201222433 六、發明說明： 【發明所屬之技術領域】 [0001] 本發明是有關於一種身分驗證裴置及其方法，特別是有 關於一種基於生物特徵之身分驗證裝置及其方法。 【先前技術】 [0002] 一直以來使用鑰匙或個人密碼作為安全防護的工具容易 因小偷、駭客入侵或遺忘密碼而使安全出現漏洞及不便 性。 [0003] 隨著生物特徵辨識技術日益的進步，使用生物特徵作為 辨識個人身分的方式相當普及\目前習知的生物特徵辨 識方法多為利用圖樣比對（pattern matching)或特徵 點比對（minutiae matching)此兩種方式進行身分辨識 。然而利用圖樣比對之生物特徵辨識技術，其缺點在於 圖像的資料量大，且於比對過程中還有位移、旋轉角度 及圖形形變等問題需要處理，故比對速度缓慢，容易浪 費使用者等待身分驗證的時.間。 [〇〇〇4]另一方面，利用特徵點比對之生物特徵辨識技術，其優 點雖較圖樣比對快速，但缺點在於當使用者的習慣改變（ 以指紋為例但不限於：按壓或滑動的角度/速度）或環境 改變（譬如但不限於：溫座度）’則可能造成所㈣的兄 徵點數量不足，故難以藉由特徵點比對之生物特徵辨特 技術取得身分驗證，如此—來，不僅造成使用者的=熾 ，更失去了利用獨特的生物特徵作為身分驗證的後勢便 _]，综上所述，習知的生物特_識技術無法快迷而有々 辨識-待辨識標的，以進行身分驗證的動作 敦的 1 表單編號A0101 第4頁/共31頁 匕基於 201222433 上述原因考量’本發明之發明人思索並設計—種基於生 物特徵之身分驗證裝置及其方法，關針對現有技術之 缺失加以改善，進而增進產業上之實施利用。 【發明内容】 [0006] 有鑑於上述習知枯蓺夕”韻，太麻α 枝*之門喊本發明之目的就是在提供 -種基於生物特徵之身分驗證裝置及其方法，以解決圖 樣比對過慢且資料量大以及特徵點過少而無法利用特徵 點比對法通過身分驗證的問題。201222433 VI. Description of the Invention: [Technical Field] [0001] The present invention relates to an identity verification device and a method thereof, and more particularly to a biometric-based identity verification device and method thereof. [Prior Art] [0002] The use of a key or a personal password as a security tool has been prone to security breaches and inconveniences due to thieves, hackers, or forgotten passwords. [0003] With the increasing progress of biometrics, the use of biometrics as a means of recognizing personal identity is quite popular. The current methods of biometric identification mostly use pattern matching or feature point comparison (minutiae). Matching) These two ways to identify the identity. However, the biometric identification technology using pattern comparison has the disadvantage that the amount of data of the image is large, and there are problems such as displacement, rotation angle and graphic deformation in the comparison process, so the comparison speed is slow and easy to waste. Waiting for the time of identity verification. [〇〇〇4] On the other hand, the biometric identification technology using feature point comparison has the advantages of faster comparison with the pattern, but the disadvantage is that when the user's habit changes (taking the fingerprint as an example but not limited to: pressing or Sliding angle/speed) or environmental changes (such as but not limited to: warm seat) may result in insufficient number of sibling points in (4), so it is difficult to obtain identity verification by feature point comparison biometric identification technology. In this way, not only does the user's = blazing, but also loses the use of unique biometrics as the post-emphasis of identity verification _]. In summary, the traditional bio-technical technology can't be fascinated and identified. - The object to be identified for the purpose of identity verification 1 Form No. A0101 Page 4 / Total 31 pages based on 201222433 Considering the above reasons, the inventor of the present invention ponders and designs a biometric-based identity verification device and The method is to improve the lack of existing technology, thereby enhancing the implementation and utilization of the industry. SUMMARY OF THE INVENTION [0006] In view of the above-mentioned conventional habits, the door of the invention is to provide a biometric-based identity verification device and method thereof to solve the pattern ratio. The problem of passing the identity verification by using the feature point comparison method is too slow and the amount of data is large and the feature points are too small.
[_根據本發明之目的，提出¥種基於生物特徵之身分驗證 裝置，其包括一儲存單元、一影像擷取單元、一生物特 徵擷取單元以及—處理單元。儲存單元係儲存—第一生 物特徵圖像及對應H特徵师訊。影像擁取單元 係對一待驗證標的擷取一第二生物特徵圖像。生物特徵 擷取單元係分析第二生物特徵圖像之複數個特徵點並對 應提供-第二特徵點資m。處理單元係>別電性連接儲 存單元、影像擷取單元以及生物特徵擷取單元。處理單 元係比對第一特徵點資訊與第二特徵點資訊，以得出一 第一比對結果及一對位資訊，根據對位資訊進行第一生 物特徵圖像與第二生物特徵圖像的對位，再進行第—生 物特徵圖像與第二生物特徵圖像的圖樣比對，以得出一 第二比對結果，然後根據第一比對結果與第二比對結果 判斷待驗證標的是否通過身分驗證。 [0008] 其中，處理單元更根據一預設權重比例，將第一卜According to the purpose of the present invention, a biometric-based identity verification device is proposed, which includes a storage unit, an image capture unit, a biometric capture unit, and a processing unit. The storage unit is stored - the first biometric image and the corresponding H feature teacher. The image capturing unit captures a second biometric image for a to-be-verified object. The biometric capture unit analyzes a plurality of feature points of the second biometric image and provides a second feature point m. The processing unit is electrically connected to the storage unit, the image capturing unit, and the biometric capturing unit. The processing unit compares the first feature point information with the second feature point information to obtain a first comparison result and a pair of bit information, and performs the first biometric image and the second biometric image according to the alignment information. The alignment of the first biometric image and the second biometric image are compared to obtain a second alignment result, and then the first comparison result and the second comparison result are determined to be verified. Whether the target is verified by identity. [0008] wherein the processing unit further according to a preset weight ratio, the first
••U 果與第二比對結果透過一運算式計算一第三比對妹果， 並根據第三比對結果與一門檻值，判斷待驗證樑的是否 099141103 表單編號A0101 第5頁/共31頁 0992071533-0 201222433 通過身分驗證。 [0009] 其中，處理單元更根據第一生物特徵圖像與第二生物特 徵圖像的相似度得出第二比對結果，並根據第二比對結 果透過一運算式計算一門檻值，再根據第一比對結果與 門檻值，判斷待驗證標的是否通過身分驗證，其中運算 式係使第二比對結果對應之相似度與門檻值呈負相關。 [0010] 其中，處理單元係根據對位資訊對位第一生物特徵圖像 與第二生物特徵圖像，特徵擷取單元更根據第二特徵點 資訊辨識出第二生物特徵圖像之一奇異點（S i ngu 1 ar Po i n t)，處理單元以奇異點為中心向四周展開一特定區 域範圍，並以圖樣比對的方式比對特定區域範圍内之第 一生物特徵圖像與第二生物特徵圖像，而得到第二比對 結果。 [0011] 其中，處理單元係根據對位資訊對位第一生物特徵圖像 與第二生物特徵圖像，並以縮圖的方式進行第一生物特 徵圖像與第二生物特徵圖像的圖樣比對，而得到第二比 對結果。 [0012] 根據本發明之目的，再提出一種基於生物特徵之身分驗 證方法。基於生物特徵之身分驗證方法適用於一基於生 物特徵之身分驗證裝置，此基於生物特徵之身分驗證裝 置包含一儲存單元、一影像擷取單元、一生物特徵擷取 單元以及一處理單元。基於生物特徵之身分驗證方法包 含下列步驟：首先以儲存單元儲存第一生物特徵圖像及 對應之第一特徵點資訊；然後利用影像擷取單元對一待 099141103 表單編號A0101 第6頁/共31頁 0992071533-0 201222433 驗證標的擷取一第二生物特徵圖像；再藉由生物特徵擷 取單元分析第二生物特徵圖像之複數個特徵點並對應提 供一第二特徵點資訊；接著透過處理單元比對第一特徵 點資訊與第二特徵點資訊，以得出一第一比對結果及一 對位資訊；接下來以處理單元根據對位資訊進行第一生 物特徵圖像與第二生物特徵圖像的對位，再進行第一生 物特徵圖像與第二生物特徵圖像的圖樣比對，以得出一 第二比對結果；以及利用處理單元根據第一比對結果與 第二比對結果判斷待驗證標的是否通過身分驗證。 Ο [0013] 其中，基於生物特徵之身分驗證方法更包含下列步驟： 根據一預設權重比例，利用處理單元將第一比對結果與 第二比對結果透過一運算式計算一第三比對結果；然後 藉由處理單元根據第三比對結果與一門檻值，判斷待驗 證標的是否通過身分驗證。 [0014] ❹ 其中，基於生物特徵之身分驗證方法更包含下列步驟： 利用處理單元根據第一生物特徵圖像與第二生物特徵圓 像的相似度得出第二比對結果；並以處理單元根據第二 比對結果透過一運算式計算一門檻值；以及藉由處理單 元根據第一比對結果與門檻值，判斷待驗證標的是否通 過身分驗證。其中，運算式係使第二比對結果對應之相 似度與門檻值呈負相關。 [0015] 其中，以處理單元進行圖樣比對之步驟中，更包含下列 步驟：利用處理單元根據對位資訊對位第一生物特徵圖 像與第二生物特徵圖像；然後以特徵擷取單元根據第二 特徵點資訊辨識出第二生物特徵圖像之一奇異點 099141103 表單編號Α0101 第7頁/共31頁 0992071533-0 201222433 (Singular Point);以及利用處理單元以奇異點為中 心向四周展開一特定區域範圍，並以圖樣比對的方式比 對特定區域範圍内之第一生物特徵圖像與第二生物特徵 圖像，而得到第二比對結果。 [0016] 其中，以處理單元進行圖樣比對之步驟中，更包含下列 步驟：利用處理單元根據對位資訊對位第一生物特徵圖 像與第二生物特徵圖像；以及透過處理單元以縮圖的方 式進行第一生物特徵圖像與第二生物特徵圖像的圖樣比 對，而得到第二比對結果。 [0017] 承上所述，依本發明之基於生物特徵之身分驗證裝置及 其方法，其可具有一或多個下述優點： [0018] ( 1 )此基於生物特徵之身分驗證裝置及其方法可藉由特 徵點比對所得到的對位資訊進行兩圖像對位，因特徵點 的資訊量比圖樣的資訊量少，故以特徵點（m i nu t i ae)對 位較圖樣（pattern)對位快，藉此可解決先前技術以圖 樣對位兩圖像所造成的耗時問題。 [0019] (2)此基於生物特徵之身分驗證裝置及其方法可藉由處理 單元依據不同的權重比例，將第一比對結果與第二比對 結果帶入一運算式以計算得出一第三比對結果，藉此可 解決先前技術用特徵點比對法進行身分驗證所造成特徵 點不足而無法通過身分驗證的缺點。 [0020] (3)此基於生物特徵之身分驗證裝置及其方法可藉由圖 樣比對的結果動態調整特徵點比對的門檻值，且由於上 述兩者呈負相關，藉此可解決使用者特徵點過少仍得以 099141103 表單編號A0101 第8頁/共31頁 0992071533-0 201222433 201222433 [0021] [0022] Ο 通過身分驗證的問題。 (4)此基於生物特徵之身分驗證裝置及其方法可藉由比 對-特定區域範圍内的圖樣，減少先前技術中圖樣比對 所造成資料量過大的問題，藉此提高比對的速度，再藉 由特徵點比對與圖樣比對達到身分驗證的精準度。稭 【實施方式】 請參閱第1圖，其係為本發明之基於生物特徵之身分驗證 裝置的第一實施例的方塊圖。如圖所示，基於生物特禮: 之身分驗證裝置1包含-储存單元1〇、一影像操取單元^ 、—生物特徵擷取單元12以及一處理單元]3，其中處理 單元13係可為中央處理器（Central Pr〇cessing••U fruit and the second comparison result calculate a third comparison result through an expression, and according to the third comparison result and a threshold value, determine whether the beam to be verified is 099141103 Form No. A0101 Page 5 / Total Page 31 0992071533-0 201222433 Passed identity verification. [0009] wherein the processing unit further obtains a second comparison result according to the similarity between the first biometric image and the second biometric image, and calculates a threshold value according to the second comparison result through an operation formula, and then According to the first comparison result and the threshold value, it is determined whether the target to be verified passes the identity verification, wherein the calculation formula makes the similarity corresponding to the second comparison result negatively correlated with the threshold value. [0010] wherein, the processing unit aligns the first biometric image and the second biometric image according to the alignment