201220211 六、發明說明: 【發明所屬之技術領域】 本發明係有關於一種數位監控系統的改良,特別是指— 種除了具有-般監控系統之功能外,並可以人物臉像辨識或 以曰期、時間來搜尋可疑或欲找尋人物的監控系統。 【先前技術】 目前應用之人臉影像辨識系統大都是採類比式收斂影像 框架的模式,其僅能紀錄特定應用領域小範圍内之少量有限 • 個數之人物臉像_資料,所運用的細-般是約在二公^ 寬場景的通道口或小區域門禁的人臉識別系統中,其除了範 圍較小的缺料,朋㈣景須較鮮科能過於複雜,^ 被測者的配合度也必馳高,否職法達到有效的人臉辨 識,由此可見,上述㈣的監控祕與人臉辨識方式仍有諸 多缺失,並非一實用的設計,而亟需加以改良。 【發明内容】 本發明之目的既在於提供—歸在超场景可同時對多 人進行辨識且具有人物搜尋機制及事件還原之監控系統。 可達上述功能之本發明高清影像式人臉辨識及監控系統 係使用小波處理技術及核框架下的流行學習等先進技術融合 進行人臉_,轉決自然光·影響與實際人臉辨識中的 高維數據小樣本等問題。 該系統主要是由取像模組、人臉辨識模組、資料處理模 組及搜尋模組所構成;取像模組包含有2〇〇百萬畫素以上之 攝影设備及網路視訊傳送設備並與遠端電腦主機連結;將鏡 201220211 頭及攝影機架設在特定地點,透過視訊傳送設備將影像傳送 至遠端監控之電腦,再由人臉辨識模組對經過監控範圍之人 物影像進行人臉辨識後,將車人臉影像顯示於監控電腦螢幕 上,同時由資料處理模組將經人臉辨識後之人物全身全景的 影像紀錄貯存;人物搜尋機制是由監控電腦即時按取顯示於 監控電腦螢幕上的人物臉像或另外輸入欲搜尋人物的影像圖 =,經由搜尋模組而會將該人物的全身及全場景影像顯示於 監控電腦躲上:亦或是只輸人日期、時間,經由搜尋模組 而會將行_時段所有人㈣全纽全場景影細示於監控 電腦螢幕上;另外,可事先將欲搜尋人物的影像圖片以備案 方式輸入監控電腦,當取像模組攝得該欲搜尋人物而經人臉 辨識模組分析辨識確認後,除了將該人物的影像紀錄貯存 外,會自動將該人物經過監控範圍時的全身及全場景影像顯 示於監控電腦螢幕上,並發出警訊通知處理。 【實施方式】 請參閱圖一及圖二所示,本發明高清影像式人臉辨識及 監控系統,主要是由取像模組0〇)、人臉辨識模組(20)、資 料處理模組⑽)及搜尋模組(4〇)所構成;取像歡⑽)包含 有以單機或複數個架設方式之200百萬晝素以上之攝影設備 (1)及視訊傳送設備(2)並與遠端電腦主機(6)連結,而該人臉 辨識模組(20)在人臉檢測時,主要是運用生物統計方法依靠 數子工具建立相應的人臉五官模板,藉由膚色與五官結構特 徵相結合的方法進行,並利用數學工具提取人臉中最具有代 表性的特徵(如五官間的相對應位置等訊息)來予以識別,資 201220211 料處理模組(30)則由檔案傳輸舰器⑶、貯存設傷⑷、資 料庫伺服器(5)及網頁伺服器等所設備。 實施方式是’賴f彡設備⑴架設在特定地點,每一單機 可攝得大場景多人物全身全彩晝面,且可清晰取得人臉影 像透過視訊傳送δ又備⑵集結後將影像傳送至遠端監控電腦 (6) ’同時由人臉辨識模組(2〇)對經過監控範圍之場景所攝得 之多人物(可達40〜50人)同時進行人臉辨識後,將人臉影像 顯不於監控電腦⑹螢幕或中央顯示設備(7)上,同時由資料 處理模組(30)的檔案傳輸飼服器⑶將經人臉辨識後之人物 全身全場景的影像紀錄貯存於貯存設備,而透過資料庫伺 服器(5)及網頁伺服器(8)得將影像資料傳送至經授權電腦 (9);搜尋模組(4〇)係與監控電腦(6)連結,當欲進行人物搜 哥時’可由監控電腦(6)即時按取顯示於監控電腦螢幕上的人 物臉像圖案或另外輸入欲搜尋人物的臉像圖片,經由搜尋模 組(40)運作而會將該人物的全身及全場景影像顯示於監控電 • 腦(6)螢幕或中央顯示設備(Ό上;亦或是只輸入曰期、時間, 經由搜尋模組(40)而會將該時段經過攝影設備(1)監控範圍 的人物全身及全場景的影像顯示於監控電腦(60)螢幕或中央 顯不設備(7)上;另外,可事先將欲搜尋人物的影像圖片以備 案方式輸入監控電腦(6) ’當取像模組(10)攝得該欲搜尋人物 時,經人臉辨識模組(20)分析辨識確認後,除了將該人物的 全身及全場影像紀錄貯存外,會自動將該欲搜尋人物的全身 及全場影像顯示於監控電腦(6)螢幕或中央顯示設備上, 並發出警訊通知處理。 201220211 在此種運高清影像式人臉辨識及監控系統用下,具有全 時監控錄影、一機雙大場景畫面多人同時辨識、支援最清晰 影像、影像放大不失真、數位照片式檔案管理、人物搜尋、 事故還原等功效,藉以取代人力且降低誤判,能大幅節省人 工的耗費,達更具實用的效果。 【圖式簡單說明】 圖一係本發明車牌自動辨識及監控系統示意圖 圖二係本發明車牌自動辨識及監控系統架構示意圖 【主要元件符號說明】 1攝影設備2視訊傳送設備3檔案傳輸伺服器*貯存設備 5資料庫伺服器6即時監電腦主機7中央顯示設備 8網頁伺服器9授權電腦10取項模組20人臉辨識模組 3〇資料處理模組40搜尋模組201220211 VI. Description of the Invention: [Technical Field] The present invention relates to an improvement of a digital monitoring system, in particular to a function of a general monitoring system, and can be recognized by a human face or in a flood season. Time to search for surveillance systems that are suspicious or looking for people. [Prior Art] Currently, the face recognition system used in the application mostly adopts the analogy convergence image frame mode, which can only record a small number of limited number of people's face images in a small range of specific application fields. Generally, in a face recognition system with a channel opening of a two-wide wide scene or a small area access control, in addition to a small area of missing material, the friend (four) scene must be too complicated for the fresh branch, ^ the fit of the testee It will also be high, and the job law will achieve effective face recognition. It can be seen that the above-mentioned (4) monitoring secrets and face recognition methods still have many defects, not a practical design, and need to be improved. SUMMARY OF THE INVENTION The object of the present invention is to provide a monitoring system that can identify multiple people at the same time in a super scene and has a person search mechanism and event restoration. The high-definition image-based face recognition and monitoring system of the present invention capable of achieving the above functions is integrated with advanced techniques such as wavelet processing technology and popular learning under the nuclear framework for the face _, which is high in natural light, influence and actual face recognition. Small data samples and other issues. The system is mainly composed of an image capture module, a face recognition module, a data processing module and a search module; the image capture module includes more than 2 million pixels of photography equipment and network video transmission. The device is connected to the remote computer host; the camera 201220211 head and the camera frame are located at a specific location, and the image is transmitted to the remote monitoring computer through the video transmission device, and then the face recognition module performs the human image of the monitored range. After face recognition, the car face image is displayed on the monitor computer screen, and the data processing module stores the image record of the whole body panoramic image of the person after the face recognition; the character search mechanism is displayed by the monitoring computer on the monitor. The face image on the computer screen or another image of the person to be searched = the whole body and the full scene image of the character will be displayed on the monitoring computer through