TW466452B - Method and apparatus for photographing/recognizing a face - Google Patents
Method and apparatus for photographing/recognizing a face Download PDFInfo
- Publication number
- TW466452B TW466452B TW88119671A TW88119671A TW466452B TW 466452 B TW466452 B TW 466452B TW 88119671 A TW88119671 A TW 88119671A TW 88119671 A TW88119671 A TW 88119671A TW 466452 B TW466452 B TW 466452B
- Authority
- TW
- Taiwan
- Prior art keywords
- face
- image
- viewing area
- scope
- patent application
- Prior art date
Links
Landscapes
- Closed-Circuit Television Systems (AREA)
- Image Processing (AREA)
Abstract
Description
V，146 6 4E 五、發明說明（1) 本發明係有關一種藉由一部攝影機將其所攝取到的物 件’機智地拍攝其照片的臉部攝影及辨識之方法與裝置。 本發明尤指一種，可藉由將一部能按照該物件動作而 自動調整焦距的攝影機來精確拍攝及辨識該人士的臉孔， 進而識別特定人士’以及對危險人物發出警報訊號，以便 月t應用於保全系統、影像辨識系統、影像電話機、網際網 路連線遊戲機及虛擬實境系統等的，臉部攝影及辨識之方 法與裝置。 為了要拍攝一張臉孔，因而在一典型的無人監控或臉 部辨識系統中，則須在合適的定位處，諸如天花板上或牆 壁上，安裝至少一部以上的監視用攝影機。藉由使用該攝 景^機所傳來的影像訊號，該典型的系統因而能夠從事遠端 ^現特定區域或是以無人操作的方式來記錄該區域的整個 尽象。上述的系統通常被用來判斷—特定區域的狀況。 然而 統應該要 區域的影 之間的距 外，對於 件亦可能 由其正前 並配合使 則該典型 困難。因 定位置（例如 於物件高度或 動’因此該影 者迅速地移動 。能夠加以辨 物件或是以人 裝設—方向調 若使用上述無 移動晝面中的 月匕包由一指 像訊號。由 離會有所變 一個隱藏或 會難以辨識 方來拍攝該 用遙控器或 因此， 的系統對於 a巩糸統或臉部辨識系 ’高度或距離）來捕捉該 疋位於s亥物件和攝影機 像訊號係不規則的。此 到攝影機鏡頭之外的物 識的影像訊號係能藉著 工方式來調整攝影機， 整鍵等方式來獲得。 法辨識的影像資料時， 臉孔在辨識上將會發生 ^46645, 五、發明說明（2) 為在揭取的t ,, i k 0Λ 辟存貧料中無法找到完整的於π π你4 照上述方法跖從π 正的臉孔影像畫面，依 μ + 所羧得之影像資料並不完令。+冰上 儲存的資料類创g釭$老二 门 几王。此外，由於其所 I沾古—Γ 疋動態晝面，因而必須要準備一屌太 =儲存裝置，以便儲存上述動態晝面。 ά τ二阳λ服上述問題，有必要開發—項新技術以便浐蕤 由可按照物件動作而銘叙谣專，她—+ 只丨议町从使把猎 攝位在前方之臉孔影像。 才 =明之主要目的在於提供一種拍攝臉孔的方法盘裝 ▲置，其中，精由偵測攝影機所攝取到之物件膚色和動作， 迠拍攝到臉孔的靜止畫面，能自動選擇最適當的物件，並 可讓攝影機自動追蹤該物件而不需依賴任何人工 及一種辨識臉孔之方法與裝置，其中，藉由分析該攝取物 件之臉孔影像，即可準確地辨識特定人士。 本發明之另一目的在於提供一種辨識臉孔之方法與裝 置’可透過分析邊選擇物件的臉孔影像訊號，而識別具有 危險要素的臉孔（諸如，奇怪的面罩、深色太陽眼鏡、下 壓的帽子等），並能對可危險的人物發出警報，故能預先 處理任何可能會發生的緊急事件。 本發明之一實施例提供一種拍攝臉孔的方法，其包 括： 藉由預先輸入於一安裝在預設位置的攝影機中的影像 訊號中所取得之臉孔的取樣資料’進而確認該臉孔位置’ 然後再選定檢視區域以便能追蹤需要識別之臉孔的初步分 析步驟；V, 146 6 4E V. Description of the invention (1) The present invention relates to a method and a device for facial photography and recognition of an object ′ taken by a camera wisely to take pictures of it. The invention particularly relates to a camera that can accurately capture and recognize the face of a person by using a camera capable of automatically adjusting the focal length according to the movement of the object, thereby recognizing a specific person 'and issuing an alarm signal to a dangerous person, so that Methods and devices for facial photography and recognition applied to security systems, image recognition systems, video telephones, Internet-connected game consoles, and virtual reality systems. In order to capture a face, in a typical unmanned surveillance or face recognition system, at least one surveillance camera must be installed at a suitable location, such as on a ceiling or wall. By using the image signal from the camera, the typical system can therefore work remotely to a specific area or to record the entire image of the area in an unmanned manner. The above-mentioned system is usually used to judge the condition of a specific area. However, the system should require the distance between the shadows of the area, and it is typically difficult for the parts to be brought forward and cooperate with each other. Due to a fixed position (such as at the height or movement of the object), the photographer moves quickly. The object can be identified or installed by a person-the direction is adjusted if the moon dagger in the non-moving daytime surface is used as a signal. The distance will change from a hidden or difficult to recognize side to shooting with a remote control or, therefore, the system's height or distance to the system or face recognition system to capture the object and camera image The signal is irregular. The image signal from the outside of the camera lens can be obtained through manual adjustment of the camera, full keys, etc. When identifying the image data, face recognition will occur ^ 46645, V. Description of the invention (2) For t, ik 0Λ, you cannot find a complete picture of π π you 4 photos The above method does not complete the order from the image of the positive face image based on the image data obtained by μ +. + The data category stored on the ice will be $ 2, the second child, and the few kings. In addition, because of its dynamic diurnal surface, it is necessary to prepare a storage device to store the dynamic diurnal surface. ά 阳 二 阳 λ is necessary to develop the above problems, it is necessary to develop a new technology in order to describe the rumors by following the action of the object, she— + only the image of the face of the head of Yoshimachi from hunting. The main purpose of Cai = Ming is to provide a method of photographing faces. The device is to detect the skin color and movement of objects captured by the camera, and 静止 to capture the still images of faces, and automatically select the most appropriate objects. , And allows the camera to automatically track the object without relying on any human and a method and device for identifying faces. Among them, by analyzing the face image of the captured object, a specific person can be accurately identified. Another object of the present invention is to provide a method and device for identifying faces, which can identify faces with dangerous elements (such as strange masks, dark sunglasses, Pressure hats, etc.), and can alert dangerous people, so it can deal with any emergency that may occur in advance. An embodiment of the present invention provides a method for photographing a face, including: confirming the position of the face by sampling data of the face obtained in advance from an image signal input in a camera installed in a preset position. 'Then select the viewing area so that you can track the initial analysis steps of the face to be identified;
第6頁 4 6 645 2Page 6 4 6 645 2
'上/下或前/後等方向移動， 中需要識別臉孔的傾斜式轉盤或 五 '發明說明（3) 將攝影機朝著左/右 以便能追縱位在選定區域 變焦距式鏡頭的控制步驟 藉由取樣及分析位於檢 以便判斷出該臉孔為已存在 可儲存已判斷過之臉孔 直接傳送至另一套辨識系統 該實施例進一步包括5 區域中的臉部輪廓後，即可 驟。 視區域中要識別臉孔的輪廊， 或為異常之判斷步驟； 影像資料，或可將該影像資料 之儲存及傳送步驟。 在完成分析及判斷取樣自檢視 針對異常的臉孔發出警報的步 括： 本發明之另一實施例提供— 種辨識臉孔的方法 其包 藉由預先輸入於一安裝在預設位置的攝影機中的影 訊號中所取得之臉孔取樣資料，進而確認該臉孔的位置， 然後再运定檢視區域以便能追蹤需要識別之臉孔的初步分 析步驟； 將攝影機朝著左/右、上/下或前/後等方向移動， 以便能追蹤位在選定區域中需要識別臉孔的傾斜式轉盤或 變焦距式鏡頭的控制步驟； 藉由取樣及分析位於檢視區域中要識別之臉孔，以便 判斷出該臉孔為已存在或為異常之判斷步驟；假使該臉孔 已存在時，則它便可藉由在檢視區域中產生一個三次元的 臉孔影像資料之方式來構成一個個人的面模，然後藉由互 相比較此個人的面模和先前所儲存之個人的面模來辨識一'Move up / down or front / back and other directions, the tilt dial or five that needs to recognize faces in the middle' Description of the invention (3) Point the camera to the left / right so as to track the control of the zoom lens in the selected area Steps: Sampling and analysis are used to determine whether the face is already present. The judged face can be stored and transmitted directly to another recognition system. This embodiment further includes facial contours in 5 areas. . In the viewing area, the contour of the face to be identified may be an abnormal judgment step; the image data may be stored and transmitted. The steps for completing the analysis and determining the sampling self-examination to send an alarm to an abnormal face include: Another embodiment of the present invention provides a method for identifying a face, which includes inputting in advance a camera installed in a preset position The face sampling data obtained in the video signal of the camera, and then confirm the position of the face, and then determine the viewing area so as to track the preliminary analysis steps of the face to be identified; point the camera to the left / right, up / down Or forward / backward direction to track the control steps of the tilt dial or zoom lens that needs to identify faces in the selected area; by sampling and analyzing the faces to be identified in the viewing area, to determine It is judged that the face is already existing or abnormal; if the face already exists, it can form a personal facial model by generating a three-dimensional face image data in the viewing area. , Then identify a person ’s facial model and the previously stored personal facial model by comparing