information, and the feature capturing unit further identifies one of the second biometric images according to the second feature point information. Point (S i ngu 1 ar Po int), the processing unit expands a specific area range around the singular point, and compares the first biometric image and the second creature in the specific area in a pattern comparison manner. Feature images, and a second alignment result is obtained. [0011] wherein, the processing unit aligns the first biometric image and the second biometric image according to the alignment information, and performs the pattern of the first biometric image and the second biometric image in a thumbnail manner. Compare and get a second comparison result. [0012] According to the object of the present invention, a biometric-based identity verification method is further proposed. The biometric-based identity verification method is applicable to a biometric-based identity verification device. The biometric-based identity verification device includes a storage unit, an image capture unit, a biometric capture unit, and a processing unit. The biometric-based identity verification method comprises the following steps: first storing the first biometric image and the corresponding first feature point information in a storage unit; and then using the image capturing unit to wait for a 099141103 form number A0101 page 6 of 31 Page 0992071533-0 201222433 verifying the target captures a second biometric image; and analyzing the plurality of feature points of the second biometric image by the biometric extraction unit and correspondingly providing a second feature point information; The unit compares the first feature point information with the second feature point information to obtain a first comparison result and a pair of bit information; and then the processing unit performs the first biometric image and the second creature according to the alignment information Aligning the feature image, performing a pattern comparison between the first biometric image and the second biometric image to obtain a second comparison result; and utilizing the processing unit according to the first alignment result and the second The comparison result determines whether the subject to be verified passes the identity verification. [0013] wherein the biometric-based identity verification method further comprises the following steps: calculating, by a processing unit, the first comparison result and the second comparison result by using an arithmetic expression to calculate a third comparison according to a preset weight ratio The result is then determined by the processing unit according to the third comparison result and a threshold value, whether the target to be verified passes the identity verification. [0014] wherein the biometric-based identity verification method further comprises the following steps: using the processing unit to obtain a second comparison result according to the similarity between the first biometric image and the second biometric circular image; And calculating, according to the second comparison result, a threshold value by an operation formula; and determining, by the processing unit, whether the target to be verified passes the identity verification according to the first comparison result and the threshold value. Wherein, the arithmetic expression makes the similarity corresponding to the second comparison result negatively correlated with the threshold value. [0015] wherein, in the step of performing pattern matching by the processing unit, the method further comprises the steps of: using the processing unit to align the first biometric image and the second biometric image according to the alignment information; and then using the feature capturing unit Identifying, according to the second feature point information, one of the second biometric images, the singular point 099141103, the form number Α0101, the seventh page, the total of 31 pages, 0992071533-0 201222433 (Singular Point); and using the processing unit to expand around the singular point A specific region range is obtained, and the first biometric image and the second biometric image in a specific region are compared in a pattern comparison manner to obtain a second comparison