200530943 玖、發明說明 【發明所屬之技術領域】 本發明係關於一種數位保全系統，且更具體地，本發 明係關於一種利用臉部三維影像進行辨識之數位保全^ 統。 ’、 【先前技術】 現今一般之保全系統，大多使用刷卡方式作為管制人員 進出之方式。但此種採用刷卡機制之保全系統，經常會發 生卡片遺失或遭冒用之情形，使得保全系統之管制機制產 生漏洞。故隨之發展出採用生物特徵作為辨識基礎之保全 系統。因每個人所擁有之生物特徵均不同，故使用生物 特徵作為辨識機制之保全系統將具有較高之專一性與安 全性。使用生物特徵作為辨識之方式包含臉部辨識，語音 辨識，指紋辨識，視網膜辨識，與虹膜 用臉部特徵作為身份辨識之優點在於可身；二； 之可能性減至最低，可更精確地辨識人員之身份。 第1圖係繪示習知技藝之臉部辨識方法。當欲進入者 10欲進入保全系統時，雷射光源12乃產生一雷射格柵14， 投射於欲進入者10之臉部。雷射格栅14乃作為一座標尺 規，藉由紀錄臉部特徵於雷射格柵14之位置，以測量臉部 特徵間之相對距離。但此種臉部辨識方式具有數個缺點。 首先’雷射光源12所產生之雷射格柵14解析度有限，故 無法精確地測量臉部特徵間之相對距離。且#雷射格拇14 200530943 投射於欲進人纟1G之臉科，因人體臉部並非—平碎表 面，將造成雷射格栅u變形成為—曲面，而造成测量:之 誤差，進一步於比對時產生辨識錯誤之情況。 因此需要-種改進之保全系統，得以更精確地辨識臉 部特徵間之相對距離，以降低發生辨識錯誤之情 言 保全系統之可靠性。 间 【發明内容】 維影像進 因此本發明之一目的係提供一種藉由臉部三 行辨識之數位保全系統。 種藉由臉部三維影像辨識 本發明之另一目的係提供一 之數位保全系統管控方法。 …根據本發明之上述目的，提出_種藉由臉部三維影像 =仃辨識之數位保全系統。根據本發明之數位保全系統， 各有一維影像成像裝置，三維影像擷取裝置，資料庫，三 維〜像比對裝i ’以及管制裝置。三維影像成像裝置乃用 以成像奴進入者之臉部三維影像。三維影像擷取裝置用以 擷取臉部三維影像中，臉部特徵間之相對距離。資料庫乃 儲存允卉進入者之臉部紀錄，每一臉部紀錄含有每一允呼 進入者之臉部特徵間之相對距離。三維影 ^ 者臉部特徵間之相對距離，逐一與資料庫中，允 、者之臉部紀錄比較，並產生一比較結果。隨後，其 制裝置根據三維影像比對裝置產生之比較結果 : 否允許欲進入者進入。 兴疋疋 200530943 根據本發明之另-目的，提出一種藉由臉部三維影像 辨識之數位保全系統管控方法。根據本發明之數位保全系 統管控方法，首先成像欲進入者之臉部三維影像。接著， 擷取臉部三維影像中，臉部特徵間之相對距離。隨後，針 對欲進入者臉部特徵間之相對距離，盘一 允許進人者之臉部紀錄比較，並產生—比較結 根據此比對結果，以決定是否允許欲進入者進入。 ,、、根據本發明之數位保全系統，可更快速且準確地辨識 欲進入者之身❾。即使欲進人者藉由化妝改變其外貌，但 其臉部特徵之相對距離並不會因此而改變，故仍可正確地 進打辨識。且根據本發明之臉部特徵辨識方法，無須使用 田射秸柵，故亦無測量誤差而造成辨識錯誤之情形發生。 根據本發明之數位保全系統，可用於許多場合之門荦管 制，例如公家軍警政單位、大型企業工廠、教育單Z， J i行號、社區、員工進出管制、一般居家防盜設備等。 【實施方式】 請參照第2圖，係繪示依照本發明之數位保全系統示 意圖。根據本發明之數位保全系統2〇，含有三維影像成像 裝置22，三維影像擷取裝置24，資料庫26，三維影像比對 裝置28，與管制裝置29。三維影像成像裝置22乃用以獲 取欲進入者之臉部三維影像。三維影像擷取裝置24接著對 於此臉部三維影像進行辨識，擷取此臉部三維影像中，所 有臉部特徵之位置與彼此間之相對距離。三維影像比對裝 200530943 置28接著將三維影像擷取裝置24所擷取之臉部特徵間相 對距離，與資料庫26中所儲存之臉部紀錄進行比較，資料 庫26所儲存之臉部紀錄乃含有所有允許進入之人員，其臉 部特徵間相對距離之資料。隨後，管制裝置29根據三維影 像比對裝置28產生之比較結果，以決定是否允許欲進入者 進入，或啟動其他之保全措施。 第3圖與第4圖乃以一較佳具體實施例，進一步說明 根據本發明之數位保全系統。第3圖係繪示根據本發明之 臉部辨識方式示意圖。當一欲進入者1〇欲進入數位保全系 統20時，三維影像成像裝置22，例如一雙鏡頭數位立體 成像相機，對於欲進入者10進行成像，以獲得欲進入者1〇 之臉部三維影像。三維影像成像裝置22可進一步設計為可 自動感應名欠進入I 10之接近，並提示欲進入纟10若欲進 入，需先進行影像攝取之步驟。 三維影像擷取裝置24接著對於三維影像成像裝置22 所攝取之臉部二維影像進行辨識。首先係辨識所攝得之 部三維影像中，各個臉部特徵之位置。此臉部特徵例如為 眉毛’眼睛，袅早，峻pr 甘 # M ^ 耳木專，並標示出每個臉部特 位址。例如，帛4圖⑩示㈣取之-臉部三維 純，於此臉部三維f彡像巾，將右眉靠近 心 …一夺另一鳊點軚示為h，將左眉靠近臉部 ^ 點標不為Ν，並將另一端點標示為b2。 :似地’可將右眼靠近臉部中心之端點標 另一端點標示為c 验士日土 將 為2將左眼罪近臉部中心之端點標示為七， 200530943 下端點右耳之上端點標示為… 二了為左耳之上端點標示為fl，下端點標示為 二亡山77'尖標不為h，並將嘴巴之右端點標示為gI， 將左端點標示為g2。 =，任意擁取兩端點間之距離，例如，擷取端點〜 至1點ba距離’此距離乃表示兩眉中心端點間之距離。 於此具體實施例_，可擷取之端點有a,、a2、bl、b2、Ci、 '62 6、gl、g2、h，故任意擷取兩端點 =料形成c，/種組合。監控人員可根據所需精確度而 ^取適¥之端點，當擁取之端點越多時，所形成之距離组 口可月Μ生也越多，則於進行比對之步驟時，產生辨識錯誤 之可能性也越低。 例如’於本較佳具體實施例中，為說明目的，乃選取 七個端點間之距離’分別為aibi、aic丨、4、一丨、以丨、 〜gi、aih。當三維影像擷取裝置24獲得欲進入者ι〇之此 七個端點距離後’接著，三維影像比對裝置28便鱼資料庫 %中所儲存之所有允許進入者之此七個端點距離逐一比 較。於三維影像比對裝4 28中，可設定—相符程度⑽值， 例如’當欲進入者10之此七個端點距離中，係有五個距離 與資料庫26所儲存之某—允許進人者相同，則表示此欲進 入者10為資料庫中26之此允許進入者，故可允許其進入。 可根據實際需要選取臉部特徵之端點，以及設定不同 =相符程度門檻值。例如’當數位保全系统2〇乃用於管制 高機密或貴重裝置時，便可增加所選取之臉部特徵之端 200530943 點。當所選取之臉部特徵端點越多，則產生之可能距離組 合越多，同時亦可提高相符程度門檻值，則產生辨識錯誤 之可能性就越低。當數位保全系統20用於一般門禁用途 時，則可降低所擷取之臉部特徵端點，同時降低相符程度 門檻值，以減少進行辨識時所花費之時間。 三維影像比較裝置28接著產生一比較結果，產生一允 許進入絲絕進人之比較結果。t制裝置29接著根據此比 較結果，以啟動各種不同之保全機制。例如，當管制裝置 29接收到允許進入之比較結果後，便允許此欲進入者⑺進 入，同時紀錄此欲進入者1〇進入之時間，並儲存此進入者 1〇之臉部三維影像，以作為日後追蹤人員出人紀錄之用。 當管制裝置29接收到拒絕進入之比較結果後，可提干 欲進入者1〇再次進行攝影之動作，以重新進行辨識。或者， 可於要求欲進人者1G進行重新辨識時， 員，告知有欲進入者10嘗試進入作失 皿t人 ★ + _ 疋1一天敗，需即時加以觀察。 或者，亦可設計為當連續三次辨識後均失敗，則啟動保全 防4措施，例如啟動警報裝置，或通知監控人員等。王 此^ ’亦可與現有之其他辨識機制結合 合輸入密碼之機制，或更 了配 識，虹膜辨識等m^ 5 s”辨識’指紋辨 之數位保全系統 多重'證機制，以提供更周延 第5圖係為一流程圖，用以說明 全系統管控方法。首先乃對於欲進幻之數位保 欲進入者之臉部三維影像（步驟5。。)。：;成：，以獲得 )接耆，由所獲得之 200530943 臉部三維影像中，擷取臉部特徵間之相對距離（步驟502 )。 