-M3 62474 * 五、新型說明： 【新型所屬之技術領域】 本創作係關於一種應用於提款機（ATM)之人臉辨識認證 系統，特別是指一種利用儲存擁有人臉部特徵之晶片提款 卡，經與提款機（ATM)上所擷取之人臉特徵相互驗證比對之 身份認證系統，用以防止提款卡擁有人與使用者不為同一 '人，發生冒領或盜領的狀況。 ·'【先前技術】 一般的提款機（ATM)，提款流程是插入晶片提款卡，然 後提款機（ATM)提示輸入密碼，鍵入正確密碼後，便可開始 進行提款或轉帳的操作；這種密碼驗證方式雖然方便，但 是卻有很大的風險，因輸入密碼是認卡，僅確定提款卡與 輸入的密碼是同一組，並無法確認提款卡是目前使用者所 擁有，因為密碼可以轉述、冒用，因此在提款安全認證上 ® 有很大的漏洞。 後來有人將人臉辨識認證方式導入提款機（atm)中，其 做法為持卡人在開卡時做一次人臉影像之訓練，可獲得持 卡人之特徵/權重資料，存放在資料庫内，做為以後提款時 使用；而在提款時，從持卡人金融卡之卡號及密碼驗證， 即可到本地/遠端監控中心查尋持卡人之特徵/權重資料，據 以建立持卡人之類神經網路，再將持卡人新的特徵向量輸 入，若運算辨識結果之輸出值2臨限值，即表示辨識成功’ 3 •-M362474 =可以：行提款/其他服務，否則，即警示監控人員有異 “領或冒領之情況’這種做法已揭露於專利「類神經網 路為主之金融卡櫃員機防止冒領與預警監控系統」（證書 號.m52G5)巾；「持卡人於卡片開切，做—次人臉影像 之訓練，以獲得持卡人之特徵/權重資料」，此做法會衍生 -些問題’因為卡片持卡人並不了解這套系統，完全由使 用者於提款機（ATM)開卡時取得人臉特徵，極可能會取到錯 誤的特徵’像是使用者與旁人說話未正視攝影機，或戴墨 鏡、戴口罩等’無法確保所獲得持卡人之特徵/權重資料是 正確的；另外’先前技術中將人臉特徵/權重資料放置於本 地/遠端監控中心，在比對驗證時要透過網路連線至本地/遠 端監控中心做查詢與回覆動作，除了多花費時間、金錢與 網路頻寬的使用外’也增加了網路擁塞、斷線所帶來的風 險，加上以本地/遠端監控中心的這種查詢方式，當本地/遠 知控中〜同時面對多方提款機（atm)同時做查詢時，會有 忙碌延遲的狀況；而且當提款時不是在本地端監控中心而 疋在另一縣市的提款機時’此種做法會造成需要在每—個 本地端都散佈每個人的人臉特徵/權重資料，不然在另一縣 市提款時，會無法做人臉特徵的比對，如此設計系統又更 加複雜、容易出錯。 由此可見’上述習用技術仍有諸多缺失，實非一良善之 设什者，而亟待加以改良。 -M362474 本案創作人鑑於上述習用技術所衍生的各項缺點，乃亟 思加以改良創新，並經多年苦心孤諸潛心研究後，終於成 功研發完成本件應用於提款機（ATM)之人臉辨識認證系統。 【新型内容】 本創作之目的即在於提供一種應用於提款機（ATM)之人 臉辨識認證系統，改善一般提款機（ATM)與晶片提款卡的身 .份驗證方式，加入人臉辨識的生物特徵，做為驗證提款卡 籲擁有人與使用者是否為同一人的機制，用以防止冒領或盜 領的發生。 可達成上述新型目的之應用於提款機（ATM)之人臉辨識 認證系統’包括有一晶片提款卡，該晶片内設置有一臉部 特徵儲存模組，該臉部特徵儲存模組係儲存擁有人臉部特 徵資料；-提款機(ATM) ’該提款機内包含有一特徵讀取單 元、一拍攝單元、一擷取與特徵抽取單元及一特徵驗證單 凡；使用者在申辦晶片提款卡時，先要拍攝一連串的臉部 影像’經程式軟體或人工篩選角度適合的人臉影像，再抽 取人臉影像中之臉部特徵，將发 a 將’、冩入晶片&款卡之臉部特 徵儲存模組内；當欲接描畔A〒 呆棱忒機（ATM)時’使用者操作流程與 傳統方式完全相同，插入摇±A 一 棚入钕放卡並按下個人密碼，同時於 提款機（ATM)内之特徵古啬跑留_矣 做凟取単疋透過晶片提款卡上之輸出入 介面，也將晶片提款卡之擁 下<擁有人臉部特徵資料讀取出來， 準備與使用者進行臉部特徵 文比對，另外，於提款機（ΑΤΜ)内 5 •M362474-M3 62474 * V. New Description: [New Technology Area] This creation is about a face recognition authentication system applied to an ATM (ATM), especially a chip that uses the characteristics of the owner's face. The card is authenticated by the identity authentication system that is mutually verified with the face features captured on the ATM (ATM) to prevent the owner and the user of the ATM card from being the same person. The status of the collar. · '[Prior Art] General ATM (ATM), the withdrawal process is to insert the chip ATM card, and then the ATM (ATM) prompts for the password. After entering the correct password, you can start the withdrawal or transfer. Operation; this password verification method is convenient, but it has a great risk. Because the password is a card, it is only determined that the ATM card is the same group as the password entered, and it cannot be confirmed that the ATM card is owned by the current user. Because the password can be narrated and fraudulent, there is a big loophole in the withdrawal security certification. Later, some people introduced the face recognition authentication method into the ATM (atm). The practice is that the cardholder can do a face image training when the card is opened, and the cardholder's feature/weight data can be obtained and stored in the database. In the future, when used for withdrawal, the card number and password of the cardholder's financial card can be used to check the cardholder's characteristics/weight data at the local/remote monitoring center. A neural network such as a cardholder, and then input the new feature vector of the cardholder. If the output value of the operation identification result is 2, the identification is successful. 