第7頁 五、發明說明（4) 位特定的人 本發明 以下的步驟 利用一 區域，籍由 中之方式來 其所選定區 —個臉孔是 偵測其影像 理過的影像 著在資料庫 的個人臉孔 經註冊登記 本發明 以下的步驟 從利用 行各個膚色 視區域，藉 相關臉孔元 選擇一個特 特性之方式 以及判斷一 向；對已經 訊號相關的 4 6 64 5 ；： 士。 之又-實施例提供一種辨識臉孔的方法，包括 部攝影機來攝取其中相關影像訊號的多個檢視 判斷-個臉孔是否已經存在於相對應檢視區域 選擇一個特定的檢視區域，並按照一個物件在 域中的動作來調整攝影機的方向；在完成判斷 否存在之後，在存有該臉孔之檢視區域内進行 訊號之特性以便重新整理其影像，依照重新整 之間的移動結果來產生三次元的資訊；並且藉 内的個人的面模註冊檔案中進行搜尋其所構成 面楔之方式來判斷某一特定人士的影像是否已 過了。 之再一實施例提供一種辨識臉孔的方法，包括 一部攝影機所輸入的影像訊號中，藉著分別進 資料及動作資料的取樣作業方式來選擇多個檢 著檢查位在相對應的檢視區域中的影像訊號之 件’並透過恢復一個特定檢視區域之方式，來 定檢視區域，藉著恢復已經選定之檢視區域之 來重新理影像，並配合使用重新整理的影像 個臉孔是否存在之方式來進行調整攝影機之方 存有一個臉孔的檢視區域重新進行整理和影像 影像資料，依照重新整理過的影像之間的移動 藉由互相比較三次元的資訊 個個人的臉孔面模；並且藉 冊登記面模，進行判斷某— 登記。 發明亦提供一種拍攝臉孔之 裝於一預設位置的攝影機所I 一連串的影像訊號之影像輪 構上的攝影機朝著左/右、 傾斜式轉盤或變焦距式鏡頭 式轉盤或變焦距式鏡頭的驅| 地追蹤該物件的動作，並可 機構的影像訊號中之該物件 視區域’亦可自位在檢視區 及能儲存和顯示該臉部之主 機構中的影像訊號或自影像 機構；以及一用來將影像訊 之相對應臉部的顯示機構。 步包括：一種能在完成分析 部輪廓後’即可以針對異常 成裝置。 塊圖。 主控制器的功能性方塊圖。 的信號流程圖。 …46645 2 五、發明說明（5) 結果來產生三次元的資訊，並 和參考資料之方式，來構成一 著搜尋位於資料庫中之個人註 特定人士的影像是否已經註冊 為了要完成上述目的，本 装置’包括：一用來榻取由安 拍攝到之物件影像，並能接收 入機構；一可將該影像輪入機 上/下或前/後等方向移動之 的驅動機構；一用來控制傾斜 動機構’因而能讓攝影機順利 藉由偵測先前輸入於影像輸入 的膚色和動作，來選擇一個檢 域之影像訊號中擷取臉部，以 控制機構；一用來儲存主控制 訊號中擷取臉部之儲存或擷取 號輸出至由主控制機構中輸出 該拍攝臉孔之裝置可進一 及判斷取樣自檢視區域中的臉 的臉孔發出警報訊號的警報生 【圖式簡單說明】 第1圖係本發明之概要方 第2圖係第1圖中所示之 第3圖係一影像輸入程序Page 7 V. Description of the invention (4) Specific people The following steps of the present invention use an area to select the selected area by the method-a face is to detect the image of its image in the database The personal face of the present invention is registered. The following steps of the present invention are to select a specific feature and judge the way by using the relevant face element by using each skin color viewing area; for the already related 4 6 64 5;: taxi. Another embodiment provides a method for identifying a face, including a camera to capture multiple views of related image signals to determine whether a face already exists in a corresponding viewing area, and selecting a specific viewing area, and according to an object Action in the domain to adjust the direction of the camera; after determining whether it exists, perform the characteristics of the signal in the viewing area where the face is stored in order to rearrange its image, and generate a three-dimensional element according to the result of the movement between the reformations And to determine whether a particular person ’s image has passed by searching the individual facial mask registration files in the individual facial mask registration file. Yet another embodiment provides a method for identifying a face, including selecting a plurality of inspection and inspection positions in a corresponding viewing area by sampling operations of data and motion data in an image signal input by a camera. The image signal in the image and the method of restoring a specific viewing area to determine the viewing area. By restoring the selected viewing area to reorganize the image, and using the method of reorganizing the face of the image To adjust the camera, there is a face view area to reorganize and reorganize the image data. According to the movement between the reorganized images, compare the three-dimensional information of each individual's face mask; and borrow Register the face mold to make a judgement—registration. The invention also provides a camera mounted on a preset position for shooting faces. The camera on the image wheel structure of a series of image signals is facing left / right, a tilting dial or a zoom lens dial or a zoom lens. The tracking of the object can be tracked, and the object's view area in the image signal of the organization can also be located in the viewing area and the image signal in the main organization that can store and display the face or from the image organization; And a display mechanism for matching the image information to the face. The steps include: a device that can be targeted for abnormalities after the outline of the analysis section is completed. Block diagram. Functional block diagram of the main controller. Signal flow chart. … 46645 2 V. Description of the invention (5) The result is to generate three-dimensional information and the method of reference materials to constitute a search for personally located images in the database Note whether the image of a specific person has been registered in order to accomplish the above purpose, The device includes: a driving mechanism for taking an image of an object photographed by An and receiving the mechanism; a driving mechanism for moving the image into the machine up / down or front / back direction; The 'tilt control mechanism' thus allows the camera to successfully capture the face from the image signal of a detection area by detecting the skin color and motion previously input in the image input to control the mechanism; one is used to store the main control signal The stored or retrieved number of the captured face is output to the device that outputs the photographed face in the main control mechanism, and it can determine the alarm signal from the face of the face sampled from the viewing area. Fig. 1 is an outline of the present invention. Fig. 2 is an image input program. Fig. 3 is an image input program.
*4 6 64s 2 五、發明說明（6)* 4 6 64s 2 V. Description of the invention (6)
第4圖係〜^ A π . 切步分析程序的信號流程圖。 第5圖係〜主丨& β斷程序的信號流程圖。 第6圖係在％ y &首 〜傾斜式轉盤或變焦距式鏡頭的控制程庠 中的一個螢幕以爲 ^ 第7圖係當及—個勞幕的活動位置等之參考圖。 .' 物件被攝取至一拍攝區域時’由適當距 離拍！-位在蝥幕中心端的物件之程序參考圖。 第8圖係一儲存及傳送影像程序的功能性方塊圖。 第9圖係一可用於擷取一臉孔部位之臉孔樣板（a)及 一輸入影像（b)的圖式。 第1 0圖係用來說明攝影/辨識臉孔之方法的處理影 像訊號之概要步驟流程圖。 第1 1圖係一臉孔資料標準化程序的流程圖。 第1 2圖係一臉孔資料輪廓取樣陣列。 第1 3圖係一依照不同亮度來調整臉孔資料輪廓強度 的流程圖。 第1 4圓係一眼部周圍之樣板。 第1 5圖係一嘴部周圍之樣板（a)及輪廓資料（b)。 第1 6圖係本發明另一實施例’拍攝/辨識臉孔之方 法的方塊圖。 第1 7圖係如第1 6圖所示之拍攝/辨識臉孔之方法 的流程圖。 第1 8圖係如第1 6圖所示之另一拍攝/辨識臉孔之 方法的流程圖。 【圖式標號說明】Figure 4 is a signal flow chart of ~ ^ A π. Fig. 5 is a signal flow chart of the main & Fig. 6 is a screen in the control program ％ of the y & first ~ tilting dial or zoom lens. ^ Fig. 7 is a reference picture of the active position of a curtain. . 'When an object is captured into a shooting area ’, shoot from a proper distance! -Process reference drawing of the object located at the center of the screen. FIG. 8 is a functional block diagram of a program for storing and transmitting images. Figure 9 is a diagram that can be used to capture a face template (a) and an input image (b) of a face part. Fig. 10 is a flowchart showing the outline steps of processing image signals for the method of photographing / recognizing faces. Figure 11 is a flowchart of the face data standardization process. Figure 12 is a contour sampling array of face data. Figure 13 is a flowchart of adjusting the intensity of the contours of face data according to different brightness. The 14th circle is a template around the eye. Figure 15 shows the template (a) and outline data (b) around a mouth. FIG. 16 is a block diagram of a method of capturing / recognizing a face according to another embodiment of the present invention. FIG. 17 is a flowchart of a method for capturing / recognizing faces as shown in FIG. 16. FIG. 18 is a flowchart of another method for capturing / recognizing faces as shown in FIG. 16. [Schematic label description]
Λ 6 B 45 2 五、發明說明（7) 10---- 影像輸入裝置 20 —— -驅動裝置 3 0---- 主控制器 31--- 一初步分析裝置 3 2---- 判斷裝置 33 —— -物件管理單元 3 4---- 控制裝置 35 —— —影像儲存及傳送裝置 3 6---- 臉孔辨識及擷取装置 4 0--- —顯示裝置 5 0---- 儲存或擷取裝置 6 0--- —警報生成裝置 100 —— 影像輸入裝置 110-- -CCD攝影機 120 —— 影像擷取裝置 130-- -攝影機驅動控制裝置 140 —— 攝影機驅動襄置 20 0 -- 一預處理單元 3 0 0--- 特性取樣裝置 400-- -資訊取樣裝置 500 —— 70 0--- 面模產生裝置 形狀辨識裝置 600-- -資料庫 ^發明之較佳實施例將配合圖式詳細說明如下： 至1 ί -15圖所示’為本發明之一實施例。而第16 至1 8圖所不，則為本發明之另一實施例。 如第1圆所示，本發明之裝置， 里 1 〇，係利用一裝設於葙w 〜像輸入裒置 以接收-連串f: ! 置的攝影機來攝取物件的影像 輸入裝㈣上的攝影機可：亦^括-可使上述影像 移動之傾斜式轉盤或變 $、上/下或前/後方向 也包括-用來控制該驅動裝^碩的驅動裝置2G。該裝置 ==件動作之主控制器 ^便能讓該攝影機順利 五、發明說明（8) ~~ ---— -- 之檢視區域中的影像訊號擷取臉部，然後再將此 —顯示裝置4°。本發明之裝'置亦提供: ,儲存或擷取裝里50，用以從主控制器3〇中儲存影 或自影像訊號中擷取臉部。該顯示裝置4〇係用後像1 號輪出至由主控制器30傳輪至相對應的臉部。 京^诼Λ ，除此之外，本發明之裝置亦可包括一能利用虛擬顯示 器來產生警報訊號的警報生成裝置6〇，以及一可依照從 控制器3 0端榻取臉孔之結杲來運作之音響裝置。 影像輸入裝置1 0可藉由設置於外側之傾斜式轉盤或變 焦距式鏡頭的j動裝置2〇，而能分別移動至一個朝向左/ 右、上/下或前/後的方向。此外，影像輸入裝置1〇還包 括-部攝影機（即此發明中的數位式攝影機），用以將鏡頭 所摘取到的影像直接轉換成—個定義成數位 或Yim格式的影像訊號。影像輸人裝置10也可包括一U 處理裔’肖以處理一般框架間隔t的影像訊號，並提供一 連串的數位式影像資料（即每秒丨5〜3 〇個框架）。 ^第2圓所不，係第工圖所示主控制器3 〇的功能性内 部Θ構之方塊圖。如圖所示，主控制器3〇包含一個初步分 著由影像輸入裝置10所輸入之影像膚色或動 k仃7 ·以選擇一檢視區域π主控制器3 0也同時包括 -個可自，視區域操取臉孔輪廊，並能將此輪廓拿來與預 f的臉孔J板互作比較’並能依照比較結果而判斷該臉孔 疋Γ ί ί -子ot或為異常的，然後再選擇準備追縱的物件等 之1早TG 。