隨後’將㈣取之這些臉部特徵間相對距離，逐—與資料 庫中所儲存之允許進人者之臉部紀錄進行比較（步驟 504 )臉紀錄乃含有所有允許進入纟，其臉部特徵間相 對距離。 將欲進入者之臉部特徵相對距離與資料庫比較之後， 產生-比較結果。若此比較結果高於一預設之相符程度門 檻值（步驟506 )，則允碑Α卜处、心 ；則兄訐此欲進八者進入（步驟508 )。之 後，可進-步紀錄此欲進入者之臉部三維影像，以及此欲 進入者進人保全系統之時間（步驟5ig)，以作為日後追縱 出入記錄之用。 若比較結果乃低於預設之相符程度門櫃值，則要求欲 進入者再次進行臉部三維影像成像與辨識（步驟511)。同 :’亦可根據使用需求與設定’進—步警示保全系統之監 控人員（步驟512)，或是啟動警報裝置（步驟514)。 根據本發明之數位保全系統， 欲進入者之身份。即使欲進人者^2速且準確地辨識 考错由化妝改變其外貌 其臉部特徵之相對距離並不會因此而改變故仍可正確地進 =。=本發明之臉部特徵辨識方法，無須使用雷 身格栅’㈣無測量誤差而造成辨識錯誤之情形發 據本發明之數位保全系統，可用 x 例如公家軍警政單位、大型企業工廠、教」：，制， 行號、社區、員工進出管制、 —@早 小型 雖然本發明已P較佳盜設備等。 貫知例揭露如上，然其並非用 200530943 以限定本發明，任何熟習此技藝者，在不脫離本發 神和範圍β，當可作各種之更動與潤飾，因此本發明之^ 遵乾圍當視後附之申請專利範圍所界定者為準。 【圖式簡單說明】 由以上本發明中較佳具體實施例之細節描述，可以 對本發明之目的、觀點及優點有更佳的了解。同 下列本發明之圖式加以說明： 第1圖係繪示習知技藝之臉部辨識方法示意圖。 第2圖係繪示依照本發明—較佳具體實施例之數 全系統方塊圖。 保 第3圖係繪示依照本發明一 識示意圖。 ⑽具體貫施例之臉部辨 第4圖係繪示依照本發明—較佳具體 臉部三維影像示意圖。 J彌取之 第5圖係繪示依照本發明一較佳具體實施例之 王系統官控方法流程圖。 ’、 【元件代表符號簡單說明】 10 12 14 20 欲進入者 雷射光源 雷射格桃 數位保全系統 三維影像成像裝置 12 22 200530943 24 三維影像擷取裝置 26 資料庫 28 三維影像比對裝置 29 管制裝置200530943 发明 Description of the invention [Technical field to which the invention belongs] The present invention relates to a digital security system, and more specifically, the present invention relates to a digital security system using three-dimensional facial image recognition. ‘Previous technology’ Most security systems today use a swipe card method as a way for control personnel to get in and out. However, such a security system that uses a card swipe mechanism often results in lost or fraudulent cards, making loopholes in the security system's control mechanism. Therefore, a security system using biological characteristics as the basis for identification was developed. Because each person has different biological characteristics, a security system using biological characteristics as an identification mechanism will have higher specificity and security. The use of biometrics as a means of identification includes facial recognition, voice recognition, fingerprint recognition, retinal recognition, and iris. The advantages of using facial features as identity recognition are that it can be worn. Second, the possibility is minimized and can be identified more accurately. Identity of personnel. Figure 1 shows the face recognition method of the conventional technique. When the entrant 10 wants to enter the security system, the laser light source 12 generates a laser grid 14 to be projected on the face of the entrant 10. The laser grille 14 is used as a ruler to measure the relative distance between facial features by recording the position of facial features on the laser grille 14. However, this face recognition method has several disadvantages. First, the resolution of the laser grid 14 generated by the 'laser light source 12 is limited, so the relative distance between facial features cannot be accurately measured. And #laser grid thumb 14 200530943 is projected on the face of the person who wants to enter 1G, because the human face is not a flat surface, the laser grille u will be deformed into a curved surface, which will cause the measurement: Misidentification occurred during comparison. Therefore, an improved security system is needed to more accurately recognize the relative distances between facial features, so as to reduce the occurrence of recognition errors. The reliability of the security system. [Summary of the invention] Dimensional image advancement It is therefore an object of the present invention to provide a digital security system that recognizes three lines of the face. A three-dimensional