the search module: or only the date and time of the person will be entered. Through the search module, the line_time owner (four) all the new scenes will be displayed on the monitor computer screen; in addition, the image of the person to be searched can be searched in advance. The film is input into the monitoring computer in the form of record. When the image capturing module captures the character to be searched and is identified and confirmed by the face recognition module, the character will be automatically monitored except when the image record of the character is stored. The full body and full scene images are displayed on the monitor computer screen and alert notifications are processed. [Embodiment] Please refer to FIG. 1 and FIG. 2, the high-definition image type face recognition and monitoring system of the present invention mainly includes an image capturing module (0), a face recognition module (20), and a data processing module. (10)) and the search module (4〇); take the image (10)) containing more than 200 million elements of photographic equipment (1) and video transmission equipment (2) in a single or multiple erection The computer host (6) is connected, and the face recognition module (20) mainly uses a biometric method to establish a corresponding facial five-point template by using a biometric method, and the skin color and the facial features are related to each other. Combine the method and use the mathematical tool to extract the most representative features of the face (such as the corresponding position between the five senses) to identify, the 201220211 material processing module (30) is transferred by the file transmission (3) , storage and injury (4), database server (5) and web server and other equipment. The implementation method is that the 'Lai F彡 device (1) is installed at a specific place, each single machine can capture a large scene, a multi-person full-color full face, and can clearly obtain the face image through the video transmission δ and prepare (2) to assemble the image to Remote monitoring computer (6) 'At the same time, the face recognition module (2〇) will face the people who have been photographed in the monitored range (up to 40~50 people) and face recognition at the same time. It is not visible on the monitor computer (6) screen or the central display device (7), and the file transfer feeder (3) of the data processing module (30) stores the image record of the full-body scene of the person after the face recognition in the storage device. The image data is transmitted to the authorized computer (9) through the database server (5) and the web server (8); the search module (4) is connected to the monitoring computer (6), when the character is to be selected When searching for a brother, the monitor computer (6) can instantly press the face image displayed on the monitor computer screen or input the face image of the person to be searched. The search module (40) will operate the person's body. And full scene image is displayed in the monitor • Brain (6) screen or central display device (on the top; or just enter the time, time, via the search module (40), the person who will pass the time through the camera (1) monitors the whole body and the full scene The image is displayed on the monitor computer (60) screen or the central display device (7). In addition, the image of the person to be searched can be input to the monitor computer in advance (6) 'When the image capture module (10) is taken When searching for a character, after the face recognition module (20) analyzes the identification and confirmation, in addition to storing the full body and full-field image records of the character, the whole body and the full-field image of the person to be searched are automatically displayed on the image. Monitor computer (6) on the screen or central display device, and send out alert notification processing. 201220211 Under this kind of high-definition video-based face recognition and monitoring system, it has full-time surveillance video, one machine and two large scenes. Identify and support the most clear images, image distortion without distortion, digital photo file management, character search, accident reduction, etc., in order to replace manpower and reduce misjudgment, can greatly save labor costs Figure 1 is a schematic diagram of the automatic identification and monitoring system of the license plate of the present invention. FIG. 2 is a schematic diagram of the automatic identification and monitoring system of the license plate of the present invention. [Main component symbol description] 1 photographic device 2 video transmission Device 3 file transfer server * storage device 5 database server 6 real-time monitoring computer host 7 central display device 8 web server 9 authorized computer 10 pick-up module 20 face recognition module 3 data processing module 40 search module group