result. [0016] wherein, in the step of comparing the patterns by the processing unit, the method further comprises the steps of: using the processing unit to align the first biometric image and the second biometric image according to the alignment information; and shrinking through the processing unit The pattern of the first biometric image is compared with the pattern of the second biometric image to obtain a second alignment result. [0017] As described above, the biometric-based identity verification device and method thereof according to the present invention may have one or more of the following advantages: [0018] (1) the biometric-based identity verification device and The method can perform two image alignment by comparing the obtained alignment information by feature points. Since the information amount of the feature points is less than the information amount of the pattern, the feature points (mi nu ti ae) are compared with the pattern (pattern The alignment is fast, thereby solving the time-consuming problem caused by the prior art pattern-aligning two images. [0019] (2) The biometric-based identity verification device and the method thereof can calculate the first comparison result and the second comparison result into an arithmetic expression according to different weight ratios by the processing unit to calculate one The third comparison result can solve the disadvantage that the prior art feature point verification method for identity verification is insufficient to pass the identity verification. [0020] (3) The biometric-based identity verification device and the method thereof can dynamically adjust the threshold value of the feature point comparison by the result of the pattern comparison, and since the two are negatively correlated, the user can be solved The feature points are too small to be 099141103 Form No. A0101 Page 8 / Total 31 Page 0992071533-0 201222433 201222433 [0021] [0022] 问题 Problems by identity verification. (4) The biometric-based identity verification device and the method thereof can reduce the problem of excessive data volume caused by the pattern comparison in the prior art by comparing the patterns in the specific region, thereby increasing the speed of the comparison, and then The accuracy of identity verification is achieved by comparing the feature points with the pattern. [Embodiment] Please refer to Fig. 1, which is a block diagram of a first embodiment of a biometric-based identity verification device of the present invention. As shown in the figure, the biometric gift-based identity verification device 1 includes a storage unit 1 , an image manipulation unit ^, a biometric extraction unit 12, and a processing unit 3, wherein the processing unit 13 can be Central processor (Central Pr〇cessing
Unit，CPl〇 或微處理器（Micr〇_pr〇cessing ’並分別電性連接儲存單元、影像擁取單元u以及生 物特㈣取單幻2，儲存單元1G係可為各式儲存媒體如 Μ式記憶體、外接式記憶卡（體）、磁或光碟片 等’並儲存—第—生物特_像剛及對應之—第—特徵 點資訊UU，但所娜的第一生物特徵圖像及對應之第一 特徵點資㈣數量並不以此輕，也就是說，儲存單元 可儲存複數個第-生物特徵圖像及對應之複數個第一 特徵點資訊。 [0023] 099141103 打观迅标即4如.扣队、虹膜、人臉或手 、，文）進行身分驗證時，此基於生物特徵之身分驗證裝置！ 係先利用影_取單元n對待驗證標的2操取第二生物特 徵圖像⑴。接著，處理單元13控制生物特徵齡單元12 諸第二生物特徵圖像UG的複數個特徵點，藉此取得對 表單編號 A01O1 « n _ 第9頁/共31頁 0992071533-0 201222433 應第二生物特徵圖像11 0的第二特徵點資訊i 11。然後， 處理單元1 3以特徵點比對法（minut i ae matchi ng)比對 弟一特徵點資§fl 1 01與苐二特徵點資訊111，以得出一第 一比對結果130及一對位資訊131。接下來，處理單元13 根據對位資訊131進行第一生物特徵圖像1〇〇與第二生物 特徵圖像110的對位，當對位完成後，處理單元1 3以圖樣 比對法（pa11ern matching)比對第一生物特徵圖像1 〇〇 與第二生物特徵圖像110 ’以得出一第二比對結果132， 並根據第一比對結果130與第二比對結果132判斷待驗證 標的2是否通過身分驗證。 [0024] 承上所述’處理單元13係根據一預設權重比例，將第_ 比對結果13 0與弟一比對結果13 2帶入一運算式，經過計 算後得到第三比對結果133，又處理單元13係設定一門檻 值134 ’並根據第三比對結果133與門檻值134判斷待驗 證標的2是否通過身分驗證。 [0025] 其中，影像擷取單元11係可為各式可辦取影像之裝置， 例如：指紋感測器、虹膜感測器:、攝影機或相機等。另 外，處理單元13及生物特徵擷取單元12可為個別獨立的 運算晶片，也可以整合成同一運算晶片。 [0026] 請參閱第2圖，其係為本發明之基於生物特徵之身分驗證 裝置的第一實施例的示意圖。如圖所示’影像操取單元 係以一面積式指紋感測器30來實施，儲存單元' 生物特 徵擷取單元以及處理單元係以一主機32來實施。當使用 者以手指31按壓在面積式指紋感測器30時，首先面積式 指紋感測器30擷取手指31的指紋圖像33，並傳送指紋圖 099141103 表單編號A0101 第10頁/共31頁 0992071533-0 201222433 =33至主機32。接著，生物特徵擷取單元分析指紋圖像 之複數個特徵點，並對應提供第二特徵點資訊。 [0027] Ο [0029] PO 一 兀可儲存複數個第一指紋圖像以及各個第一指紋Unit, CPl〇 or microprocessor (Micr〇_pr〇cessing ' and electrically connected to the storage unit, image capturing unit u and bio-specific (four) take the single magic 2, the storage unit 1G can be a variety of storage media such as Μ Memory, external memory card (body), magnetic or optical disc, etc. 