image recognition through the face Another object of the present invention is to provide a digital security system management method. … According to the above-mentioned object of the present invention, a digital security system that recognizes a three-dimensional image of a face = 仃 recognition is proposed. According to the digital security system of the present invention, each has a one-dimensional image imaging device, a three-dimensional image capture device, a database, a three-dimensional to image comparison device i ', and a control device. The three-dimensional image imaging device is used to image the three-dimensional image of the face of the entrant. The 3D image capture device is used to capture the relative distance between facial features in a 3D image of a face. The database stores the facial records of the entrants. Each facial record contains the relative distance between the facial features of each entrant. The relative distances between the three-dimensional shadows' facial features are compared with the facial records in the database one by one, and a comparison result is generated. Subsequently, the manufacturing device according to the comparison result produced by the three-dimensional image comparison device: whether to allow the person who wants to enter. Xingye 200530943 According to another object of the present invention, a method for controlling a digital security system by recognizing a three-dimensional image of a face is proposed. According to the digital security system management and control method of the present invention, a three-dimensional image of the face of a person who wants to enter is first imaged. Then, the relative distance between the facial features in the three-dimensional image of the face is captured. Then, for the relative distance between the facial features of the person who wants to enter, Pan 1 allows the person's facial records to be compared, and a comparison is made based on the comparison result to decide whether to allow the person who wants to enter. According to the digital security system of the present invention, the identity of the person who wants to enter can be identified more quickly and accurately. Even if the person who wants to enter changes his appearance through makeup, the relative distance of his facial features will not change because of this, so he can still correctly recognize it. Moreover, according to the facial feature recognition method of the present invention, it is not necessary to use a field shoot stalk, so there is no recognition error caused by measurement errors. The digital security system according to the present invention can be used for door and door management of many occasions, such as public military police units, large-scale enterprise factories, education bills Z, Ji line number, community, employee access control, general home anti-theft equipment, etc. [Embodiment] Please refer to FIG. 2 for a schematic view of a digital security system according to the present invention. The digital security system 20 according to the present invention includes a three-dimensional image imaging device 22, a three-dimensional image acquisition device 24, a database 26, a three-dimensional image comparison device 28, and a control device 29. The three-dimensional imaging device 22 is used to obtain a three-dimensional image of the face of a person who wants to enter. The three-dimensional image capturing device 24 then recognizes the three-dimensional