3 • M362474 = Yes: Bank withdrawal / other services Otherwise, it is a warning that the monitoring personnel have different "collar or fraudulent conditions". This practice has been exposed in the patent "Nu-Nerve Network-based Financial Card Teller Machine Preventing Precaution and Early Warning Monitoring System" (Certificate No. m52G5) "The cardholder opens the card and does the training of the face image to obtain the cardholder's characteristics/weight information." This practice will lead to some problems - because the card cardholder does not understand the system. , completely withdraw by the user When the machine (ATM) opens the card, the face features are obtained, and it is very likely that the wrong feature will be obtained. 'If the user talks with someone else, the camera is not facing the camera, or wearing sunglasses, wearing a mask, etc.' cannot ensure the characteristics of the card holder obtained / The weighting data is correct; in addition, in the prior art, the face feature/weight data is placed in the local/remote monitoring center, and the comparison and verification are performed through the network connection to the local/remote monitoring center for query and reply actions. In addition to spending more time, money and network bandwidth, it also increases the risk of network congestion and disconnection, plus the local/remote monitoring center's query method, when local/far In the knowledge control - when facing the multi-party cash machine (atm) at the same time, there will be a busy delay; and when the withdrawal is not at the local monitoring center and in the cash machine of another county and city' This approach will result in the need to spread the facial features/weights of each person on each local end. Otherwise, in another county and city, when the withdrawal is made, the face features will not be matched. The design system is more complicated. Easy to come out wrong. It can be seen that there are still many shortcomings in the above-mentioned conventional technologies. It is not a good one, and it needs to be improved. -M362474 In view of the shortcomings derived from the above-mentioned conventional technologies, the creators of this case have improved and innovated, and after years of painstaking research, they finally succeeded in developing and developing the face recognition for the ATM machine. Certification system. [New content] The purpose of this creation is to provide a face recognition and authentication system for cash dispensers (ATM), to improve the body verification method of the ATM and the ATM card, and to join the face. The identified biometrics are used as a mechanism to verify whether the owner and the user are the same person to prevent fraud or theft. A face recognition authentication system for an