'and store - the first - biometric _ image just and corresponding - the first feature point information UU, but the first biometric image of The number of corresponding first feature points (4) is not so light, that is, the storage unit can store a plurality of first biometric images and corresponding plurality of first feature point information. [0023] 099141103 That is, 4, such as the buckle, iris, face or hand, text), the identity verification device based on biometrics when performing identity verification! First, the second biometric image (1) is processed by the image-taking unit n to be verified. Next, the processing unit 13 controls the plurality of feature points of the second biometric images UG of the biometric age unit 12, thereby obtaining the pair of forms No. A01O1 « n _ 9 pages / total 31 pages 0992071533-0 201222433 should be the second creature The second feature point information i 11 of the feature image 110. Then, the processing unit 13 compares the feature points §fl 1 01 and the second feature point information 111 with a feature point comparison method (minut i ae match), to obtain a first comparison result 130 and a Registration information 131. Next, the processing unit 13 performs the alignment of the first biometric image 1〇〇 and the second biometric image 110 according to the alignment information 131. When the alignment is completed, the processing unit 13 compares the patterns (pa11ern). Matching the first biometric image 1 〇〇 with the second biometric image 110 ′ to obtain a second comparison result 132 , and determining, according to the first comparison result 130 and the second comparison result 132 Verify that the target 2 is verified by identity. [0024] The processing unit 13 carries the first comparison result 13 0 and the brother-one comparison result 13 2 into an arithmetic expression according to a preset weight ratio, and obtains a third comparison result after calculation. 133. The processing unit 13 further sets a threshold value 134' and determines whether the target 2 to be verified passes the identity verification according to the third comparison result 133 and the threshold value 134. [0025] The image capturing unit 11 can be any device capable of capturing images, such as a fingerprint sensor, an iris sensor, a camera or a camera. In addition, the processing unit 13 and the biometrics capture unit 12 may be separate arithmetic wafers or integrated into the same operational wafer. Please refer to FIG. 2, which is a schematic diagram of a first embodiment of a biometric-based identity verification device of the present invention. As shown in the figure, the image manipulation unit is implemented by an area fingerprint sensor 30, and the storage unit 'biometric extraction unit and processing unit are implemented by a host 32. When the user presses the fingerprint sensor 30 with the finger 31, the area fingerprint sensor 30 first captures the fingerprint image 33 of the finger 31 and transmits the fingerprint image 099141103 Form No. A0101 Page 10 of 31 0992071533-0 201222433 =33 to host 32. Next, the biometric extraction unit analyzes the plurality of feature points of the fingerprint image and provides the second feature point information correspondingly. [0029] PO 兀 can store a plurality of first fingerprint images and respective first fingerprints
Ml徵點資訊’處理單^係進行第二特徵 f資訊與第-特徵點資訊的特徵點比對(minutiae tching)，並得出一第一比對結果及—對位資訊， 中對位資訊至少包含—旋轉角度及—位移量。處理單元 係:據第二特徵點資訊與第一特徵點資訊的相似度得出 第:比對結果，在本實施例中，第—崎結果係為一比 對分數，例如：她度達3_比對分數為⑽分。 :處理單第_比對結果後’難據紐點比對所 得到的對位資訊進行第—指紋圖像與指紋圖像33的對位 ’且由於特徵點資訊的資料量較圖像的f料量小，因此 以特徵點比對法對位圖像係比簡樣比對法對位圖像來 的决速如此-來，藉由本發明以特徵點比對所得到的 對位資訊進行圖像的對位，可減少進行圖樣比對前的圖 像對位時間。 其中’第—特徵點資訊及第二特徵點資訊至少包含各特 徵點之類型資訊、坐標資訊、角度資訊、曲率資訊或方 向資訊’類型資訊包含各個特徵點的端點（Endings)、 分岔點（BifUrcations)、核心（c〇re)或三角點 (Delta)。 [0030] 當處理單元完成第一指紋圖像與指 處理單元即根據第一指紋圖像與指 紋圖像33的對位後， 紋圖像33進行圖樣比 099141103 表單編號A0101 第Π頁/共31頁 0992071533-0 201222433 對（pattern matching)。值得一提的是，在進行圓樣 比對的過程中’為了減少全指紋的圖樣比對所造成的耗 時問題’本發明之發明人特職出由特_取單元根據 第二特徵點資訊辨識出第二生物特徵圖像之一奇異點 330(SingUlar P〇int)，亦即特徵擷取單元可根據核心 或三角點等特徵點的類型分析複數個特徵點而決定一奇 異點330，處理單元再以奇異點33〇為中心向四周展開一 特定區域範圍33卜並根據特定區域範圍331内之指紋圖 像33與第一指紋圖像進行圖樣比對，而得出一第二比對 結果，藉此可縮短圖樣比對的速度並可誠少所儲存之圖 像的資料量，又因該特定區域範圍内之指紋圖像的形變 機率小於全指紋圖像的形變機率，故可藉由比對特定區 域範圍内之指紋圖像減低全指紋圖像比對時，若有圖形 形變所造成的驗證問題。 [0031] [0032] [0033] 099141103 值得注意的是，雖然前述之特定區域範圍係以方形為例 說明本發明之概念，但熟悉此項技藝者當可輕易理解並 可延伸至圓形、橢圓形或多邊形，端看使用者之設定。 然後，處理單元可根據第一指紋圖像與指紋圖像33的相 似度得出第二比對結果，在本實施例中，第二比對結果 係為一比對分數，例如：相似度達8〇%則比對分數為8〇分 〇 接下來，本發明之發明人可依據經驗對應特徵點比對所 付到的第一比對結果給予—第一權重值，並對應圖樣比 對所得到的第二比對結果給予一第二權重值。在本實施 例中，處理單元設定第—權重值與第二權重值皆為〇 . 5， 表單編號A0101 第12頁/共31頁 201222433 但並不以此為限。此外，處理單元更設定一門檻值為45 分，並透過一運算式計算一第三比對結果，如下所示： [0034] 30 X 0,5 + 80 X = 55 [0035] 如上式可得出第三比對結果為55分，然後處理單元再判 斷第三比對結果是否大於Η權值，由於第二比對結果為 55分大於門檻值為45分，故處理單元判定此使用者通過 身分驗證。 _ [0036] 〇 ❹ 如上述實施例可檢視出在先前技術中以特徵點比對法所 無法通過身分驗證的赛驗證標的，可藉由本發明之基於 生物特徵之身分驗證裝置根據特徵點比對法及圖樣比對 法所得到的比對結果，給予預設權重比例，藉此提高身 份驗證的準確率，也可避免因待辨識標的使用習慣改變（ 例如但不限於：當使用者站立時所擷取到的指紋多為指 尖、當使用者坐著時所擷取到的指紋多為指腹、滑動時 速度過快或過慢、滑動時角度偏移荨）或環境改變（譬如 但不限於：溫涯度義.:;測器過髒..或老舊等），造成所掏取 的特徵點資訊不足而無法利用特徵點比對法進行身分驗 證的問題。 [0037] 請參閱第3圖，其係為本發明之基於生物特徵之身分驗證 裝置的第二實施例的方塊圖。圖中，其架構大致與第一 實施例雷同，兩者之差異在於’處理單元13係根據第二 比對結果132透過一運算式’經過計算以得出一門播值 134 ’故此門摇值134係隨者第一比對結果132而動態調 整’且此運算式係使第二比對結果13 2對應之相似度與門 099141103 表單編號A0101 第13頁/共31頁 0992071533-0 201222433 檻值134呈負相關，也就是說當相似度越高，則門檻值 134越低，反之當相似度越低，則門檻值134越高。接著 ，當處理單元13計算出門檻值134後，處理單元13即根據 第一比對結果130與門檻值134判斷待驗證標的2是否通過 身分驗證。 [0038] 請參閱第4圖，其係為本發明之基於生物特徵之身分驗證 裝置的第二實施例的示意圖。如圖所示，影像擷取單元 係以一掃掠式指紋感測器40來實施，但並不以此為限。 當使用者將手指31以滑動的方式滑過掃掠式指紋感測器 40時，掃掠式指紋感測器40即擷取複數個指紋圖像片段 ，並經過處理單元分析複數個指紋圖像片段之重合部份 ，並合成為完整的指紋圖像41。值得注意的是，本實施 例與第一實施例的差異在於，處理單元係以縮圖的方式 進行第一指紋圖像與指紋圖像41的圖樣比對，而得到第 二比對結果。其中縮圖的方式可包含縮小圖像的尺寸並 可搭配改變圖像的解析度等，且以縮圖的方式進行圖樣 比對的優點在於可節省儲存單元的儲存空間並可藉由縮 圖有效加快圖樣比對過程中位移及旋轉的速率。 [0039] 此外，處理單元係根據第一指紋圖像100與指紋圖像41的 相似度得出第二比對結果，在本實施例_，第二比對結 果係為一比對分數，假設比對分數為70分。接著，處理 單元將第二比對結果帶入一運算式計算一門檻值，得到 門檻值為21分，如下所示： _〇] 70X0,3=21 099141103 表單編號A0101 第14頁/共31頁 0992071533-0 201222433 [0041] 處理單元係根據第一比對結果與門檻值判斷使用者的指 紋是否得以通過身分驗證，亦即當第一比對結果超過門 檻值的21分，處理單元則判斷該使用者通過身分驗證， 反之，當第一比對結果小於門檻值的21分，處理單元則 判斷該使用者身分驗證失敗。 [0042] 承上所述，運算式係使第二比對結果對應之相似度與門 檻值呈負相關，如下表所示： 第二比對結果（比對分數） 乘（運算符號） 大於90 0.0 70-90 0.3 60-70 0. 35 50-60 0. A 40-50 0. 7 30-40 0. 9 20-30 1.2 低於20 1 [0044] 如表所示，當第二比對結果越高，經過運算式計算後， 所得出的門檻值則越低，亦即當以圖樣比對的相似度越 高，則可容許以特徵點比對的相似度較低，由於第二比 對結果與門檻值呈負相關，故處理單元可根據第二比對 結果動態調整門檻值，除了克服先前技術當以特徵點比 對時所擷取的特徵點數量不足而無法通過身分認證的困 擾，還可藉由以特徵點比對所得到的對位資訊克服以圖 像對位所造成的耗時問題，且更藉由縮圖的方式提高圖 099141103 表單編號A0101 第15頁/共31頁 0992071533-0 201222433 樣比對的速率。 [0045] 儘管前述在說明本發明之基於 ^ ^ , 物特徵之身分驗證裝置 的過耘中，亦已同時說明 3之基於生物特徵之身分 驗迅方法的概念，但為求清 圖詳細巧見’以下仍另繪示流程 [0046] [0047] 2⑷圖，其係為本發明之基於生物特徵之身分驗證 tr流程圖。如圖所示，本發明之基於生物特徵之身 t方法適用於—基於生物特徵之身分驗證裝置，其 以-儲存單元、—影像擷取單元、―生物特徵操取單 兀以及-處理單①’基於生物特徵之身分驗證方法包含 下列步驟： S10. Μ料單元儲存第—生物特徵圖像及對應之第一特 徵點資訊； [0048] [0049] su:利用影像擷取單元對—待臀證標的擷取第二生物特 徵圖像； S12 ·藉由生物特徵擷取單元分析第二生物特徵圖像之複 數個特徵點並對應提供第二特徵點資訊； [0050] S13 .透過處理單元比對第一特徵點資訊與第二特徵點資 訊，以得出第一比對結果及對位資訊； [0051] S14 ·以處理單元根據對位資訊進行第一生物特徵圖像與 第二生物特徵圖像的對位再進行兩者的圖樣比對，以得 出第二比對結果；以及 [0052] S15 :利用處理單元根據第一比對結果與第二比對結果判 099141103 表單編號A0101 第16頁/共31頁 0992071533-0 201222433 斷待驗證標的是否通過身分驗證。 [0053] 請參閱第6圖，其係為本發明之基於生物特徵之身分驗證 方法的另一流程圖。如圖所示，本發明之基於生物特徵 之身分驗證方法更包含下列步驟： [0054] S20 :根據一預設權重比例，利用處理單元將第一比對結 果與第二比對結果透過運算式計算第三比對結果；以及 [0055] S21 :藉由處理單元根據第三比對結果與門檻值，判斷待 驗證標的是否通過身分驗證。 ❹ [0056] 請參閱第7圖，其係為本發明之基於生物特徵之身分驗證 方法的再一流程圖。如圖所示，本發明之基於生物特徵 之身分驗證方法更包含下列步驟： [0057] S30 :利用處理單元根據第一生物特徵圖像與第二生物特 徵圖像的相似度得出第二比對結果； [0058] S31 :以處理單元根據第二比對結果透過一運算式計算一 門檻值，其中運算式係使第二比對結果對應之相似度與The Ml sign information 'processing unit' performs the feature point comparison (minutiae tching) of the second feature f information and the first feature point information, and obtains a first comparison result and - alignment information, the middle alignment information At least - the angle of rotation and the amount of displacement. Processing unit: according to the similarity between the second feature point information and the first feature point information, the first: comparison result, in the embodiment, the first-salt result is a comparison score, for example, she reaches 3 The _ comparison score is (10) points. : After processing the single _ aligning result, the 'difficulty data is compared with the obtained alignment information to perform the alignment of the first fingerprint image and the fingerprint image 33' and the data amount of the feature point information is compared with the image f The amount of material is small, so the feature point comparison method is compared with the simple image comparison method, and the alignment information obtained by