在-實施例中’㈣統可包含—載有使用者或被授權使 用者之臉部影像之資訊的資料庫。 適當地,該系統可經由網際網路而連接至該資 配番+ 吁 ^ *之-優點為可自遠離該系統之位置維護及更新該資料 庫。 。有利地,該系統可包含用於識別使用者之構件。該系統 可包含用於偵測識別構件之構件。識別構件大體可呈卡之 / ^式,該等卡載有該系統之各別被授權使用者之識別資訊 :提供至該系統之各別被授權使用者。或者,偵測構件可 4凋適以偵測各別使用者獨有的其他特徵。 根據本勒明之第二態樣’提供_用於允許或不允許使用 ,進入之門禁系統’其包含如上文所描述之系統。門禁系 可匕3閘道或—門。在確認使用者之真實性後’該閘 道或門可根據臉部辨識系統之指令而打開。 根據本發明之第三態樣,提供藉由使用上文所描述之系 統來鑑別使用者身份的方法。 根據本發明之第四態樣,提供藉由使用臉部辨識系統來 ‘別使用者身份的方法,其包含如下步驟:⑷自試圖通過 d系統之使用者獲取第—組資料;⑻基於該第-組資料判 定使用相機中之哪—者來捕獲使用者之臉部影像;⑷捕獲 使用者之影像;(d)基於所捕獲之影像產生第二組資料; (e)將該第二組資料與載有系統之被授權使用者之識別的資 料庫中之資4進行比較;(f)基於該比較而同步產生第三組 貝料’ & ig)基於該第三組資料,允許或拒絕使用者通過該 I23695.doc 1338866 系統。在一實施例中,該方法可進一步包含在顯示構件中 顯示使用者之臉部影像的步驟。 【實施方式】In the embodiment, the '(4) system may include a database containing information on the facial images of the user or authorized user. Suitably, the system can be connected to the resource via the Internet - the advantage is that the database can be maintained and updated from a location remote from the system. . Advantageously, the system can include components for identifying the user. The system can include components for detecting the identification member. The identification members can generally be in the form of a card containing identification information for each authorized user of the system: to each authorized user of the system. Alternatively, the detection component can be adapted to detect other features unique to the individual user. According to the second aspect of the present invention, 'providing _ for allowing or disallowing access to the access control system' includes the system as described above. The access control system can be used for 3 gates or gates. After confirming the authenticity of the user, the gateway or door can be opened according to the instructions of the face recognition system. According to a third aspect of the present invention, there is provided a method of authenticating a user's identity by using the system described above. According to a fourth aspect of the present invention, there is provided a method for 'individual user identity' by using a face recognition system, comprising the steps of: (4) obtaining a first set of data from a user attempting to pass the d system; (8) based on the first - the group data determines which of the cameras is used to capture the user's facial image; (4) captures the user's image; (d) generates a second set of data based on the captured image; (e) the second set of data Comparing with the capital 4 in the database containing the identification of the authorized user of the system; (f) synchronizing generation of the third group of beryllium ' & ig based on the comparison, based on the third set of data, allowing or rejecting The user passes the I23695.doc 1338866 system. In one embodiment, the method can further include the step of displaying a facial image of the user in the display member. [Embodiment]
如上文所述,臉部辨識系統已存在一些時間。用於將 所捕獲之使用者之臉部影像與儲存在資料庫中之影像進行 比較並接著判定έ玄使用者是否為被授權使用者之電腦軟體 並不陌生。習知之臉部辨識系統通常設計有一終端,使用 者藉由該终端與該系統相互作用。在使用時,使用者站在 终端前,以便駐留在終端處之相機可捕獲使用者的臉部影 像。然而’因為該终端通常固定地安裝在牆壁上或柱體 上,且在該終端處將相機固定至距地面之某一高度,所以 在影像捕獲處理期間常常出現㈣。因使用者有不同身高 =機被固定地安裝,所以適當地捕獲使用者之臉部影像 吊吊不令人滿意。本發明或如下文所描述之根據本發明之 實施例試圖解決與此問題相關聯之問題。As mentioned above, the face recognition system has been around for some time. It is no stranger to compare the captured user's face image with the image stored in the database and then determine whether the user is an authorized user's computer software. Conventional facial recognition systems are typically designed with a terminal through which the user interacts with the system. In use, the user stands in front of the terminal so that the camera residing at the terminal can capture the user's facial image. However, because the terminal is usually fixedly mounted on a wall or a cylinder and the camera is fixed at a certain height from the ground at the terminal, it often occurs during the image capturing process (4). Since the user has different heights = the machine is fixedly mounted, it is unsatisfactory to properly capture the user's facial image. The present invention or embodiments of the present invention as described below attempt to solve the problems associated with this problem.