主控制器3 0又包含一個位於拍攝地區中心 4ϋ〇45 2 五、發明說明（9) '~~~ --- 的物件官理單元3 3，以便用來判斷該物件是否係由判斷單 元32所選擇的追蹤目標，並可計算及輸出攝影機之調整值 以便能追蹤相對應物件的動作。主控制器3 〇又包括一傾斜 式轉盤或變焦距式鏡頭的控制裝置34，用以產生一傾斜式 轉盤或變焦距式鏡頭的控制訊號，因而能夠依照物件管^ 單元33的調整值而將影像輸入裝置1〇移向—左/右、上/ 下或前/後等方向。除此之外，主控制器3〇還具有一影像 儲存及傳送裝置3 5，用以儲存經由影像輪入裝置1 〇所操取 3臉孔影冑，並能將該影像分別傳送至一顯示裝置4〇及儲 擷取裝置5 〇。主控制器3 〇也同時含有一臉孔辨識及擷 取裝置36，用以辨識及擷取儲存的臉孔影像，然後將辨識 結果傳送給警報生成裝置6 0。 如第3圖所示，初步分析裝置31可藉由，將從輸入影 ^，號S313擷取回來之現有影像資料S31 4與先前儲存的影 ，實料S 3 1 5互作比較，因而能偵測一移動的部位^ 31 6。然 後初步刀析裝置31則可以偵測一個來自於影像输入f 1 K、綠色G、藍色B或YUV)皆是在膚色的範圍之内。接著， :步分析裝置3!即會選擇—其中膚色和動作皆會被摘測之 檢視區域的部位S318。 如第4圖所示’判斷單元32係藉相步分析裝置31而由 =機拍攝區域S中所選定的檢視區域D取得—個臉孔輪廊 if後’判斷單元32即會將先前健存臉孔樣板 的尺寸轉換成才欢視區域D中的尺寸’然後再將樣板與輪廓Λ 6 B 45 2 V. Description of the invention (7) 10 ---- Video input device 20 -----Driving device 3 0 ---- Main controller 31 --- A preliminary analysis device 3 2 ---- Judgment Device 33 ——- Object management unit 3 4 ---- Control device 35 ——— Image storage and transmission device 3 6 ---- Face recognition and capture device 4 0 --- —Display device 5 0-- -Storage or retrieval device 6 0 --- --Alarm generation device 100-Image input device 110--CCD camera 120-Image acquisition device 130--Camera drive control device 140-Camera drive 20 0-a pre-processing unit 3 0 0 --- characteristic sampling device 400--information sampling device 500-70 0 --- surface pattern generation device shape identification device 600--database ^ the best of the invention The embodiment will be described in detail with reference to the drawings as follows: 1 to 15 ′ is an embodiment of the present invention. What is not shown in Figs. 16 to 18 is another embodiment of the present invention. As shown in the first circle, the device of the present invention, 10, uses a camera mounted on 葙 w ~ image input to receive-a series of f :! to capture the image of the object on the input input device. The camera can: also include-a tilting turntable that can move the above-mentioned image or change the $, up / down or front / rear direction also-a drive device 2G for controlling the drive device. This device == the main controller of the action ^ will make the camera smooth. V. Invention Description (8) ~~ ------The image signal in the viewing area captures the face, and then displays this- Device 4 °. The device of the present invention also provides: a storage or retrieval device 50 for storing images from the main controller 30 or retrieving faces from image signals. The display device 40 is used as the No. 1 wheel after being used to pass the main controller 30 to the corresponding face. In addition, in addition to this, the device of the present invention may also include an alarm generating device 60 capable of generating an alarm signal by using a virtual display, and a knot that can be used to obtain faces from the controller 30. To operate the audio device. The image input device 10 can be moved to a left / right, up / down, or forward / backward direction by a tilting dial or a variable-motion lens 20 provided on the outside. In addition, the image input device 10 also includes a camera (that is, a digital camera in this invention), which is used to directly convert the image captured by the lens into an image signal defined as a digital or Yim format. The image input device 10 may also include a U processing frame to process image signals at a general frame interval t, and provide a series of digital image data (that is, 5 to 30 frames per second). ^ The second circle is a block diagram of the functional internal Θ structure of the main controller 30 shown in the second drawing. As shown in the figure, the main controller 30 includes a preliminary skin color or motion k7 inputted by the image input device 10 to select a viewing area. The main controller 30 also includes a Observe the face contour in the viewing area, and compare this contour with the pre-f face J-plate, and judge whether the face is abnormal or not according to the comparison result. Then select the item you want to chase, such as TG. The main controller 30 also includes an object management unit 3 3 located at the center of the shooting area 4ϋ45 2 5. Invention description (9) '~~~ --- In order to determine whether the object is determined by the determination unit 32 The selected tracking target can calculate and output the camera's adjustment value so as to track the movement of the corresponding object. The main controller 3 also includes a control device 34 for a tilting dial or a zoom lens, which is used to generate a control signal of a tilting dial or a zoom lens, so that it can be adjusted according to the adjustment value of the object management unit 33 The image input device 10 moves to the left / right, up / down, or front / back direction. In addition, the main controller 30 also has an image storage and transmission device 35, which is used to store the 3 face images accessed by the image rotation device 10, and can transmit the images to a display separately. Device 40 and storage retrieval device 50. The main controller 30 also includes a face recognition and capture device 36 for recognizing and capturing the stored face images, and then sends the recognition result to the alarm generating device 60. As shown in FIG. 3, the preliminary analysis device 31 can compare the existing image data S31 4 retrieved from the input image number S313 with the previously stored image S 3 1 5 to compare with each other. Detect a moving part ^ 31 6. Then, the preliminary analysis device 31 can detect that an image input f 1 K, green G, blue B, or YUV is within the range of the skin color. Next, the step analysis device 3! Will select-the part S318 of the viewing area where the skin tone and motion are picked up. As shown in FIG. 4, the 'judgment unit 32 is obtained from the selected viewing area D in the camera shooting area S by the phase analysis device 31-after a face contour, the judgment unit 32 will save the previous health The size of the face template is converted to the size in the area D only, and then the template and outline
第13頁 46645 2 五、發明說明（ίο) 互作比較。然而’如果在檢視區域D中偵測到有一個輪廓 近似於臉孔樣板’則判斷單元3 2即會認定該臉孔存在於檢 視區域D中S412。 ，外’如第5圖所示’藉由樣板與輪廓S5 1 1判斷有一 臉孔存在時，則判斷留-。π ^ η ^ 』斷早凡32即會判斷是否還有先前的追蹤 μ $ #次=在Iί判定之後，判斷單元3 2會依照現有選擇 二，1 二及「則儲存的影像資料間的比較結果，選擇追 :U公、Si>14。此時，為了防止會偵測到二個或更多 存的追縱物件影像擇一個最接近於先前儲 先前的追縱物件存在^作^的追縱物件S514。如果並無 # :½ i π a i、在則判斷單元32將選擇一個最接近 ； 二二、位置的影像來作為新的追蹤物件S51 3。 件管』單元33所=，位於一個螢幕上的位置（SX，SY)之物 # I _ π % 1將可拍攝到整個螢幕區域，並能判斷經由 U，y)的中心位置上。同時物件:否會位於攝影機拍攝區域 將會移動的位臟，二夸若：，理單元33會計算螢幕 鏡頭的控帝1，攝影機大”右霜利二傾斜式轉盤或變 並能將攝影機的調整值耠I ί =整個區域（Pmax，Tmax)， 的控制裝置34。 輪出給傾斜式轉盤或變焦距式鏡頭 如第7圖所示，傾斜式 置3 4係能產生一個傾斜式轉 便能依照得自於物件管理單 影像輸入裝置1 〇的攝影機可 轉盤或變焦距式鏡頭的控制裝 盤或變焦距式鏡頭的訊號，以 元3 3之攝影機調整值，而使得 以移向左/右、上/下或前/Page 13 46645 2 V. Description of Invention (ίο) Interaction comparison. However, if 'an outline similar to a face template is detected in the viewing area D', the judging unit 32 will determine that the face exists in the viewing area D S412. Outside, as shown in Fig. 5, when it is judged by the template and the contour S5 1 1 that a face exists, it is judged to stay-. π ^ η ^ "Break early Fan 32 will judge whether there is any previous tracking μ $ # 次 = After Iί judgment, judgment unit 3 2 will compare with the existing choices 2, 12 and" the stored image data As a result, choose chase: U male, Si> 14. At this time, in order to prevent two or more stored chase object images from being detected, choose a chase that is closest to the existence of the previous chase object. Vertical object S514. If there is no #: ½ i π ai, the judgment unit 32 will choose the closest one; 22, the position image will be used as the new tracking object S51 3. The piece tube "unit 33", located in a Position on the screen (SX, SY) # I _ π% 1 will capture the entire screen area, and can determine the center position via U, y). At the same time, the object: whether it will be located in the camera shooting area will move The position is dirty, the second quarantine :, the processing unit 33 will calculate the control lens 1 of the screen lens, the camera is large, and the right frost is tilted, or the camera ’s adjustment value can be changed. ) ， 的 控制 装置 34。 The control device 34. Rotate the tilting dial or zoom lens as shown in Figure 7. The tilting 3 or 4 series can produce a tilting rotation and can be turned or changed according to the camera obtained from the object management single image input device 10. The signal of the control lens of the focal length lens or the zoom lens is adjusted by the camera of RMB 33, so that it moves to the left / right, up / down or front /