the feature point comparison is plotted by the present invention. The alignment of the image reduces the image alignment time before the pattern is compared. The 'first feature point information and the second feature point information include at least type information, coordinate information, angle information, curvature information or direction information of each feature point. The type information includes end points of each feature point, and a branch point. (BifUrcations), core (c〇re) or triangle (Delta). [0030] When the processing unit completes the first fingerprint image and the finger processing unit, that is, according to the alignment of the first fingerprint image and the fingerprint image 33, the pattern image 33 is patterned to be 099141103. Form No. A0101 Page 3 of 31 Page 0992071533-0 201222433 pair (pattern matching). It is worth mentioning that in the process of round sample comparison, in order to reduce the time-consuming problem caused by the pattern matching of the full fingerprints, the inventor of the present invention specializes in the information of the second feature point according to the second feature point. Identifying a singular point 330 (SingUlar P〇int) of the second biometric image, that is, the feature extraction unit may determine a singular point 330 according to the type of feature points such as the core or the triangle point, and determine a singular point 330. The unit then expands a specific area range 33 around the singular point 33〇 and compares the fingerprint image 33 in the specific area range 331 with the first fingerprint image to obtain a second comparison result. Therefore, the speed of the pattern comparison can be shortened and the amount of data of the stored image can be reduced, and the deformation probability of the fingerprint image in the specific area is smaller than the deformation probability of the full fingerprint image, so When the fingerprint image in a specific area is reduced and the full fingerprint image is compared, if there is a verification problem caused by the graphic deformation. [0033] 099141103 It is to be noted that although the foregoing specific region ranges are illustrated by squares as an example, those skilled in the art can easily understand and extend to a circle or an ellipse. Shape or polygon, look at the user's settings. Then, the processing unit may obtain a second comparison result according to the similarity between the first fingerprint image and the fingerprint image 33. In this embodiment, the second comparison result is a comparison score, for example, the similarity is up to 8〇%, the comparison score is 8〇. Next, the inventor of the present invention can give the first weight value according to the first correspondence result of the empirical corresponding feature point comparison, and corresponding to the pattern comparison The resulting second alignment result is given a second weight value. In this embodiment, the processing unit sets the first weight value and the second weight value to be 〇. 5, Form No. A0101 Page 12 of 31 201222433 However, it is not limited thereto. In addition, the processing unit further sets a threshold value of 45 points, and calculates a third comparison result through an arithmetic expression, as follows: [0034] 30 X 0, 5 + 80 X = 55 [0035] The third comparison result is 55 points, and then the processing unit determines whether the third comparison result is greater than the weight value. Since the second comparison result is 55 points and the threshold value is greater than 45 points, the processing unit determines that the user passes the Identity verification. _ [0036] 〇❹ As in the above embodiment, the game verification target that cannot be verified by the feature point comparison method in the prior art can be inspected, and the biometric-based identity verification device of the present invention can be compared according to the feature points. The comparison result obtained by the method and the pattern comparison method is given a preset weight ratio, thereby improving the accuracy of the identity verification, and also avoiding the change of the usage habit of the object to be identified (for example, but not limited to: when the user stands The