【正文】 inx Inc., “Virtex4 Data Sheets: Virtex4 Family Overview,” Sep. 2020. DOI= [6] Y. Wei, X. Bing, and C. Chareonsak, “FPGA implementation of AdaBoost algorithm for detection of face biometrics,” In Proceedings of IEEE International Workshop Biomedical Circuits and Systems, page S1, 2020. [7] M. Yang, Y. Wu, J. Crenshaw, B. Augustine, and R. Mareachen, “Face detection for automatic exposure control in handheld camera,” In Proceedings of IEEE international Conference on Computer Vision System, , 206. [8] V. Nair, P. Laprise, and J. Clark, “An FPGAbased people detection system,” EURASIP Journal of Applied Signal Processing, 2020(7), pp. 10471061, 2020 [9] C. Gao and S. Lu, “Novel FPGA based Haar classifier face detection algorithm acceleration,” In Proceedings of International Conference on Field Programmable Logic and Applications, 2020. 譯文二 基于半邊臉的人臉檢測 2 概要： 圖像中的人臉檢測是人臉識別研究中一項(xiàng)非常重要的研究分支。s face, locate to is followed a person39。s face examination district inside, passing the formula(7) e to something to (|| Wkcandidate － Wface||,|| Wkcandidate － Wnonface||),(7) among them the Wkcandidate is to trains( test) the characteristic space inside to the k a candidate a district, and Wface, Wnonface respectively is training( test) characteristic space middleman face with not person39。s face or not person39。s skin district, skin color is not suitable for in the person an use of to act information can away with this weakness the sake of the precision, after the skin color divides into section, consider the skin color district of the containment action , the action information of the bination skin color model leads a binary system for a containment scene( person39。s face candidate for election districts first, bine exploitation PCA technique to judge the real person a , make use of the characteristic technique( eigen － technique) follow to confirmed person39。02) According to the method of the edge useds for the edge that follow a picture preface row, but these edgeses is usually the boundary line of the main , because were musted shine on with the light at the color by the on the trail of object the term descends to display the obvious edge changes, so these methodses will fall among the color with the variety that light shine addition, be a background of picture contain very obvious edge,( follow the method) dependable result in very difficult this type of method that a lot of cultural heritages all involve e from the Kass et the snake form rate of exchange motion [ 5 the achievement of ] see the scene of to acquire from included various the noise of varieties solid the hour the resemble the machine of, therefore many systems is very rare to dependable person39。s face with recent drive the person who examine the characteristic space inside of the a. Useding for a for following resembles the controller the work in such way: Make use of equilibrium/ tilt to one side (pan/ tilt) the terrace, examine drive of person a district controls at hold the act central. This method cans also expand to go to in the other system, for example telemunication meeting, invader check system etc. 1 preface Seeing the signal of handles many applications, for example owing to the munication can see the telemunication meeting that turn, for disable and sick person service of the lips reads the system. In up many systems that mention, the facial examination in person drink to follow to see to can39。