ATM machine that achieves the above novel object includes a chip cash card having a face feature storage module disposed therein, the face feature storage module being stored and owned Face feature data; - ATM (ATM) 'The cash dispenser includes a feature reading unit, a shooting unit, a capture and feature extraction unit, and a feature verification unit; the user withdraws funds from the application wafer When the card is used, it is necessary to take a series of facial images 'programmed software or artificially screen the face image suitable for the angle, and then extract the facial features in the face image, and send a, 'put into the chip & card In the facial feature storage module; when the user wants to pick up the A〒 忒 忒 A A A ' ' 使用者 使用者 使用者 使用者 使用者 使用者 使用者 使用者 使用者 使用者 使用者 使用者 使用者 使用者 使用者 使用者 使用者 使用者 使用者 使用者 使用者 使用者 使用者 使用者 使用者 使用者 使用者 使用者 使用者 使用者 使用者 使用者 使用者 使用者At the same time, the characteristics of the ATM in the cash machine (ATM) are _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ Read out, prepare and use Facial features for text comparison, addition, in the dispenser (ΑΤΜ) 5 • M362474
。。 ~連串使用者的臉部影像，接著擷取與特 徵抽取単元針對這一連串的影像先擷取出角度適合的臉部 影像、，再從適合的臉部影像中抽取出該使用者臉部特徵資 科’並利用特徵驗證單元直接與特徵讀取單元所讀取出來 之擁有人臉部特徵資料做驗證；若比對兩者臉部特徵之相 似度高過—門襤值時’即認為晶片提款卡擁有人與使用者. . ~ a series of facial images of the user, and then extracting and extracting the facial image suitable for the series of images from the series of images, and extracting the facial features of the user from the appropriate facial images And use the feature verification unit to directly verify the feature data of the owner's face read by the feature reading unit; if the similarity between the two face features is higher than the threshold value, the chip is considered to be Card owner and user
為同一人’可繼續進行提款或轉帳等服務；若比對兩者臉 寺徵之相似度低於⑽值時，則認為晶片提款卡擁有人 與使用者不為同一人，可能有冒領或盜領的狀況發生，系 統可即時發出告警資訊給銀行的監控中心或警政單位或僅 留下記錄供未來異常查詢。 【實施方式】 請參閱圖一，本創作所提供之應用於提款機（ATM)之人 臉辨識認證系統之晶片提款卡架構圖，主要包括有· -晶片提款卡卜該晶片内設置有—臉部特徵儲存模组 11 ’該臉部特徵儲存模組n係儲存擁有人臉部特徵資料； 係透過晶片提款卡之輸出人介面，對臉部特徵儲存模_ 進行寫入與讀取之動作；使用者在臨櫃申辦提款卡時，先 要拍攝-連串人臉影像，經程式軟體或人工篩選角度適合 的人臉影像’經由人臉操取與特徵抽取程序，將該卡片擁 有人臉部特徵資料透過晶片提款卡之輸出入介面寫入晶片 内之臉部特徵儲存模㈣内’·其中，臉部特徵儲存模㈣ 6 M362474 ' 的容量需足以存入擁有人臉部特徵資料；當欲操作提款機 (ATM)時，使用者除了透過晶片提款卡之輸出入介面驗證其 卡號/密碼外，還會將儲存於臉部特徵儲存模組11内的擁有 人臉部特徵資料讀取出來，準備進行人臉身份驗證。 請參考圖二所示，為本創作應用於提款機（ATM)之人臉 辨識認證系統之晶片提款卡與提款機（ATM)認證流程圖，主 要包括有· 一提款機（ATM)2，該提款機（ATM)2内包含有： 一特徵讀取單元2 1，該特徵讀取單元2 1係透過晶片提款 卡1上之輸出入介面，將晶片提款卡1上儲存於臉部特徵儲 存模組内之擁有人臉部特徵資料讀取出； 一拍攝單元22，拍攝一連串使用者的臉部影像； 一擷取與特徵抽取單元23，係由拍攝單元23拍攝之影像 中，擷取出角度適合的臉部影像，再從這些適合的影像中 • 抽取出該使用者臉部特徵資料； 一特徵驗證單元24 ;該特徵驗證單元24係對特徵讀取單 元21所讀取出來之擁有人臉部特徵資料與目前使用者臉部 特徵資料做一對一的驗證比對； 當欲操作提款機（ATM)2時，使用者之操作流程與一般 提款方式完全相同；晶片提款卡1插入並按下個人密碼後， 透過卡號讀取與密碼驗證確認晶片提款卡1與個人密碼是否 符合，若不符合，則提供密碼錯誤警示訊息，三次後沒入 •:M362474 卡二，㈣，在使用者插卡的觸發下，特徵讀取單元叫 過晶片提款卡1上之輪Ψ 入 輪出入介面，將晶片提款卡！上儲存於 臉部特徵儲存模組U内之擁有人臉部特徵資料讀取出來暫 存，準備與使用者進行臉部特徵比對；另外，拍攝單元a 開始拍攝一連串使用者的臉部影像，並以類取與特徵抽取 .^23針對這—連串的影像中，榻取出角度適合的臉部影 y再仏些適合的臉部影像中抽取出該使㈣臉部特徵 I貝科，這裡的梅取與特徵抽取單元23與晶片提款卡i擁有人 在臨櫃申辦時所使用係為一致的軟體模組，如此所榻取的 臉^特徵才會一致；利用特徵驗證單元24對特徵讀取單元 21所讀取出來之擁有人臉部特徵資料與目前使用者臉部特 徵資料做一對一的驗證比對’特徵驗證單元24比對兩者臉 部特徵相似度高過—門棰值時，即認為晶片提款卡i擁有人 f 用者為同人，且當晶片提款卡1與個人密碼也符合的 清形下’便可繼續進行提款或轉帳等服務；若比對後兩者 臉部特徵相似度低於過門檻值時，則系統認為晶片提款卡i 擁有人與使用者不為同一人’可能有冒領或盜領的狀況發 系統可拒絕使用者繼續操作並即時發出告警資訊給 監控單位’或系統可讓使用者繼續操作，只留下異常記錄 供歷史資料查詢。 