4 6 6 45 2 五、發明說明（π) ----- 後的方向（在將來時將會是面向/背向著紙張）。如第7 中之Β所示，對於攝影機所偵測到的物件4，則傾斜式 或變焦距式鏡頭的控制裝置34便會將該物件d放置到第 圖中C所示之螢幕中心位置上。 如第8圖所示之訊號流程，影像儲存與傳送裝置”會 將由影像輸入裝置10所擷取之臉孔影像S6U轉換成一固定 的尺寸S612，然後會以點，圖1：)1111)的類型來儲存該臉孔的 影像S613。然後，影像儲％與傳送裝置35便會以一個預設 之縮小格式（即是JPEG)來長期儲存該臉孔的影像。除此之 外’影像儲存與傳送裝置3 5也會將追蹤物件之影像傳送至 臉孔辨識及擁取裝置36 ’以供用來辨識臉孔wig。 臉孔辨識及擷取裝置3 6可擷取一位人士之特性，以便 能自輸入的臉孔資料中找出該特定人士的臉孔。於是，臉 孔辨識及擷取裝置3 6便會將一個臉孔影像資料傳送至相對 應的特性上（第9圖中的臉孔樣板（a)與臉孔影像皆具 有相同的特性）。 如第1 0圖所示，係為本發明臉部攝影及辨識之方法 的實施例’其中包括初步分析步驟S1 〇丨〜S1 09，判斷步驟 SI 1 1〜si 1 9，以及一項影像儲存及傳送步驟5〗2】。初步分 析步驟S1 01〜S1 09能透過安裝在預設位置處的影像輸入裝 |置1 0來輸入拍攝到的影像，並能藉著一個含有膚色與動作 的取樣部位，來選擇一個檢視區域。初步分析步驟5丨〇〗〜 S1 0 9係能夠自選擇的檢視區域中所取樣得到的臉孔輪廓， 來選擇一個追蹤物件，比較輪廓與一個預設的臉孔樣板，4 6 6 45 2 V. Description of the invention (π) ----- Backward direction (in the future it will be facing / backing towards the paper). As shown in B in FIG. 7, for the object 4 detected by the camera, the control device 34 of the tilt or zoom lens will place the object d on the center position of the screen shown in C in FIG. . As shown in the signal flow shown in Figure 8, the image storage and transmission device "will convert the face image S6U captured by the image input device 10 into a fixed size S612, and then use dots, Figure 1 :) 1111) type To store the image of the face S613. Then, the image storage and transmission device 35 will store the image of the face in a preset reduced format (that is, JPEG) for a long time. In addition, 'Image storage and transmission The device 35 will also send the image of the tracking object to the face recognition and capture device 36 'for identifying the face wig. The face recognition and capture device 36 can capture the characteristics of a person so that Find the face of the specific person from the input face data. Therefore, the face recognition and capture device 36 will send a face image data to the corresponding feature (face model in Figure 9) (A) Both have the same characteristics as the face image.) As shown in FIG. 10, this is an embodiment of the method of facial photography and recognition according to the present invention, which includes the preliminary analysis steps S1 〇 丨 ~ S1 09, judge Steps SI 1 1 to si 1 9 and one item Image storage and transmission step 5〗 2]. Preliminary analysis steps S1 01 to S1 09 can input the captured image through the image input device installed at the preset position, and set it to 10, and can use a skin tone and motion To select a viewing area. The preliminary analysis steps 5 丨 〇 ～ S1 0 9 are able to select a tracking object from the face contours sampled in the selected viewing area, and compare the contour with a preset Face template
第15頁 46645 ^ 五、發明說明（12) 然後再依據其 的。影像儲存 1 〇所摘取之臉 臉孔影像及取 儘 可包括 控制步 追蹤物 用來追 轉盤或 變焦距 出之攝 上/下 管未見 一物件 驟。物 件是否 蹤相對 變焦距 式鏡頭 影機調 或前/ 比較結 及傳送 孔影像 樣自臉 於相關 管理步 件管理 位於攝 應追蹤 式鏡頭 的功能 整值， 後等方 果來判斷 步驟S121 進行辨識 孔影像的 圖式中， 驟及一傾 步驟可判 影機拍攝 物件動作 的控制步 ，以便能 將影像輸 向。 是否有臉 可以針對 與擷取， 追蹤物件 是本發 斜式轉盤 斷，在判 區域的中 之·攝影機 驟，可以 依據於物 入裝置10 孔存在 經由影 然後再 之輪廓 明另一 或變焦 斷步驟 心處> 的調整 控制傾 件管理 移動至 像輪入巢置 儲存輪入的 〇 實施例，係 距式鏡碩的 中所選擇的 並可計算出 值。傾斜式 斜式轉盤或 步驟所計苜 向左/右、 本發明之另一實施例，進 及掏取取樣區域中的臉孔輪廓 能產生一警報訊號的警報生成 除此之外，本發明之又一 在檢視影像時產生三次元的臉 於檢視區域時’可配合使用該 人的面模，並可藉由比較該建 存之個人的面模，諸如資料庫 特定的人士’故可順利地應用 此外’透過初步的分析步 鏡頭的控制步驟，本發明可儲 —步包括一項可在經過分析 ，並於發現臉孔為異常時， 步驟S120。 實施例，亦可包括，—項可 孔，像資訊，當該臉孔存在 二次元的資訊而建構成一個 構成一個人的面模與先前儲 ，因而可以進行辨識某一位 於臉孔影像的辨識工作上。 驟和傾斜式轉盤或變焦距式 存個在檢視區域中所準確Page 15 46645 ^ V. Description of the Invention (12) and then based on it. Image storage 1 〇 Extracted faces Face images and removal can include control steps Tracking objects Used to track the turntable or zoom out of the camera No object is seen in the up / down tube. Whether the object is tracked relative to the zoom lens camera or the front / comparison and transmission hole images from the face are managed by the relevant management steps, and the function of the camera tracking lens is set to the entire value. Then, wait for the result to determine step S121 for identification. In the pattern of the hole image, the step and the tilting step can determine the control steps of the object shooting action of the camera, so that the image can be output. Whether there is a face that can be pointed and captured. The tracking object is the oblique turntable of the hair. In the judgment area, the camera step can be based on the presence of 10 holes in the device, and then the contour is cleared or zoomed. The adjustment control of the center of the step> The tilting piece management is moved to the embodiment like the wheel-in nest storage wheel-in, which is selected from the distance mirror master and can be calculated. Inclined sloping turntable or step left / right, another embodiment of the present invention, the face contour in and out of the sampling area can generate an alarm signal. In addition to the alarm generation, Another three-dimensional face generated when viewing the image 'can be used in conjunction with the person's facial mask in the viewing area, and can be compared by comparing the facial masks of the stored individual, such as a specific person in the database', so it can be smoothly In addition, through the preliminary analysis step of the lens control step, the present invention can store-step includes a step S120 after analyzing and finding that the face is abnormal. The embodiment may also include: an item with holes, like information. When there is second-dimensional information on the face, it is constructed to form a person's face model and the previous storage, so that the identification of a certain face image can be performed. on. Step and tilt dials or zoom type
46 645 2 五、發明說明（13) 拍攝之瞬日寸臉孔影像（在一張靜止照片中的臉孔影像）於一 個另外δχ置的儲存裝置，並可將它應用於一個影像辨識裝 置，以作為調查之用，一個經過特別設計可利用辨識一個 臉孔等來調整儀器的保全設備等。 請參閱第1 1圖’係利用由臉孔資料中所擷取之眼睛 的排列’來檢驗一個傾斜度S71丨。臉孔資料於是被標準化 S712 ’所以會出現具有一固定尺寸及形狀的臉孔資料可供 擷取S7〗3。 首先，請參閱第1 3圖，係一項含有眼睛、鼻子、嘴 巴、眉毛等輪廓特性之輸入臉孔資料S8丨丨。取樣回來之輪 廓樣品整理好之後，各以圖素S81 2來替代^此外，為了能 決定免度的強度’因而利用輸入臉孔資料S813來計算出一 項平均克度°之後’在將此平均亮度直接與圖素S814的亮 度互相比較，以便產生一項水平線的輪廓取樣資料S8 i 5。 首先，將利用影像輸入裝置丨〇中的攝影機所輸入之一 個數位式的景彡像1定義成—種格式，以供從事數位化的處 理程序（例如RGB或YUV)，並可提供可適用於影像處理器的 解折度（如，具有640 X 480圖素）。此時，將現有的輸入影 像與一個先4的輸入影像相互比較，然後將它更新成為一 參考影像（於下次時可當作另一個先前的影像）。此外，透 過比較現有輸入資料及先前之輸入資料等之間的方式，來 偵測一個追縱物件的膚色和動作’並藉此而選擇一個檢視 區域。 之後，藉由影像輸入裴置1 〇所輸入的影像資料中取得46 645 2 V. Description of the invention (13) The instant-inch face image (the face image in a still photo) is stored in another δχ storage device, and it can be applied to an image recognition device. For investigation purposes, a security device designed to adjust the instrument by identifying a face, etc. Please refer to FIG. 11 'to check an inclination S71 丨 using the arrangement of eyes captured from the face data'. The face data is then standardized S712 ′, so a face data with a fixed size and shape will appear for S7. 3 First, please refer to Figure 13 for input face data S8, which contains contour features such as eyes, nose, mouth, and eyebrows. After finishing the sampling of the contour samples, each pixel is replaced by the pixel S81 2 ^ In addition, in order to determine the intensity of the degree of exemption, 'the input face data S813 is used to calculate an average gram degree °'. The brightness is directly compared with the brightness of the pixel S814 in order to generate a horizontal line profile sample data S8 i 5. First, a digital landscape image 1 input by a camera in the image input device is defined as a format for digital processing programs (such as RGB or YUV), and can be applied to Resolution of the image processor (eg, with 640 X 480 pixels). At this time, the existing input image is compared with a first 4 input image, and then it is updated into a reference image (it can be used as another previous image at the next time). In addition, by comparing the existing input data with the previous input data, etc., it is used to detect the skin color and movement of a tracking object 'and select a viewing area. After that, it is obtained from the image data input by the image input Pei Chi 10.