一 q〜矛一I孢例之臉部辨識系統 I:使用者與其相互作用之终端2。終端以合適高度固定 :安裝在—牆壁上。圈1及圖2分別展示展示终端2之正面 =透視圖之示意圖。終物狀大體為長方形,且垂直 牆壁上。自終端2之上端至下端,提供-第-影 件卜:彳(諸如相機4)、用於顯示使用者之臉部影像之構 什(睹如監視器6)、_坌_呈 $i m 第一衫像捕獲構件(諸如相機8)、複 數個指不器10及用於偵洌 E. 19, ^ 使用者之識別的構件(諸如偵測 為12)。監视器6具有多種功能, 。亥4功能中之一功能為向 I23695.doc 資訊概況之部分。為解釋該系統之特 境中假定使用者係類独之人。 ’、作’在如下情 參看圓3,處於其待用捃 .、 、传用槟式之糸統14準備好(類型!·!# ® 者之)臉部進行辨識或驗證。 (類抑使用 於禎涔6滲技 田使用者走近終端2時,將在 BL視益6處長:示其在伯,目丨丨哭丨。士 ,± 彺偵測夺12處登錄其識別卡。作為0 應’使用者將其_卡置 接著首先試圖驗證該卡是…二2處(見16),器 右崎丰 卞疋否為有效卡(見〗8)。若該卡非 有效卡’則其將輸出信號 在就至里視裔6用於將錯誤訊息顯示 在皿視益6上(見20),且李 22) 系、先接耆返回至其待用棋式(見 ) 右該卡為有效卡,則:Β:將冲六―丄 操取其他資訊並確定'之將/圖自儲存在卡中之概況 驗證之人的身高類型(亦即,類型Η s ,之使用者)(見24)。接著當該人被識別為類抑時 (如上文在此實施例中所指示,使用者係類❹之人且識別 :在資讯概况中向其指派此類型Η資訊),偵測器將輸出一 信號以便啟動位於較高位置之第一相機4(見2〇) 一旦相機 4啟動,其將經調適以在相機距離(自,約5,6”至6,)捕獲使用 者之臉部影像。在捕獲影像之後,所捕獲之影像將轉換為 資》fl以供與資料庫中之資訊進行比較’從而試圖確定具有 該特定所捕獲臉部影像之人是否為被授權通過該系統^人 (見28)。右發現所捕獲之臉部影像並不對應於記錄上之任 何資訊’則系統將中止辨識處理(見3〇)並輸出信號以將錯 决訊息提不至監視器(見32)’且接著立即返回至其待用模 式(見34)。該系統亦將與此,,不成功"辨識有關之資訊登錄 在日誌表上為供將來參考(見36)。若發現所捕獲之臉部影 I23695.doc I338866 像不對應於被授權使用者之資訊’則系統將輸出―信號將 個人化問候訊息提示至監視器(見38)。在系統連接至入口 間道之情境中,該系統亦可經調適以輪出一用於自動開啟 該閉道或打開該閘道之信號。該系統同樣亦將與此,,成功” 辨識有關之資訊登錄在日誌、表上為供將來參考(見,。系 統接著返回至其待用模式用於下一辨識實施(見42)。 ’、 當然’當系統用以辨識類η之被授權使用者的臉部影 像時’將替代地啟動下部相機8’但其餘演算法大雜類 似。 上文所描述之系統在多種方式上係技術性重大的。首 先,在習知臉部辨識系統中僅提供—相機。若相機相對於 使用者過高地安裝在一固定物上’則相機傾向於捕獲使用 者臉部之頂部透視圖’或者若使用者很矮,則相機傾向於 捕獲使用者前額之影像。應理解,連接至系統之資料庫載 有使用者(或被授權之使用者)臉部影像之資訊,且若所捕 獲之影像並非真正為使用者之臉部影像而是其前額影像, =管使用者很可能為被授權使用者,但辨識系統將不能 適备地區別使用者並允許該使用者通過該系統。若相機相 對於使用者過低地安裝在一固定物 使用者臉部之下部透視圖,或者若使用=機傾向於捕獲 韦右便用者很鬲,則相機傾 ==用者頦之影像。再者’此將類似地導致不成功 雙相機配置目反㈣上文所描述之實施例’藉由使用 雙相機配置’該系統能夠在兩個相機之間選擇適 以更精確地捕獲使用者之臉部影像。 田 123695.doc 10- 1338866 弟二’在習知臉部辨識系統中無用於向使用者提供反饋 之構件可用。通常情況為,使用者並未站在相機正前方, 且未意識到其站在對於辨識處理不適當之位置,從而該系 、·’先將不此適s地執行辨識實施。相反,根據上文所說明之 貫施例藉由使用一監視器,使用者可看到在其走近終端 時其影像是否出現在螢幕中。歸因於使用者在監視器中尋 找其影像之自然傾向,使用者會在終端前移動自身直至其A face recognition system of a spear-I spore case I: a terminal 2 with which the user interacts. The terminal is fixed at a suitable height: mounted on the wall. Circle 1 and Figure 2 show a schematic view of the front side = perspective view of the display terminal 2, respectively. The final shape is generally rectangular and vertical on the wall. From the upper end to the lower end of the terminal 2, a -photo-image is provided: 彳 (such as the camera 4), a display for displaying the facial image of the user (such as the monitor 6), _坌_ is presented as $im A shirt is like a capture member (such as camera 8), a plurality of fingers 10, and a member for detecting E. 19, ^ user identification (such as detection of 12). The monitor 6 has a variety of functions. One of the features of the Hai 4 function is part of the I23695.doc information profile. To explain the characteristics of the system, it is assumed that the user is unique. ‘,作’ See the circle 3 in the following case, and use it for identification or verification of the face of the 槟 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 14 (When the class is used, the user of the 渗6 技 技 田 走 走 走 走 , , 使用者 使用者 使用者 BL BL BL BL BL BL BL BL BL BL BL BL BL BL BL BL BL BL BL BL BL BL BL BL BL BL Identification card. As 0 should be 'users set its _ card and then first try to verify that the card is ... two (see 16), whether the right is a valid card (see 8). If the card is not valid The card's output signal is then used to display the error message on the dish 6 (see 20), and the Li 22) system is first returned to its pending game (see) Right, the card is a valid card, then: Β: will take the other information and determine the type of height of the person who verified the profile from the card stored in the card (ie, type Η s, use () (see 24). Then when the person is identified as class-like (as indicated above in this embodiment, the user is a class of people and identifies: assigning this type of information to the information profile), the detector will output A signal to activate the first camera 4 at a higher position (see 2〇). Once the camera 4 is activated, it will be adapted to capture the user's facial image at camera distance (from, about 5, 6" to 6,). After capturing the image, the captured image will be converted to "followed for comparison with the information in the database" to attempt to determine whether the person with the particular captured facial image is authorized to pass the system ( See 28). Right find that the captured facial image does not correspond to any information on the record' then the system will abort the identification process (see 3〇) and output a signal to lift the wrong message to the monitor (see 32). 