ϋ之拍攝單元’可以是利用硬體感應器觸發拍照，亦 可以是利用軟體程式自動積測影像變化的方式觸發。 8 :M362474 上述之擷取與特徵抽取單元可以是習知的影像切割與特 徵抽取方法’並可以針對單張或數張人臉影像組合成一個 人臉特徵。 上述之擁有人臉部特徵資料係為從一連串的臉部影像 中’經程式軟體或人H選角度適合的單張或數張人臉影 像，再抽料些人臉影像巾之擁有人臉料徵資料，並將 其寫入晶片提款卡之臉部特徵儲存模組。For the same person, you can continue to make withdrawals or transfer services. If the similarity between the two faces is less than (10), the owner of the ATM card is not the same person as the user. When the situation of the collar or the thief takes place, the system can immediately send an alarm message to the monitoring center or policing unit of the bank or leave only records for future abnormal inquiry. [Embodiment] Please refer to Figure 1. The chip decanter card architecture diagram of the face recognition authentication system applied to the ATM (ATM) provided by this creation mainly includes the - wafer withdrawal card. a facial feature storage module 11 'the facial feature storage module n stores the owner's facial feature data; the face feature storage module _ is written and read through the output interface of the chip ATM card Take the action; when the user applies for the ATM card in the counter, the user must first take a series of face images, and the face image of the appropriate angle can be filtered by the program software or manually. The face feature data of the card owner is written into the face feature storage module (4) in the wafer through the input and output interface of the wafer cash card, and the capacity of the face feature storage module (4) 6 M362474' is sufficient to be stored in the face. Partial profile data; when the ATM is to be operated, the user will verify the card number/password through the input interface of the chip ATM card, and will also store the owner in the face feature storage module 11. Facial features The data is read out and ready for face authentication. Please refer to Figure 2 for the flow chart of the ATM card and ATM (ATM) authentication for the face recognition and authentication system applied to the ATM (ATM), including the ATM (ATM) 2, the cash dispenser (ATM) 2 includes: a feature reading unit 2 1 , the feature reading unit 2 1 is through the input and output interface on the wafer cash card 1 , the chip cash card 1 The owner's facial feature data stored in the facial feature storage module is read out; a shooting unit 22 captures a series of facial images of the user; and a capture and feature extraction unit 23 is captured by the imaging unit 23. In the image, the face image with the appropriate angle is taken out, and the user face feature data is extracted from the suitable images; a feature verification unit 24; the feature verification unit 24 reads the feature reading unit 21 The extracted facial feature data is compared with the current user facial feature data for one-to-one verification; when the cash machine (ATM) 2 is to be operated, the user's operation flow is exactly the same as the general withdrawal method. ; the chip cash card 1 is inserted and pressed After the