第17頁 ^466^ ^466^Page 17 ^ 466 ^ ^ 466 ^
五、發明說明（14) 影像樣品，並偵測一個含有RGB比例及動作的部位 再以主控制器3 0的初步的分析裝置3丨來選擇—項擷取的 圍。在完成偵測部位與動作之後，利用初步的分析裝置31 來儲存相關的資料，然後使用此資料以便用來偵測^ 一個 影像。其次，藉由使用動作之資料並將檢視區域的資料 送給判斷單元32等方式，再利用初步的分析裝置31來進 選擇檢視區域。 藉由榻取經過初步的分析裝置31分析過的檢視區域， 並以判斷單元32進行判斷到底該臉孔是否係存在。第丄2 圖中所示，為了達到上述之目的’首先利用判斷單元32來 谓測一個，其中心位置處係為呈黑暗的而位於檢視區域之 影像訊號的下緣部位則呈明亮的。偵測輪廓之結果顯示 則將可順利地擷取到其中具有明顯的眉毛 '眼睛、鼻子的 下半部及嘴巴等的一個輪廓S815。接著，判斷單元32便需 判斷疋否有臉孔存在於檢視區域中，以防止在經過和先前 储存之臉孔樣板相互比較之後’會顯示出檢視區域中的輪 廓的數值會高於參考值之現象。由於檢視區域並無固定的 尺寸，因而在比較之前’對於先前儲存的樣板尺寸則應該 要變更。在那之後’藉由比較一個現有的臉孔取樣資料和 一個先前的資料，以使得物件管理單元33可以選擇追蹤物 件。假使再一個新的螢幕中偵測到一個臉孔時，當偵測到 二個或更多個臉孔時，則該臉孔即會變成一個追蹤物件， 並且移中一個最接近於先前之物件的，即會被選擇作為— 個追蹤物件。如果其中並無可供追蹤的先前物件時，則最V. Description of the invention (14) Image sample, and detect a part containing RGB ratio and motion, and then use the preliminary analysis device 3 丨 of the main controller 30 to select the range of item capture. After the detection of parts and actions is completed, the preliminary analysis device 31 is used to store relevant data, and then this data is used to detect ^ an image. Secondly, by using the action data and sending the data of the viewing area to the judgment unit 32, etc., the preliminary analyzing device 31 is used to select the viewing area. The inspection area analyzed by the preliminary analysis device 31 is taken, and the judgment unit 32 determines whether the face exists. As shown in FIG. 2, in order to achieve the above-mentioned purpose, firstly, the judgment unit 32 is used to measure one, and its center position is dark and the lower edge of the image signal located in the viewing area is bright. The result of detecting the contour shows that a contour S815 with obvious eyebrows, eyes, the lower half of the nose, and the mouth can be captured smoothly. Then, the judging unit 32 needs to judge whether or not a face exists in the viewing area, so as to prevent the value of the outline in the viewing area from being higher than the reference value after being compared with the previously stored face template. phenomenon. Since the viewing area does not have a fixed size, the size of the previously stored template should be changed before the comparison. After that, 'the object management unit 33 can choose to track the object by comparing an existing face sampling data with a previous data. If a face is detected on a new screen, when two or more faces are detected, the face will become a tracking object, and the one closest to the previous object will be moved , It will be selected as a tracking object. If there are no previous objects to track,
第18頁 46645 2 五、發明說明（15) 接近於螢幕中心位置處的一個臉孔，將被選出 蹤物件。 田f碲迫 其次’傾斜式轉盤或變焦距式鏡頭的控制裝置34係 供用來操作攝影機，以便追蹤物件，並防止該追蹤物件^ 偏離攝影機的中心位置，傾斜式轉盤或變焦距式鏡 制裝置34包括傾斜式轉盤或變焦距式鏡頭的驅動裝二 =是被安裝在影像輸入裝置10的攝影機之上 攝影機即可以接受主控制器上的傾斜式轉盤或變 之指令而進行操作。傾斜式轉盤或變二距 驅動裝置2G則是能夠以串列#、序料 '萬用、 列式匯W排或無線通訊等方式， 串 焦距式鏡頭的控制裝置34。 ^°傾斜式轉盤或變 置35便將可一個：二：：：之後’影像儲存及擷取裝 個別的儲存裝置中，並可鱼以圖像的類型’儲存至 傳送至一個辨識系統。將景/像與取樣的特性鄧資料-起 臉孔影像指頁敢^ ^罢Q。 便能找出儲存；^ ^ ^ 可偵洌—位特定人士的特性，以 仗犯？X出燔存於輪入臉孔 τ 人 取裝置36可以將—個孤β 、甲的—個臉孔。臉孔影像擷 中。 ㈣孔影像輪出至相對應的特性資料檔 此時，處理的資料 _ 其中唯有臉孔部位才可· '八貝料）係為—矩形影像資料， 寸而有所不同，然而資料傾向】=或許可依照臉孔尺 因此，執行檢視一個臉 f不需一致。 ϋ之程序’然後擷取眼睛及臉Page 18 46645 2 V. Description of the invention (15) A face near the center of the screen will be selected as a tracking object. Tian f Telluride's second control device 34 for tilting the turntable or zoom lens is used to operate the camera to track the object and prevent the tracking object ^ from deviating from the center position of the camera, tilting turntable or zoom lens device 34. Drive assembly including a tilting dial or a zoom lens == is mounted on the camera of the image input device 10, and the camera can accept the tilting dial on the main controller or change the command to operate. The tilting turntable or variable-pitch drive 2G is a control device 34 for tandem-focus lenses in the form of tandem #, sequencer, universal, in-line sink, or wireless communication. ^ ° tilting dial or setting 35 will be able to one: two ::: after that ‘image storage and retrieval is installed in a separate storage device, and the fish can be stored as an image type’ and transferred to an identification system. The characteristics of scene / image and sampling Deng materials-from the face image refers to the page dare ^ ^ strike Q. You can find the storage; ^ ^ ^ Can detect the characteristics of a specific person to war criminals? Xouts are stored in the face of the wheel. Τ The person taking device 36 can put a lone β and a face. Face image capture. The perforated image is rotated out to the corresponding characteristic data file. At this time, the processed data _ only the face parts can be used. 'Eight shell material） is-rectangular image data, the size varies, but the data tends to be] = Or it may be allowed to follow the face rule. Therefore, it is not necessary to perform a view on a face f. ϋ 的 程序 ’Then capture eyes and face
第19頁 IUH: 4 6 645 ? 五、發明說明（16) 孔資料中之臉孔元件的特性等資料，如s丨〇 i、S1 〇 3所示。 然後’配合使用眼睛之陣列以進行處理臉孔資料標準化程 序’以便能使得臉孔的尺寸及形狀都能一致化，如第1 〇 圖所示，S1 0 5。 之後’針對尺寸和形狀S107、S1 〇9，一致的臉孔中的 各個兀件之輪廓進行取樣，然後配合使用一個眼睛樣板 S11 1 ’進行擷取眼睛部位，因而得以藉著檢驗其所擷取之 細節資料S Π 3來判斷是否有誤失存在。 循序地’配合使用一個嘴部四周的樣板sn 5，進行擷 取嘴部’因而得以藉著檢驗其所擷取之嘴部資料s 11 7而來 判斷是否有誤失存在。 此外，配合使用一個鼻子四周的樣板及眉毛和眉毛四 周的樣板（圖式中並未明示），進行擷取鼻子部位和眉毛部 位的資料，因而得以藉著檢驗其所擷取之鼻子和眉毛的資 料而來判斷是否有誤失存在。 ' 然後，利用眼睛和嘴部（大概是鼻子和眉毛）的檢驗社 果，同步判斷是否有誤差存在S119。 .假使在眼睛， 存在時，則進行該 此擷取之資料傳送 系統，以供辨識之 時，便能配合使用 人士的資料。 鼻子，嘴部眉毛或綜 人士臉孔圖像資料之 至一位門房的保全系 用（S121)。因此，如 任一種操取程式，輕 合的部位皆無誤差 寫入與儲存，並將 統或是一訪客識別 果有任何問題發生 易地擷取一位可疑 換句話說，如果在有關眼睛、鼻子、嘴邹、眉毛或綜Page 19 IUH: 4 6 645? V. Description of the invention (16) The characteristics of the face elements in the hole data, etc., are shown in s 丨 〇i, S103. Then, “the eye array is used in conjunction with the face data normalization program” to make the size and shape of the face uniform, as shown in FIG. 10, S105. Afterwards, the size and shape of S107, S1 〇9, the contours of each element in the uniform face were sampled, and then used with an eye template S11 1 to capture the eye parts, so it can be checked by inspection Details S Π 3 to determine whether there is a mistake. Sequentially, 'with the use of a template sn 5 around the mouth to retrieve the mouth', it is possible to determine whether there is a mistake by checking the captured mouth data s 11 7. In addition, by using a template around the nose and a template around the eyebrows and eyebrows (not shown in the figure), the data of the nose and eyebrows can be retrieved, so the nose and eyebrows can be checked by examining Data to determine if there is a mistake. 'Then, using the inspection results of the eyes and mouth (presumably the nose and eyebrows), it is simultaneously determined whether there is an error in S119. If it is in the eyes, the captured data transmission system is carried out for identification, and it can cooperate with the user's data. The nose, mouth, eyebrows, or face image data of a comprehensive person is used for security of a porter (S121). Therefore, if any kind of manipulation program is used, the lightly-closed parts are written and stored without error, and the system or a visitor can easily identify if there is any problem. In other words, if it is related to the eyes, nose, etc. , Mouth, eyebrow, or comprehensive
第20頁 46 6,Page 20 46 6,
五、發明說明（π) 合性的檢驗結果有誤差存在時’即可判斷該位人士具有危 險要素，例如戴一個面罩’深色的太陽眼鏡，一個下拉g 蓋的帽子。在這情況之下，處理程序即會產生警報訊號， 讓主管人員何立即採取緊急措施應付（S1 20)。 综上所述’本發明係可以配合使用一部安裝在一處 口處的攝影機’快速又準確地拍攝一個臉孔，以便能辨A ' 出陌生人。除此之外’事實上’儲存裝置的容量係=夠^ 著地降低，由於本發明所儲存的僅是靜態的圖像，而非 同既有的專利技術中所採用的動態圖像。 a 而且，為了防止有異常人士佩帶著一個面罩，.、果色' 太陽眼鏡或一個下拉覆蓋的帽子進入，即是，—加曰_^的 險要素的人士出現，本發明即會預先產生警報訊號給主^ 或維修人員，以便採取緊急措施。 ~’_α Τ 此外’因為可用來追蹤臉孔之動作的攝影機係安右 一扇門上，因而本發明便能對罪犯產生嚇阻致用/ χ在 何其他的監督系統更有效率。 比任 本發明可克服既有專利技術需要以臉孔前去對^ _ 固定式的攝影機等不方便性。當應用在影像電古舌時 口 攝影機係能機智地移動至上/下、左/太夕士二由於 ^ 乃向，以便能 讓臉孔得以準確地出現在其正前方。 更方便的功能，由於該攝 而4疋既有專利技術無 此外，本發明係可以提供一 衫機係能對應至1¾或矮的人皆可 法辦到的。 本發明同時也具有更大的效益，因它可以結合一套門V. Description of the invention (π) When there is an error in the test result of the suitability, it can be judged that the person has dangerous elements, for example, wearing a mask, dark sunglasses, and a hat with a g cover pulled down. In this case, the processing procedure will generate an alarm signal, so that the supervisor can take emergency measures immediately (S1 20). To sum up, ‘the present invention can be used with a camera installed at a mouth’ to shoot a face quickly and accurately, so that A ′ can be a stranger. In addition, the capacity of the storage device is actually reduced enough, because the present invention stores only static images, not dynamic images used in the existing patented technology. a Moreover, in order to prevent abnormal persons from wearing a mask, fruit-colored sunglasses, or a drop-covered hat to enter, that is, a person with a risk factor of _ ^ appears, the present invention will generate an alarm in advance Signal to the owner ^ or service personnel for emergency action. ~ ’_Α Τ In addition, because the camera that can be used to track the movement of the face is mounted on a right door, the present invention can be used to deter criminals / χ, what other monitoring systems are more efficient. The invention can overcome the inconveniences of the existing patented technology, such as the need to face the fixed camera. When applied to the image of the ancient tongue, the mouth camera system can intelligently move to the up / down and left / too shishiji ^ direction, so that the face can appear accurately in front of it. More convenient functions. Because of this camera, there is no patented technology. In addition, the present invention can provide a shirt machine that can correspond to 1¾ or short people. The invention also has greater benefits because it can be combined with a set of doors