image of the face, and captures the relative distances between the positions of all facial features in the three-dimensional image of the face. The three-dimensional image comparison device 200530943 is set to 28, and then the relative distance between the facial features captured by the three-dimensional image capture device 24 is compared with the face records stored in the database 26, and the face records stored in the database 26 It contains information on the relative distance between facial features of all persons allowed to enter. Then, the control device 29 decides whether to allow the person who wants to enter, or activates other security measures according to the comparison result generated by the three-dimensional image comparison device 28. Figures 3 and 4 illustrate a digital security system according to the present invention with a preferred embodiment. Fig. 3 is a schematic diagram showing a face recognition method according to the present invention. When an entrant 10 wants to enter the digital security system 20, the three-dimensional image imaging device 22, such as a dual-lens digital stereo imaging camera, images the entrant 10 to obtain a three-dimensional image of the face of the entrant 10. . The three-dimensional image imaging device 22 may be further designed to automatically sense the proximity of the name to the I 10 and prompts that if you want to enter the Y 10, you need to perform the image capture step first. The three-dimensional image capturing device 24 then recognizes the two-dimensional image of the face captured by the three-dimensional image imaging device 22. First, identify the position of each facial feature in the 3D image captured. This facial feature is, for example, eyebrows, eyes, early morning, prpr 甘 # M ^ Auricularia, and the special address of each face is marked. For example, Figure 4 shows how to take-the face is three-dimensionally pure. Here, the face is three-dimensionally f-shaped like a towel, bringing the right eyebrow closer to the heart ... grabbing another point as h, and bringing the left eyebrow closer to the face ^ The point mark is not N, and the other end point is marked as b2. : Similarly, the end point of the right eye near the center of the face can be marked with the other end point as c. The tester Sun soil will be 2 The end point of the left eye near the center of the face is marked as seven, 200530943 The upper endpoint is marked as ... Second, the upper endpoint of the left ear is marked as fl, the lower endpoint is marked as the second dead mountain 77 'tip is not h, the right endpoint of the mouth is marked as gI, and the left endpoint is marked as g2. =, The distance between the two ends of the arbitrary capture, for example, the distance between the end point of the capture and the point 1 to ba. This distance represents the distance between the center ends of the two eyebrows. In this specific embodiment, the endpoints that can be retrieved are a ,, a2, bl, b2, Ci, '62 6, gl, g2, h, so arbitrary endpoints are retrieved = material forms c, / combinations . The monitoring personnel can ^ take the endpoints suitable for ¥ according to the required accuracy. When the more endpoints it holds, the more distance groups can be formed, the more the number of months can be generated. When performing the comparison step, The lower the possibility of identification errors. For example, in the preferred embodiment, for the purpose of illustration, the distances