'And then immediately return to its standby mode (see 34). The system will also log in with the information about this, unsuccessful " identification on the log table for future reference (see 36). If found Face shadow I23695.doc I338866 does not correspond to being taught The user's information 'The system will output the signal to prompt the personalized greeting message to the monitor (see 38). In the context of the system connected to the entrance lane, the system can also be adapted to rotate one for automatic opening The signal that closes or opens the gateway. The system will also log in with the information about the successful "identification" in the log and table for future reference (see, the system then returns to its standby mode for Next identification implementation (see 42). 'And of course' when the system is used to identify the face image of an authorized user of class η, 'the lower camera 8' will be activated instead, but the rest of the algorithms are similar. The described system is technically significant in a number of ways. First, in the conventional facial recognition system, only the camera is provided. If the camera is mounted too high on the fixture relative to the user, the camera tends to capture the user. The top perspective of the face' or if the user is short, the camera tends to capture the image of the user's forehead. It should be understood that the database connected to the system contains the user (or is authorized to make it The information of the facial image, and if the captured image is not the user's facial image but the forehead image, the tube user is likely to be an authorized user, but the identification system will not be suitable Distinguishing the user and allowing the user to pass through the system. If the camera is mounted too low on the underside of the face of the user's face relative to the user, or if the use of the machine tends to capture the Wei right, the user is embarrassed. Then the camera tilts == user's image. Again, this will similarly lead to unsuccessful dual camera configuration. (4) The embodiment described above 'by using a dual camera configuration' the system can be used in two cameras The choice is to capture the user's facial image more accurately. Tian 123695.doc 10- 1338866 Dior's no component for providing feedback to the user in the conventional facial recognition system. Usually, the user is not standing in front of the camera, and is unaware that the station is not in a position suitable for the identification processing, so that the system will perform the identification implementation without first. In contrast, according to the embodiment described above, by using a monitor, the user can see whether or not his or her image appears on the screen as it approaches the terminal. Due to the natural tendency of the user to find their image in the monitor, the user will move itself in front of the terminal until it
正面景/像充分在監視器中顯示。監視器之引入本質上產生 生物反饋’藉此使用者可因辨識系統之作用而被引至適當 位置。 根據本發明之第二實施例與第一實施例類似,除了監視 器經調適以顯示使用者之鏡像,使得隨著使用者向右移 “V像將跟隨並向右移動,且反之亦然。因為使用者 可跟隨影像之移動並相應地調整其位置,所以此設計係有 利的。The front view/image is fully displayed on the monitor. The introduction of the monitor essentially produces biofeedback' whereby the user can be brought to the appropriate location by the action of the identification system. A second embodiment in accordance with the present invention is similar to the first embodiment except that the monitor is adapted to display a mirror image of the user such that as the user moves to the right the "V image will follow and move to the right, and vice versa. This design is advantageous because the user can follow the movement of the image and adjust its position accordingly.