466452 五、發明說明（is) 禁系統共同來擔任辨識陌生人的主& ^ m A名 王人的貝任’或是能使用於一間 製造工廠或使用自動影像追蹤枯 . *延峨技術的連線即時模擬電玩。 以下’係為本發明臉部辨嗜太。土 p ^ 1汧°我方法之另一實施例，並以 第1 6至1 8圖闡述如下。 第1 6圖係為本發明另—實施例中，用來拍攝及辨識 臉孔之方法的方框圖’卩闡明一種臉部拍攝及辨識系統的 架構。 如圖所示’一種影像輸入裝置100係藉著多部朝向預 設方向的攝影機’以供用來拍攝一個物件。大體上來*兒， 影像輸入裝置100至少配置二部CCD攝影機11〇 β 一個預處理單兀200可擷取一個經由影像輸入裝置1〇〇 而輸入之一個影像訊號。預處理單元2〇〇可以消除影像訊 號裡的雜訊’然後擁取一個有關該影像訊號的一個輪廊的 樣品。 一個特性取樣裝置3 0 0可以，依據自預處理單元2⑽輸 入之影像訊號的輪摩’來擷取該影像訊號的樣品。 ’ 一個三次元的資訊取樣裝置4 0 0能藉由使用，由特性 取樣裝置300所取樣之特有的樣品，即可以使得預處理單 元200端的影像訊號產生旋轉’上升或移開等方式來重整 影像，並可藉著利用介於重新整理之影像間的移動距離^， 來操取該三次元的資訊。 一個面模產生裝置500，可以將自三次元的資訊取樣 裝置4 0 0所擷取之二次元資訊樣品與一預設的參考面模相 互比較，並可針對母人士可以產生—新的面模專屬權。466452 V. Description of the invention (is) Forbidden system to jointly identify strangers & ^ m A famous king's beren 'or can be used in a manufacturing plant or using automatic image tracking. * Yan'e technology Connection to simulate video games in real time. The following 'is the facial discrimination of the present invention. A further embodiment of our method is illustrated in Figures 16 to 18 below. Fig. 16 is a block diagram of a method for photographing and recognizing faces in another embodiment of the present invention ', which illustrates the architecture of a facial photographing and recognizing system. As shown in the figure, 'an image input device 100 is provided with a plurality of cameras oriented in a predetermined direction' for shooting an object. Generally speaking, the image input device 100 is configured with at least two CCD cameras 11 β. A pre-processing unit 200 can capture an image signal input through the image input device 100. The pre-processing unit 200 can eliminate the noise in the video signal 'and then grab a sample of a corridor about the video signal. A characteristic sampling device 3 0 0 can acquire a sample of the image signal according to the rotation of the image signal input from the pre-processing unit 2⑽. 'A three-dimensional information sampling device 4 0 0 can be used by using the unique sample sampled by the characteristic sampling device 300, that is, the image signal at the pre-processing unit 200 can be rotated to rise or move to reform Images, and the three-dimensional information can be manipulated by using the moving distance ^ between the rearranged images. A surface mold generating device 500 can compare the two-dimensional information sample obtained from the three-dimensional information sampling device 400 with a preset reference surface mold, and can generate a new surface mold for the mother. Exclusive rights.
第22頁 46645； 五、發明說明（19) 一個資料庫600包含有獲自合格使用者的影像訊號， 以作為面模資料的多個影像資料。 一個形狀辨識裝置700通常係為一中央處理單元cpu。 形狀7識裝置70 0會將面模產生裝置5〇〇中新近產生的面模 與在資料庫6 0 0中的面模資料互作比較，然後判斷該新的 面模使否係和面模資料相同。如果有一個面模是和資料庫 6 0 0中的新面模相同時，則經由攝影機所拍下的現有的影 像將被視為是前以註冊登記的面模5而此使用者則會被 歸類於是一位合格的使用者。否則，則該影像將會被辨認 成為一位註冊登記的面模，而且該使用者亦會被歸類 個不合格的使用者。 。一個儲存的工具δ〇〇係包括可供全套系統使用之一作 業程式，以及辨識臉孔時所必須使用之多項資料。上述之 儲存工具800的操作作業係依照一項來自於形狀辨識裝置 700的控制訊號，因而可以將先前就有的儲存資料輸 形狀辨識裝置700。儲存工具8〇〇係可採用具備可能— 半導j記憶卡片、一顆硬式磁碟機、一個光碟機等。 影像輸入裝置1 〇 0係可以包含，可供用來拍攝— =的CCD攝影機11〇，一種可自攝影機中擷取訊號並能按昭 ^方式來顯示此訊號之影像擷取裝置丨2 〇 , 一種可藉由=十 鼻一個CCD攝影機之現有的旋轉範圍並配合使用一個^ 性取樣裝置300所選定之檢視區域，而輸出一個攝影啦 ，^制訊號之攝影機驅動控制裝置丨30，以及—個可以^ 照攝影機驅動控制裝置1 30的攝影機驅動控制訊號來轉^Page 22 46645; V. Description of the invention (19) A database 600 contains a plurality of image data obtained from qualified users as image data. A shape recognition device 700 is usually a CPU. The shape recognition device 70 0 compares the newly generated surface mode in the surface mode generating device 500 with the surface mode data in the database 600, and then determines whether the new surface mode is related to the surface mode. The information is the same. If a mask is the same as the new mask in the database 600, the existing image captured by the camera will be regarded as the previously registered mask 5 and the user will be Classified as a qualified user. Otherwise, the image will be recognized as a registered mask, and the user will be classified as an unqualified user. . A stored tool δ〇〇 includes a job program that can be used by the entire system, and a variety of data that must be used when identifying faces. The operation of the storage tool 800 described above is based on a control signal from the shape recognition device 700, so the previously stored data can be input to the shape recognition device 700. The storage tool 800 series can use the possible-semi-conductive j memory card, a hard disk drive, an optical drive, etc. The image input device 100 may include a CCD camera 11 for shooting — = an image capture device capable of capturing a signal from the camera and displaying the signal in a ^^ way, 2 By using the existing rotation range of a CCD camera in conjunction with the viewing area selected by a sampling device 300, a camera can be output, and a camera driving control device for the signal can be output. 30, and one can ^ According to the camera drive control signal of the camera drive control device 1 30
第23頁 «46645 五、發明說明（20) 攝影機至一指 如第1 7 攝及辨識一張 包括， 域之步 性，重 有臉孔 性》然 項方法 新整理 可按照 訊，並 臉孔面 庫的個 位特定 第 一張臉 它可利 資料和 比對樣 進行調 區域。 在檢視 重新整 可以利 驟 S200 新整理 的部位 後調整 更進一步包括一 在檢視 位於重 且能配 模。上 人註冊 人士之 1 8圖 孔之方 用攝影 動作資 板與臉 整攝影 本項方 區域中 理過之 之攝影機驅動裝 ，係為本發明之 法的流程圖。如 並選擇所欲擷取 可自相對應的檢 用所擷取的臉孔 有臉孔之部位進 方向以便判斷是 項步驟S26 0，它 具有一個臉孔之 影像的位移資訊 項三次元得資訊 則更進一步地包 尋構成的臉孔面 已經註冊登記之 明另一實施例中，另 法包括 的資料 對應區 過是否 選擇至 〜S370 影像訊 訊來產 置 140。 另一實施例中，拍 該方法係 的檢視區 測其特 括，一項若自資料 模時，即可判斷一 步驟S270。 定之方向 圖中所示 臉孔之方 用攝影機 〜S250 ， 影像，利 ，針對具 攝影機之 區域中係 新整理之 合使用該 述之方法 面模中搜 影像是否 係為本發 法的流程 機輸入的 料，配合 孔元件， 機的方向 法亦包括 存在著一 影像之間 圖。該方 影像訊號 使用在相 並再判斷 等，則可 步、S350 個臉孔的 的位移資 圖所示， 影像訊號 視區域偵 元件來偵 行取樣並 否有臉孔 可藉著其 影像訊號 來產生三 而構成一 拍 步驟S300 中，進行 域中的亮 存有一個 少一個以 ’它可重 號中的影 生三次元 測一個和 檢驗其特 存在。此 特性，重 的影像， 次元的資 個物件之 攝及辨識 〜S 3 4 0， 擷取膚色 度差，來 臉孔後， 上的檢視 新調整， 像，依照 的資訊，Page 23 «46645 V. Description of the invention (20) Camera to one finger as described in Section 1 7 Photographing and recognizing a picture, including the step of the field, regaining the faceness> However, the new method can be updated according to the news, and the face The niche of the face library specifies the first face. It can adjust the area using data and comparison samples. After reviewing the rearranged parts of the S200, adjustments can be made further. The adjustments further include a review that is located and can be adjusted. Figure 8 of the registered person of the hole Kong Zhifang uses photographic action board and face to complete photography. The camera driver installed in this area is the flowchart of the method of the present invention. For example, if you want to select the direction you want to capture from the part of the face that has been captured by the corresponding inspection, in order to determine that it is step S260, it has the displacement information of the image of a face. Then, it is further included that the formed face has been registered. In another embodiment, whether the corresponding area of the data included in the alternative method is selected to ~ S370 image news to produce 140. In another embodiment, the inspection area of the method system is taken to measure its specifics. If one item is from the data model, a step S270 can be judged. The camera used in the direction of the face shown in the fixed direction is ~ S250. The image is good. For the newly-organized area in the camera area, use the method described above to search whether the image is input by the process machine of this method. Materials, mating hole components, and machine orientation methods also include the existence of an image between images. The image signal of this party is used in phase and re-judgment, etc., as shown in the displacement map of S350 faces. The image signal view area detection element is used to detect the sample and whether there are faces can be detected by its image signal. In step S300, three generations are formed to make a single shot. In the field, there is one less one in the field, and it can be used to measure one of the three-dimensional shadows in the duplicate number and test its specific existence. This feature, the heavy image, the capture and recognition of dimensional objects and objects ~ S 3 4 0, capture the difference in skin tone, after the face, the new adjustment on the view, image, according to the information,