between the seven endpoints are selected as aibi, aic 丨, 4, a 丨, 丨, ~ gi, aih. When the three-dimensional image capturing device 24 obtains the seven endpoint distances of the person who wants to enter, then, the three-dimensional image comparing device 28 stores the seven endpoint distances of all permitted entrants stored in the fish database%. Compare them one by one. In the 3D image comparison device 4 28, you can set the value of the degree of conformity, for example, 'When the distance between the seven endpoints of the person 10 to enter, there are five distances and one stored in the database 26-allowed to enter If the person is the same, it means that the person who wants to enter 10 is the allowed person in the database 26, so he can be allowed to enter. The endpoints of facial features can be selected according to actual needs, and different thresholds for the degree of conformity can be set. For example, when the digital security system 20 is used to control highly confidential or valuable devices, the number of selected facial features can be increased by 200530943 points. When more face feature endpoints are selected, more combinations of possible distances can be generated. At the same time, the threshold of the degree of matching can be increased, and the possibility of identification errors is lower. When the digital security system 20 is used for general access control purposes, the facial feature endpoints captured can be reduced, and at the same time, the threshold of the degree of compliance can be reduced to reduce the time taken for identification. The three-dimensional image comparison device 28 then generates a comparison result, which produces a comparison result that allows entry into the world. The t control device 29 then uses the comparison result to activate various security mechanisms. For example, when the control device 29 receives the comparison result of allowing entry, it allows the person who wants to enter into the room, records the time when the person who wants to enter 10 enters, and stores a three-dimensional image of the face of the person who enters 10. It is used to track the record of people coming out in the future. When the control device 29 receives the comparison result of refusal of entry, it can be mentioned that the entrant 10 is required to perform the photographing operation again to re-identify. Alternatively, when the person who wants to enter 1G is required to re-identify, the staff member informs the person who wants to enter 10 to try to enter as the loser. ★ + _ 疋 1 day failure, you need to observe immediately. Alternatively, it can also be designed to start security measures when identification fails for three consecutive times, such as activating an alarm device or notifying a monitoring staff. Wang Yi ^ 'It can also be combined with other existing identification mechanisms to enter a password mechanism, or more m ^ 5 s, such as identification, iris identification, etc. "Identification of the fingerprint identification digital security system multiple' identification mechanism, to provide more delay Figure 5 is a flowchart to explain the whole system management and control method. The first is a three-dimensional image of the face of a digital enthusiast who wants to enter the fantasy (step 5.). Then, from the obtained three-dimensional 200530943 facial image, the relative distances between facial features are extracted (step 502). Then, the relative distances between