根據本發明之第 供用於顯示使用者 其仍係有利的,因 不同身高之使用者 二實施例與第一實施例類似,除了未提 之影像之構件。雖然此實施例欠佳,但 為使用兩個相機至少可處理較寬範圍之 很锞本發明之第四實施例與 供— 貫轭例類似’除了僅 供相機。雖然此實施例欠佳 用於生物β n A 彳一 /、仍係有利的,因為 M ^ ^ l 任下’使用者至少可在 視益輔助下對自身進行定 7在 枘诚+ ν 統進行辨識。 务明之第五實施例與第一 I她例類似,除了積 123695.doc 1338866 盗經調適以辨識被授權使用者之指紋>指紋辨識可用於(i) 識別特定被授權使用者係哪一身高類型;及/或(π)識別使 用者究竟是否為被授權使用者。The invention according to the invention is still advantageous for displaying the user, because the user of different heights is similar to the first embodiment except for the components of the image which are not mentioned. Although this embodiment is not preferred, at least a wide range can be handled for the use of two cameras. The fourth embodiment of the present invention is similar to the yoke yoke except that it is only for the camera. Although this embodiment is not good for the biological β n A / / , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , Identification. The fifth embodiment of the invention is similar to the first example of I, except for the accumulation of 123695.doc 1338866 piracy adaptation to identify the fingerprint of the authorized user> fingerprint identification can be used to (i) identify which height of a particular authorized user is Type; and/or (π) identifies whether the user is an authorized user.
可設想,在另一實施例中,具有臉部影像或被授權臉部 影像之資料庫可位於遠端位置且終端經由網際網路連接至 忒資料庫。因許多原因,此係有利的。首先,可在遠端位 置進行資料庫之維蠖。第二,因資料庫在遠端位置,可避 免盜竊終端處之個人資訊。第三,f料庫可能包括罪犯之 臉部影像,且若辨識系統谓測到此等罪犯,該系統可經調 適而自動地通知執法機關。 亦可設想,臉部辨識系統可包含在n禁系統中,用於允 許或拒絕經由該系統之入口進入建築物。臉部辨識系統可 包含在m統(諸如自動櫃員機)中’藉此允許使用者僅 在其已通過臉部辨識系統之後執行任何交易。 應注意,上文各種演算法之執行可 ^仃】經由電路、軟體程式It is contemplated that in another embodiment, a library of facial images or authorized facial images may be located at a remote location and the terminal is connected to the database via the Internet. This is advantageous for a number of reasons. First, the database can be maintained at the remote location. Second, because the database is located at a remote location, personal information at the terminal can be avoided. Third, the f-library may include facial images of the offender, and if the identification system detects such offenders, the system may be notified and automatically notified to the law enforcement agency. It is also contemplated that the face recognition system can be included in the n-ban system for allowing or denying access to the building via the entrance to the system. The face recognition system can be included in the system (such as an automated teller machine) thereby allowing the user to perform any transaction only after they have passed the face recognition system. It should be noted that the execution of the various algorithms above can be performed via circuits and software programs.
化、勒體程式化或其他硬體程式化 ϋ吨進仃,且使用任一執 订模式白並不偏離本發明之發明特徵。 【圖式簡單說明】 實知例之臉部辨識系統或門 圖;及 貫施例之系 矛'統彳呆作邏輯的流 圖1係說明根據本發明之一 禁系統之終端之透視圖; 圓2係圖1所示之終端之正面 圖3係說明根據本發明之一 程圖。 【主要元件符號說明】 123695.doc 12 1338866 2 終端 4 相機 6 監視器 8 相機 10 指示器 12 偵測器 123695.doc -13-The simplification, stylization, or other hardware stylization, and the use of any of the binding modes, does not deviate from the inventive features of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS A face recognition system or a door diagram of a known example; and a flow diagram of a lancet of a conventional embodiment; FIG. 1 is a perspective view of a terminal of a forbidden system according to the present invention; Circle 2 is a front view of the terminal shown in Fig. 1. Fig. 3 is a block diagram showing a process according to the present invention. [Key component symbol description] 123695.doc 12 1338866 2 Terminal 4 Camera 6 Monitor 8 Camera 10 Indicator 12 Detector 123695.doc -13-