未來的工作是我們將進(jìn)一步發(fā)展這種方法，通過從被檢測的人臉區(qū)域 種萃取臉部特征來為臉部活動系統(tǒng)服務(wù)。用 偽代碼來表示平衡 /傾斜處理的持續(xù)時(shí)間和攝像機(jī)的定位。使用這個(gè)距離向量，就能控制攝像機(jī)中定位和平衡 /傾斜的持續(xù)時(shí)間。投影區(qū)域的檢驗(yàn)是利用人臉類和非人臉類的檢測區(qū)域內(nèi)的最小距離，通過公式（ 7）來實(shí)現(xiàn)的。因此，為了計(jì)算的可行性，與其為 C 找出特征向量，不如我們計(jì)算 [YTY]中 M個(gè)特征向量 vk和特征值 ? k， 所以用 u k＝kY?vk*來計(jì)算一個(gè)基本集合，其中 k＝ 1，?， M。這個(gè) 訓(xùn)練（測試）集的平均值用 A＝ M1 ??Mi Ii1來定義。我們使用特征空間中候選區(qū)域的分量向量來為人臉檢驗(yàn)問題服務(wù)。 St 是當(dāng)前幀中膚色像素的集合，（斯坦） t 是利用適當(dāng)?shù)拈撓藜夹g(shù)計(jì)算出的閾限值 [9]。利用動作信息可以有效地去除這個(gè)缺點(diǎn)。人臉的色彩分布是在一個(gè)小的彩色的色彩空間中成群的，且可以通過一個(gè) 2 維的高斯分部來近似。為了改進(jìn)人臉檢測的精確性，我們把諸如膚色模型 [1,6]和 PCA[7,8]這些已經(jīng)發(fā)表的技術(shù)結(jié)合起來。這種方法由兩大步驟構(gòu) 成：人臉檢測和人臉跟蹤。因?yàn)橐? 1 DoJoon Jung, ChangWoo Lee, YeonChul Lee, SangYong Bak, JongBae Kim, Hyun Kang, HangJoon Kim. International Technical Conference on Circuits/Systems, Computers and Communications (ITCCSCC39。 基于邊緣的（跟蹤）方法用于跟蹤一幅圖像序列的邊緣，而這些邊緣通常是主要對象的邊界線。一般來說，根據(jù)跟蹤角度的不同，可以把跟蹤方法分為兩類。這個(gè)方法還可以擴(kuò)展到其他的系統(tǒng)中去，例如電信會議、入侵者檢查系統(tǒng)等等 。為了實(shí)現(xiàn)人臉的檢測，首先，我們要用一個(gè)膚色模型和一些動作信息 (如：姿勢、手勢、眼色 )。這種方法是以主要成分分析技術(shù)為基礎(chǔ)的。用于人臉跟蹤的攝像控制器以這樣的方法工作：利用平衡 /（ pan/tilt）平臺， 把被檢測的人臉區(qū)域控制在屏幕的中央。在本文中，涉及到一些實(shí)時(shí)的人臉區(qū)域跟蹤 [13]?；趧幼鞯母櫴且蕾囉趧幼鳈z測技術(shù)，且該技術(shù)可以被分成視頻流（ optical flow）的（檢測）方法和動作 — 能量（ motion－ energy）的（檢測）方法。當(dāng)前很多的文獻(xiàn)都涉及到的這類方法時(shí)源于 Kass et [5]的成就。 在本文中，我們提出了一種基于 PCA 的實(shí)時(shí)人臉檢測和跟蹤方法，該方法是利用一個(gè)如圖 1 所示的活動攝像機(jī)來檢測和識別人臉的。 在這一部分中，將介紹本文提及到的方法中的用于檢測人臉的技術(shù)。通過亮度區(qū)分一個(gè)彩色像素的三個(gè)成分，可以移動亮度。 動作檢測 雖然膚色在特征的應(yīng)用種非常廣泛，但是當(dāng)膚色同時(shí)出現(xiàn)在背景區(qū)域和人的皮膚區(qū)域時(shí) ，膚色就不適合于人臉檢測了。這幅二進(jìn)制圖像定義為 ，其中 It(x,y) 和 It1(x,y)分別是當(dāng)前幀和前面那幀中像素（ x,y）的亮度。此外，還需要檢驗(yàn)這個(gè)移動的對象是人臉還是非人臉。 為了簡述這個(gè)特征空間，假設(shè)一個(gè)圖像集合 I1， I2， I3，?， IM，其中每幅圖像是一個(gè) N 維的列向量，并以此構(gòu)成人臉空間。雖然矩陣 C 是 N N 維的，但是定義一個(gè) N 維的特征向量和 N 個(gè)特征值是個(gè)難處理的問題。 為了檢驗(yàn)候選的人臉區(qū)域是否是真正的人臉圖像，也會利用公式（ 6）把這個(gè)候選人臉區(qū)域投影到訓(xùn)練（測試）特征空間中。 在定義了人臉區(qū)域后 ，位于被檢測人臉區(qū)域的中心和屏幕中心之間的距離用distt（ face， screen）＝ Facet（ x， y）－ Screen（ height/2， width/2），（ 9）來計(jì)算，其中 Facet（ x， y）是時(shí)間 t 內(nèi)被檢測人臉區(qū)域的中心， Screen（ height/2， width/2）是屏幕的中心區(qū)域。參數(shù)表示的是活動攝像機(jī)的控制。在一個(gè)視頻輸入流中，首先，我們利用注入色彩、動作信息和 PCA 這類提示來檢測人臉區(qū)域，然后，用這樣的方式跟蹤人臉：即通過一個(gè)安裝了平衡 /請求平臺的活動攝像機(jī)把被檢測的人臉區(qū)域保持在屏幕的中央。s face follows according to the is several in the virtuous (Euclidian) distance of, among them the is several to reign in the virtuous distance in past drive on the trail of person39。s face is divided into according to edge of on the trail of with on the trail of [that according to district 4]. According to the on the trail of that identify is really with the object identifies technique is basal, but follow the function of the system is the restri