第24頁 '4664：Page 4'4664:
以及藉著比較該三次元資訊 個人的臉孔面模。該方法更 若藉由進行搜尋儲存暫資料 即可以判斷某一特定人士的 以上所構成之本發明的 方式如下： 與一個參考面模，而構成一個 進一步包括步驟S380〜S400， 庫中之個人的註冊登記面模， 影像是否已經註冊登記。 每一項具體化實施例，其運作 1 5 :ί ^ ί 1 7圖中所示實施例之運作，係可參考黃 d f中的相關方框圖而加以闡述。 擷取ί 00可接收來自CCD攝影機110以及影像 缺後ΐ 象訊號。預處理單元2°〇可以消除雜訊， '後執仃輪廓校正與過濾。 3 f) η :像Λ號處理的方式’係將它輸人於特性偵測裝置 性偵測裝置300可藉由比較訊號的影像中-個关 與一個現有之影像之間的結果，而進行谓測-姻 ^ “山動作，並因而產生動作的資料。特性偵測裝置3〇q :錯偵測其中一個，含有成比例之彩色元件，如R (紅 G(綠色）、B(藍色）或γυν等皆屬於膚色範圍的部位， 來產生一個膚色資料。 ^ 特性價測裝置3〇〇能夠選擇至少一個，利用影像輸入 衣置1 ο 〇來輸入於其中一個欄位之影像訊號的檢視區域， =預處理單元2〇〇則可使用經由上述之方式所產生之動作 貝料及色彩的資料’ S21 〇。特性偵測裝置3〇〇可以藉由識 別’.儲存於已選定之檢視區域S2 1 0 — 1中的，臉孔元件之 方式來選擇一個其中存有一個臉孔之區域，然後再配合使And by comparing the three-dimensional information personal face mask. In the method, if the temporary data is searched and stored, the method of the present invention constituted by a specific person can be judged as follows: With a reference surface model, a method is further included that includes steps S380 to S400. Register the mask, whether the image has been registered. The operation of each specific embodiment 15: ί ^ ί 17 The operation of the embodiment shown in the figure can be explained with reference to the relevant block diagram in Huang Df. Capture ί 00 can receive CCD camera 110 and image artifact signals. The pre-processing unit 2 ° 〇 can eliminate noise, and then perform contour correction and filtering. 3 f) η: The method of processing like Λ 'is input to the characteristic detection device. The detection device 300 can perform the comparison by comparing the result between an image in the signal and an existing image. Predicate measurement-marriage ^ "Mountain action, and thus generate data of action. Characteristic detection device 30q: One of the wrong detections, which contains proportional color elements, such as R (red G (green), B (blue ) Or γυν and other parts that belong to the skin color range to generate a skin color data. ^ The characteristic price measuring device 300 can select at least one and use the image input clothes 1 ο 〇 to enter the image signal view in one of the fields Area, = pre-processing unit 200 can use the motion material and color data generated by the above method 'S21 〇. The characteristic detection device 3OO can be identified by recognition'. Stored in the selected viewing area S2 1 0—1, the face component method to select an area where a face is stored, and then cooperate with
第25頁 ί'46645 2 五、發明說明（22) 用特性育料S220，即可重新整理，經由CCD攝影機110及預 處理單元200等所輸入之影像。 其所產生之檢視區域中的動作和色彩資料、被债測過 的特性及重新整理的檢視區域的影像資料等，皆可分別輸 入於二次元資訊取樣裝置4 〇 〇和攝影機驅動控制裝置〗3 〇。 攝影機驅動控制裝置丨3〇會比較存在於其所產生之檢 視區域中的動作與膚色資料，被偵測過的特性，並會配合 先前輸入及處理過的檢視區域資料，而重新整理影像資 料。 如上所述’攝影機驅動控制裝置丨3 〇同時也可藉由相 互比較一個檢視區域的資料，而計算出檢視區域的移動距 離S230。此外，藉由將CCD攝影機nQ加以旋轉至相當於1 移動的距離時，則其旋轉距離和角度的設定係為了配合拍 攝之物件而定。 已設定好其旋轉距離和角度的CCD攝影機，於是便可 以開始輸入資料，就如同一部攝影機驅動控制訊號被輸入 到攝影機驅動裝置140 —般。攝影機驅動裝置14〇會接收攝 影機驅動控制訊號，然後會儘可能地旋轉以 某-方向的預設角度S24。。只要一偵測到有新攝^ 時，則上述之攝影機控制的操作事項即會重複地執行。 換句話說，三次元資訊取樣裝置4〇〇會收到該檢視區 域中的動作與膚色資料，已谓測的特性，以及由特性偵測 裝置30 0所重新整理的影像資料，並且會配合使用已请測、 之特性而再重新整理一次，然後配合使用重新整理的影像Page 25 ί'46645 2 V. Description of the invention (22) With the characteristic breeding material S220, the images can be rearranged and input through the CCD camera 110 and the pre-processing unit 200. The motion and color data in the viewing area generated by it, the characteristics measured by the debt, and the image data of the rearranged viewing area can be input into the two-dimensional information sampling device 400 and the camera drive control device respectively. 3 〇. The camera driving control device 丨 30 will compare the motion and skin color data existing in the inspection area generated by it, the detected characteristics, and will reorganize the image data in accordance with the previously entered and processed inspection area data. As described above, the camera driving control device 3 can also calculate the moving distance S230 of the viewing area by comparing the data of the viewing area with each other. In addition, when the CCD camera nQ is rotated to a distance equivalent to 1 movement, the setting of the rotation distance and angle is determined to match the object being photographed. The CCD camera whose rotation distance and angle have been set can then start to input data, just like a camera driving control signal is input to the camera driving device 140. The camera driving device 14 receives the camera driving control signal, and then rotates as much as possible at a preset angle S24 in a certain direction. . As soon as a new camera is detected, the above-mentioned operations of the camera control will be repeatedly performed. In other words, the three-dimensional information sampling device 400 will receive the motion and skin color data in the viewing area, the measured characteristics, and the image data rearranged by the characteristic detection device 300, and they will be used together. I have retested the characteristics, and then refreshed the image again, and then used the refreshed image together
第26頁 4 6 6 4 5 2 五、發明說明（23) 特性’而判斷是否有一個臉孔係已經存在於檢視區域之内 S2 5 0 ° 由於必須進行判斷，在檢視區域中具有一個臉孔的影 像说號之二次元資訊，在配合使用位於多個重新整理過的 :像之間的位移資訊，#可順利產生。除此之外，若比較 f所產之二次元資訊與一個位在面模產生裝置5 00中之 >考面模，便能產生—個新的個人的面模s 2 6 〇。 π > f後將新產生的面模拿來與一個在資料庫6 0 0中之 次：：：作比肖’以便能判斷是否有相同的面模係存在於 果係已經有一相同的面模資料存在於資料庫 模，&夺，+則新的面模將被視為是一個已註冊登記過的面 使用去現搞ί攝影機所擷取之物#，亦將被歸類為合格的 的面假使在資料庫6〇°中並沒有存在相同 模、、：：，則新的面模將被視為是一尚未註冊登記的面 、=〇忒物件則會被歸類為一個不合格的使用者S27〇 s 以便M ^ 8圖中所示&，係為結合了第1 6圖中之圖例 更閣述另—實施例的方框圖。 當一個影像訊號透過CCD攝影機丨〗〇和影 中的料擷取裝置120及預處理單元2 0 0而輸入/，則 的^測裝置300即會判斷輸入的訊號，並且會比較現有 物：：與先前的螢幕的影像訊號。依照比較的結果，一個 其所伯動作係可以被偵測亦便產生一個動作資料，而且若 YUV) ^測到之一個部分之中，若其彩色元件（如R、G、β或 的比例是屬於膚色的範圍之内時’便會產生一個色彩Page 26 4 6 6 4 5 2 V. Description of the invention (23) Characteristic and determine whether a face already exists in the viewing area S2 5 0 ° Due to the necessity of judgment, there is a face in the viewing area The two-dimensional information of the video commentary of # can be generated smoothly in combination with the information of the displacement between multiple rearranged: images. In addition, if the two-dimensional information produced by f is compared with a face mask generated in the face mask generating device 5000, a new personal face mask s 2 6 can be generated. After π > f, the newly generated face model is compared with one in the database 6 0 0 :: as a comparison, so as to determine whether the same face model system exists in the fruit system and already has the same face. The model data exists in the database model, & +, the new face model will be regarded as a registered face and used to retrieve the thing captured by the camera #, and will be classified as qualified If the same surface does not exist in the database 60 °, the new surface mold will be regarded as an unregistered surface, and the object of 0% will be classified as an The qualified user S270s so that the & shown in Fig. 8 is a block diagram of another embodiment in combination with the legend in Fig. 16. When an image signal is input to / through the CCD camera, the material acquisition device 120 and the pre-processing unit 200 in the image, the measurement device 300 will judge the input signal and compare the existing objects :: Image signal from previous screen. According to the result of comparison, an action data can be detected to generate an action data, and if YUV) ^ measured in a part, if the proportion of its color components (such as R, G, β or When it is within the range of skin tone ', a color will be produced
