<span class="_ _1"></span> </span></div><div class="t m0 x7 h7 y83 ff2 fs4 fc0 sc0 ls3e wsbd">the thermal distribution of t<span class="_ _1"></span>he same face in di<span class="ls27 wsbe">ffere<span class="_ _1"></span>nt ambient te<span class="_ _1"></span>mperatures. All of<span class="_ _1"></span> the images </span></div><div class="t m0 x7 h7 y84 ff2 fs4 fc0 sc0 ls27 wsbe">are observed (the red part<span class="_ _1"></span>) from the pixel v<span class="_ _1"></span>alues ranging f<span class="_ _1"></span>rom 238 to 255. It is observed<span class="_ _1"></span> from </div><div class="t m0 x7 h7 y85 ff2 fs4 fc0 sc0 ls27 wsbf">Fig.1 that the skin <span class="_ _1"></span>temperature of the cheeks<span class="_ _1"></span>, tip of nose and hair increas<span class="_ _1"></span>es as the ambient<span class="_ _1"></span> </div><div class="t m0 x7 h7 y86 ff2 fs4 fc0 sc0 ls28 wsc0">temperature increases. The intensities<span class="_ _1"></span> of the fo<span class="ls3f wsc1">rehe<span class="_ _1"></span>ad region st<span class="_ _1"></span>art off as bright <span class="_ _1"></span>when the </span></div><div class="t m0 x7 h7 y87 ff2 fs4 fc0 sc0 ls2b wsc2">ambient temperature is low. <span class="_ _1"></span>As the ambient temperature i<span class="_ _1"></span>ncreases (above 27.9 º<span class="_ _1"></span>C ~ 28.1 ºC), </div><div class="t m0 x7 h7 y88 ff2 fs4 fc0 sc0 ls24 wsc3">the intensities of the fo<span class="_ _1"></span>rehead region dro<span class="_ _1"></span>ps dras<span class="ls26 wsc4">tically due<span class="_ _1"></span> to the effect<span class="ls40 wsc5"> of sweating. It was </span></span></div><div class="t m0 x7 h7 y89 ff2 fs4 fc0 sc0 ls21 ws8d">indicated by Blatteis (Blatteis, 1998) that the <span class="ls3f wsc6">human b<span class="_ _1"></span>ody has about three million swea<span class="_ _1"></span>t glands, </span></div><div class="t m0 x7 h7 y8a ff2 fs4 fc0 sc0 ls2b ws5c">the greatest density being found on t<span class="_ _1"></span>he palms, <span class="ls41 wsc7">soles and forehead. Thermoregulatory sweating </span></div></div><div class="pi" data-data='{"ctm":[1.991701,0.000000,0.000000,1.991701,0.000000,0.000000]}'></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/622b578715da9b288bc328e7/bg4.jpg"><div class="t m0 xd hc y2c ff7 fs7 fc0 sc0 ls18 ws40"> <span class="_ _4"> </span>Recent Advances in Fa<span class="_ _1"></span>ce Recogniti<span class="_ _1"></span>on </div><div class="t m0 x7 hd y2d ff8 fs8 fc0 sc0 ls10 ws0"> </div><div class="t m0 x7 he y2e ff7 fs4 fc0 sc0 ls19 ws0">186 </div><div class="t m0 x7 h7 y2f ff2 fs4 fc0 sc0 ls42 wsb7">increases with elevation i<span class="_ _1"></span>n core temperature (Blatteis, 199<span class="_ _1"></span>8), and therefore the forehe<span class="_ _1"></span>ad </div><div class="t m0 x7 h7 y30 ff2 fs4 fc0 sc0 ls3e wsc8">region emits sweat easil<span class="_ _1"></span>y when the ambient te<span class="ls22 wsc9">mperature increases. <span class="ls28 wsca">Evap<span class="_ _1"></span>oration takes place<span class="_ _1"></span> </span></span></div><div class="t m0 x7 h7 y31 ff2 fs4 fc0 sc0 ls26 ws93">once the sweat reaches the surf<span class="_ _1"></span>ace, hence causing the skin temperat<span class="_ _1"></span>ure to lower down. </div><div class="t m0 x7 h10 y8b ff2 fs9 fc0 sc0 ls10 ws0"> </div><div class="t m0 x10 h7 y8c ff2 fs4 fc0 sc0 ls10 ws0"> </div><div class="t m0 x7 h7 y8d ff2 fs4 fc0 sc0 lsd wscb">Fig. 1. Images taken at different ambient con<span class="_ _0"></span>dit<span class="lsb ws5a">ions (1st row: 24.8 ºC ~ 25.0 ºC, 2nd row: 25.7 </span></div><div class="t m0 x7 h7 y8e ff2 fs4 fc0 sc0 lsa wscc">ºC ~ 26.1 ºC, 3th row: 27.1 ºC ~ 27.4<span class="_ _0"></span> ºC, 4th row: <span class="ls10">27.9 ºC ~ 2<span class="_ _0"></span>8.1 ºC, 5th row: 28.4 ºC ~ 28.7 ºC, </span></div><div class="t m0 x7 h7 y8f ff2 fs4 fc0 sc0 lsc wsaf">6th row: 28.9 ºC ~ 29.3 ºC) </div></div><div class="pi" data-data='{"ctm":[1.991701,0.000000,0.000000,1.991701,0.000000,0.000000]}'></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/622b578715da9b288bc328e7/bg5.jpg"><div class="t m0 x7 hc y2c ff7 fs7 fc0 sc0 ls10 ws8b">Blood Perfusion Models for Infrared Fa<span class="_ _1"></span>ce Recognition </div><div class="t m0 xe hd y2d ff8 fs8 fc0 sc0 ls10 ws0"> </div><div class="t m0 xf he y2e ff7 fs4 fc0 sc0 ls19 ws0">187 </div><div class="t m0 x11 h7 y90 ff2 fs4 fc0 sc0 ls10 ws0"> </div><div class="t m0 x7 h7 y91 ff2 fs4 fc0 sc0 ls17 wscd">(a) Histogram in ambient temperature 24.8 ºC ~ 25.0 ºC </div><div class="t m0 x11 h7 y92 ff2 fs4 fc0 sc0 ls10 ws0"> </div><div class="t m0 x7 h7 y93 ff2 fs4 fc0 sc0 lse wsce">(b) Histogram in ambient temperature 28.9 ºC ~ 29.3 ºC </div><div class="t m0 x7 h7 y94 ff2 fs4 fc0 sc0 ls15 wscf">Fig. 2. Histograms in different ambient temperatures of the same face </div></div><div class="pi" data-data='{"ctm":[1.991701,0.000000,0.000000,1.991701,0.000000,0.000000]}'></div></div>