these facial features will be captured one by one with the data stored in the database. The face records of the entrants are allowed to be compared (step 504). The face records contain all the relative distances between the facial features that are allowed to enter the salamander. After comparing the relative distances of the facial features of the entrants with the database, a comparison is made. Result. If the result of this comparison is higher than a preset threshold of conformity (step 506), then the monument Abu will be allowed, and the heart will enter; then the brother will enter the eight (step 508). After that, you can proceed- Record this desire The three-dimensional image of the person ’s face, and the time (step 5ig) for the person who wants to enter the security system, are used for future entry and exit records. If the comparison result is lower than the preset door level value, it is required Those who want to enter again perform face three-dimensional image imaging and recognition (step 511). The same as: 'Also according to the use needs and settings'-monitor the personnel of the security system (step 512), or activate the alarm device (step 514) ). According to the digital security system of the present invention, the identity of the person who wants to enter. Even if the person who wants to enter ^ 2 speeds and accurately recognizes the error, the relative distance of the facial features changed by makeup will not change because of this. It can be correctly entered =. = The facial feature recognition method of the present invention does not require the use of a lightning grid. There is no measurement error that causes recognition errors. According to the digital security system of the present invention, x can be used, for example, public military police units. ", Large enterprise factories, teaching" :, system, line number, community, employee access control, @ 早 小 though the present invention has been better than stealing equipment, etc. The conventional examples are disclosed as above, but it is not intended to limit the present invention with 200530943. Any person skilled in this art can make various modifications and retouches without departing from the spirit and scope β of the present invention. Subject to the scope of the attached patent application. [Brief description of the drawings] From the above detailed description of the preferred embodiments of the present invention, the purpose, viewpoint, and advantages of the present invention can be better understood. It is explained with the following drawings of the present invention: FIG. 1 is a schematic diagram showing a face recognition method of a conventional technique. FIG. 2 is a block diagram of the whole system according to the preferred embodiment of the present invention. Fig. 3 is a schematic diagram showing the recognition according to the present invention.辨 Facial Recognition in Specific Embodiments Fig. 4 is a schematic diagram of a three-dimensional image of a face according to the present invention, which is a preferred embodiment. Figure 5 of J Mito is a flowchart showing the official control method of the King system according to a preferred embodiment of the present invention. ', [Simple description of component representative symbols] 10 12 14 20 Laser light source laser grid peach digital security system 3D imaging device 12 22 200530943 24 3D image acquisition device 26 database 28 3D image comparison device 29 control Device
1313