46645 2 五、發明說明（24) ' 資料S310。 此外’在上述之程序中，特性偵消 用色彩資料’而再一個攔位上的影像訊 個以上之檢視區域，因而即會產生動作 當檢視區域選定了之後，則此項具 測到一個部位，其中’若透過檢驗臉孔 且再挑選出再相對應區域中的追縱物件 存在著。於是藉著讓特性偵測裝置3〇〇 ^ 能重新整理影像S330 ’然後藉由重新整 即可辨識出再相對應的區域中，存在著 如果再相對應之區域中的追蹤物件，係 則在小心地調整攝影機並使得該影像訊 孔，能位於螢幕的中心點上S 3 5 0之後， 新整理。接著，在配合使用位在多個重 移資訊S3 60，則三次元的資訊即會因此 資訊會與在面模產生裝置500中的參考3 便產生一個因人而異之新面模S370。 如果益沒有正確的臉孔時，則攝影 之攝影機的移動量而計算出該攝影機的 回復至正常的位置（一般攝影^tjS341 接收到的影像訊號，此項具體化實施例 執行一項能產生皮膚資料及動作資料的 如上所述’若和資料庫6 0 0中的一 d 時’則新的面模於是便會順利產生，因 丨裝置300會配合使 ^號中至少會選擇一 的資料S320。 體化實施例即會偵 元件S320 _ 1，並 之後，一個臉孔係 I偵測其特性，便 理之影像，本發明 —追蹤物件S340。 具有一個臉孔時， 號所擁有的一個臉 則該影像即會被重 新整理之影像的位 而產生。三次元的 3模互相比較，以 機會藉著其所指示 5周整值，被調整而 。然後，利用其所 中所述的方法可以 步驟S310。 固面模資料作比較 而可以判斷出該新 46645 2 五、發明說明（25) 類於不合格 的面模係已經存在於資料庫600中㈡80。由於判斷結果， 如果係有相同的面模存在資料庫60〇中時，則新的面模即 會被視為是-已註冊登記的面模S39〇 ’而且讀攝影機擷取 之現有之物件也會被歸類於合格的使用者。如果再資料庫 600之中並無相同的面模時，則此新的面模則會被視貝為是 一個尚未註冊登記的面模，而該物件也會被歸 疋 的使用者。 因此’綜上所述’本發明係可藉著自—個影像訊號， 並不斷重複地調整攝影機的位置’同時將攝影^ 至"宜 -角度直到：以順利地讓臉孔完全辨識，因而二 擁取到的皮膚資料和作資料而完成辨識 自於它可以機智地，準確對準一個胳 皿王蛋疋术 去銘#、由人 賤孔’而不需勞駕使用 者移動身妝至一適當位置來配合辨增 ^ ^ ^ ^ λ. ^ m j± 糸充。本發明之臉部 攝衫與辨4之方法和裝置係已經詳細地 惟以上所揭示者，僅為本發明 例而已，不能以此限定本發明之實5J項較佳可行之實施 之申請專利範固内所述之均等變化^ 圍。即凡依本發明 專利範圍之精神與範鳴所涵ir化或修鄭’皆仍屬本發明46645 2 V. Description of Invention (24) 'Data S310. In addition, in the above-mentioned procedure, the color data for characteristic detection is used for more than one viewing area of the image on another stop, so an action will occur. When the viewing area is selected, the tool detects a part. In which, if the face is checked and the chasing object in the corresponding area is selected again, it exists. Therefore, by allowing the characteristic detection device 3〇 ^ to rearrange the image S330 ', and then to re-identify the area corresponding to the re-correspondence, if there is a tracking object in the re-corresponding area, the Carefully adjust the camera so that the image hole can be re-arranged after S 3 50 on the center point of the screen. Then, when multiple relocation information S3 60 is used in combination, the three-dimensional information will be generated with the reference 3 in the surface pattern generating device 500 to generate a new surface pattern S370 that varies from person to person. If you do n’t have the correct face, the camera ’s movement is calculated to calculate the camera ’s return to the normal position (general photography ^ tjS341 received image signal. This specific embodiment implements an item that can generate skin The data and action data are as described above 'if it is a d in the database 6 0', then the new surface model will be generated smoothly, because the device 300 will cooperate to make at least one of the data in the ^ number S320 The embodiment will detect the component S320_1, and then, a face system I detects its characteristics, and the image is reasonable. The present invention-tracking object S340. When there is a face, a face owned by No. Then the image will be generated by rearranging the bits of the image. The three-dimensional 3 modes are compared with each other, and the opportunity is adjusted by the integer value of the 5 weeks indicated by it. Then, using the method described in it can Step S310. The new surface model can be judged by comparing the solid surface model data. 2. V. INTRODUCTION (25) The surface model system that is not qualified already exists in the database 600. 80. As a result of the judgment, if When there are the same masks in the database 60, the new masks will be regarded as-registered masks S39 'and the existing objects retrieved by the camera will be classified as qualified User. If there is no identical mask in the database 600, the new mask will be regarded as an unregistered mask, and the object will be used. Therefore, according to the above description, the present invention can rely on an image signal and continuously adjust the position of the camera at the same time. At the same time, the photography can be adjusted to the " sufficient-angle until: to make the face completely recognized. Therefore, the two acquired skin data and data are used to complete the identification since it can intelligently and accurately aim at a skeletal egg sacrifice to inscription #, by people's holes, without the user having to move the makeup To a suitable position to cooperate with the identification increase ^ ^ ^ ^ λ. ^ Mj ± 糸 charge. The method and device of the face camera and identification 4 of the present invention have been described in detail except for the above, which are merely examples of the present invention. This cannot be used to limit the 5J items of the present invention. The alterations of the patent application and Fan ^ around the solid. Where i.e. under this patent disclosure the spirit and scope of the bypass fan Ming Cheng repair or ir 'are still present invention
Claims (1)
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| KR19980049468 | 1998-11-18 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| TW466452B true TW466452B (en) | 2001-12-01 |
Family
ID=19558789
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| TW88119671A TW466452B (en) | 1998-11-18 | 1999-11-11 | Method and apparatus for photographing/recognizing a face |
Country Status (1)
| Country | Link |
|---|---|
| TW (1) | TW466452B (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8159536B2 (en) | 2004-06-14 | 2012-04-17 | Agency For Science, Technology And Research | Method for detecting desired objects in a highly dynamic environment by a monitoring system |
| US8244002B2 (en) | 2009-03-18 | 2012-08-14 | Industrial Technology Research Institute | System and method for performing rapid facial recognition |
| TWI397024B (en) * | 2007-01-16 | 2013-05-21 | Asustek Comp Inc | Method for image auto-selection and computer system |
| US8520907B2 (en) | 2010-11-30 | 2013-08-27 | Inventec Corporation | Sending a digital image method and apparatus thereof |
| TWI419060B (en) * | 2009-10-15 | 2013-12-11 | Univ Nat Kaohsiung Applied Sci | Human face recognition method using a pyramid sampling algorithm |
| TWI448999B (en) * | 2010-10-13 | 2014-08-11 | Univ Nat Cheng Kung | Sketch alive method and system using the method |
| TWI479857B (en) * | 2012-10-15 | 2015-04-01 | Everfocus Electronics Corp | PTZ camera automatic tracking method |
-
1999
- 1999-11-11 TW TW88119671A patent/TW466452B/en not_active IP Right Cessation
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8159536B2 (en) | 2004-06-14 | 2012-04-17 | Agency For Science, Technology And Research | Method for detecting desired objects in a highly dynamic environment by a monitoring system |
| TWI397024B (en) * | 2007-01-16 | 2013-05-21 | Asustek Comp Inc | Method for image auto-selection and computer system |
| US8244002B2 (en) | 2009-03-18 | 2012-08-14 | Industrial Technology Research Institute | System and method for performing rapid facial recognition |
| TWI419060B (en) * | 2009-10-15 | 2013-12-11 | Univ Nat Kaohsiung Applied Sci | Human face recognition method using a pyramid sampling algorithm |
| TWI448999B (en) * | 2010-10-13 | 2014-08-11 | Univ Nat Cheng Kung | Sketch alive method and system using the method |
| US8520907B2 (en) | 2010-11-30 | 2013-08-27 | Inventec Corporation | Sending a digital image method and apparatus thereof |
| TWI479857B (en) * | 2012-10-15 | 2015-04-01 | Everfocus Electronics Corp | PTZ camera automatic tracking method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP6844038B2 (en) | Biological detection methods and devices, electronic devices and storage media | |
| WO2019214201A1 (en) | Live body detection method and apparatus, system, electronic device, and storage medium | |
| JP3271750B2 (en) | Iris identification code extraction method and device, iris recognition method and device, data encryption device | |
| JP7026225B2 (en) | Biological detection methods, devices and systems, electronic devices and storage media | |
| KR101569268B1 (en) | Acquisition System and Method of Iris image for iris recognition by using facial component distance | |
| Raghavendra et al. | Exploring the usefulness of light field cameras for biometrics: An empirical study on face and iris recognition | |
| KR101781358B1 (en) | Personal Identification System And Method By Face Recognition In Digital Image | |
| KR101530255B1 (en) | Cctv system having auto tracking function of moving target | |
| CN105894702A (en) | Invasion detecting alarming system based on multi-camera data combination and detecting method thereof | |
| KR100326203B1 (en) | Method and apparatus for face photographing and recognizing by automatic trading a skin color and motion | |
| US20120281874A1 (en) | Method, material, and apparatus to improve acquisition of human frontal face images using image template | |
| JP2004192378A (en) | Face image processor and method therefor | |
| WO2018101247A1 (en) | Image recognition imaging apparatus | |
| CN110414381A (en) | Tracing type face identification system | |
| JP2005227957A (en) | Optimal face image recording device and optimal face image recording method | |
| GB2343945A (en) | Photographing or recognising a face | |
| WO2016203698A1 (en) | Face detection device, face detection system, and face detection method | |
| KR101729966B1 (en) | CCTV camera system having image recognition and voice guidance and an operating method thereof | |
| WO2021217764A1 (en) | Human face liveness detection method based on polarization imaging | |
| KR20190093799A (en) | Real-time missing person recognition system using cctv and method thereof | |
| KR101084914B1 (en) | Indexing management system of vehicle-number and man-image | |
| CN109255282A (en) | A kind of biometric discrimination method, device and system | |
| KR100347058B1 (en) | Method for photographing and recognizing a face | |
| TW466452B (en) | Method and apparatus for photographing/recognizing a face | |
| CN112464012A (en) | Automatic scenic spot photographing system capable of automatically screening photos and automatic scenic spot photographing method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| GD4A | Issue of patent certificate for granted invention patent | ||
| MM4A | Annulment or lapse of patent due to non-payment of fees |