password, confirm whether the chip withdrawal card 1 and the personal password match by the card number reading and password verification. If it does not match, provide the password error warning message. After three times, the message is not included. • M362474 card 2, (4), the user inserts Under the trigger of the card, the feature reading unit calls the rim wheel access interface on the wafer cash card 1 and the chip withdrawal card! The facial feature data stored in the facial feature storage module U is temporarily read out, and is ready to be compared with the user for facial features. In addition, the shooting unit a starts to capture a series of facial images of the user. And take the class extraction and feature extraction. ^23 For this - a series of images, the face angle y that fits the angle of the couch is extracted and the appropriate facial image is extracted to extract the (4) facial features I. The feature module of the fetching and feature extraction unit 23 and the owner of the wafer debit card i are used in the application of the counter, so that the features of the face are consistent; the feature verification unit 24 is used for the feature The owner's facial feature data read by the reading unit 21 is compared with the current user facial feature data for one-to-one verification comparison. The feature verification unit 24 compares the facial features with higher similarity than the threshold. At the time of the value, it is considered that the owner of the chip withdrawal card i is the same person, and when the wafer withdrawal card 1 and the personal password are also in compliance with the clearing form, the service such as withdrawal or transfer can be continued; The facial features of the two are similar to each other. When the threshold is devalued, the system considers that the owner and the user of the chip withdrawal card i are not the same person. 'There may be cases of fraudulent or stolen. The system may refuse the user to continue operation and immediately send an alarm message to the monitoring unit' or the system. Allows the user to continue the operation, leaving only the exception record for historical data query. The camera unit can be triggered by a hardware sensor, or it can be triggered by a software program that automatically detects image changes. 8: M362474 The above extraction and feature extraction unit may be a conventional image cutting and feature extraction method' and may combine a single or several facial images into one facial feature. The facial feature data of the above-mentioned owner is a single or several facial images suitable for the angle of the program software or person H from a series of facial images, and then the face of the face image towel is taken. The data is collected and written into the facial feature storage module of the chip ATM card.
本創作所提供之應用於提款機（ATM)之人臉辨識認證系 、’充’、則述引5登案及其他習用技術相互比較時，更具有下 列之優點： 1.本創作可大幅提高於提款機（ATM)之身份認證安全 陡’避免冒領或盜領的狀況發生。 2·本創作晶片提款卡内之人臉特徵資料，係於使用者在 申辦提款卡時’所拍攝與選取角度適合的人臉影像進 行訓練成人臉特徵，避免錯誤的訓練影像造成錯誤的 人臉特徵，可大幅增加人臉特徵比對的正確性。 3. 本創作對於使用者在提款機（ATM)的操作，除了在申 辦提款卡時配合拍攝臉部影像外，其餘完全與一般提 ^機（ATM)的操作方式相同，對使用者操作的影響最 小’相對操作錯誤的機率也大幅降低。 4. 本創作將人臉特徵資料儲存於晶片提款卡上，直接讀 取出來與提款機（ATM)上使用者的人臉做一對一特徵 ：M362474 比對，資訊無須透過本地/遠端監控中心的網路傳送 與驗證，架構簡單且到那裡都可通用。 上列詳細說明係針對本創作之一可行實施例之具體說 明’惟該實施例並非用以限制本創作之專利範圍，凡未脫 離本創作技藝精神所為之等效實施或變更，均應包含於本 案之專利範圍中。 综上所述，本案不但在空間型態上確屬創新，並能較習 用物品增進上述多項功效’應已充分符合新賴性及進步性 之法定新型專利要件’爰依法提出申請，懇請貴局核准 本件新型專利申請案’以勵創作’至感德便。 -M362474 【圖式簡單說明】 圖一為本創作應用於提款機（ATM)之人臉辨識認證系 統之臉部特徵儲存模組架構圖， 圖二為本創作應用於提款機（ATM)之人臉辨識認證系 統之晶片提款卡與提款機（ATM)認證流程圖。 【主要元件符號說明】 1晶片提款卡 11臉部特徵儲存模組 2提款機（ATM) 21特徵讀取單元 22拍攝單元 23擷取與特徵抽取單元 24特徵驗證單元The face recognition and authentication system, the 'charge', the introduction of 5 cases and other conventional techniques provided by the creation of the ATM have the following advantages: 1. The creation can be greatly Improve the identity of the ATM (ATM) and securely avoid the situation of taking the lead or stealing. 2. The face feature data in the ATM card of the creation is based on the face image of the user who is shooting and selecting the angle to perform the training of the adult face feature, so as to avoid the wrong training image and cause the error. Face features can greatly improve the correctness of face feature comparison. 3. This creation is the same as the operation of the general lift machine (ATM) for the operation of the user at the cash machine (ATM), except for the operation of the facial image. The impact of the smallest 'relative operational errors is also greatly reduced. 4. This creation stores the face feature data on the chip ATM card and directly reads it out from the user's face on the cash machine (ATM). The M362474 is compared. The information does not need to be transmitted through the local/far. The network transmission and verification of the monitoring center is simple and can be used everywhere. The detailed description above is a detailed description of one of the possible embodiments of the present invention. The present invention is not intended to limit the scope of the present invention. Any equivalent implementation or modification that is not departing from the spirit of the present invention should be included in The patent scope of this case. In summary, this case is not only innovative in terms of space type, but also can enhance the above-mentioned multiple functions compared with the customary items. 'The statutory new patent requirements that should be fully in line with the new and progressive nature', apply in accordance with the law, and ask for your application. Approved this new type of patent application 'to encourage creation' to the sense of virtue. -M362474 [Simple description of the diagram] Figure 1 is a schematic diagram of the facial feature storage module of the face recognition authentication system applied to the ATM (ATM), and Figure 2 is applied to the ATM (ATM). The chip withdrawal card and cash dispenser (ATM) certification flow chart of the face recognition authentication system. [Main component symbol description] 1 chip ATM card 11 face feature storage module 2 cash dispenser (ATM) 21 feature reading unit 22 shooting unit 23 capture and feature extraction unit 24 feature verification unit
1111