fingerprints obtained are mostly fingertips, the fingerprints captured by the user when sitting are mostly the fingertips, the speed is too fast or too slow when sliding, the angle is shifted when sliding, or the environment changes (for example, but not Limited to: Wen Ya Du Yi.:; The detector is too dirty.. or old, etc.), causing insufficient information on the feature points that are extracted and unable to use the feature point comparison method for identity verification. [0037] Please refer to FIG. 3, which is a block diagram of a second embodiment of the biometric-based identity verification device of the present invention. In the figure, the architecture is substantially the same as that of the first embodiment, and the difference between the two is that the processing unit 13 is calculated according to the second comparison result 132 through an arithmetic expression to obtain a homing value 134 ' The system is dynamically adjusted by the first comparison result 132 and the expression is such that the second comparison result 13 2 corresponds to the similarity and the gate 099141103 Form No. A0101 Page 13 / Total 31 Page 0992071533-0 201222433 Devaluation 134 Negative correlation, that is to say, the higher the similarity, the lower the threshold value 134, and vice versa, the lower the similarity, the higher the threshold value 134. Next, when the processing unit 13 calculates the threshold value 134, the processing unit 13 determines whether the target 2 to be verified passes the identity verification based on the first comparison result 130 and the threshold value 134. [0038] Please refer to FIG. 4, which is a schematic diagram of a second embodiment of the biometric-based identity verification device of the present invention. As shown in the figure, the image capturing unit is implemented by a swept fingerprint sensor 40, but is not limited thereto. When the user slides the finger 31 over the swept fingerprint sensor 40 in a sliding manner, the swept fingerprint sensor 40 captures a plurality of fingerprint image segments and analyzes the plurality of fingerprint images through the processing unit. The overlapping portions of the segments are combined into a complete fingerprint image 41. It should be noted that the difference between this embodiment and the first embodiment is that the processing unit compares the first fingerprint image with the fingerprint image 41 in a thumbnail manner to obtain a second comparison result. The method of thumbnailing may include reducing the size of the image and changing the resolution of the image, etc., and the advantage of pattern comparison in the form of thumbnails is that the storage space of the storage unit can be saved and can be effectively reduced by the thumbnail. Speed up the rate of displacement and rotation during pattern alignment. [0039] In addition, the processing unit obtains a second comparison result according to the similarity between the first fingerprint image 100 and the fingerprint image 41. In the embodiment _, the second comparison result is a comparison score, assuming The comparison score is 70 points. Then, the processing unit brings the second comparison result into an arithmetic expression to calculate a threshold value, and obtains a threshold value of 21 points, as follows: _〇] 70X0, 3=21 099141103 Form No. A0101 Page 14 of 31 0992071533-0 201222433 [0041] The processing unit determines whether the fingerprint of the user is verified by the identity according to the first comparison result and the threshold value, that is, when the first comparison result exceeds 21 points of the threshold value, the processing unit determines the The user verifies by identity, and conversely, when the first comparison result is less than 21 points of the threshold value, the processing unit determines that the user identity verification fails. [0042] As described above, the arithmetic expression is such that the similarity corresponding to the second comparison result is negatively correlated with the threshold value, as shown in the following table: Second alignment result (alignment score) Multiplication (operation symbol) is greater than 90 0.0 70-90 0.3 60-70 0. 35 50-60 0. A 40-50 0. 7 30-40 0. 9 20-30 1.2 Below 20 1 [0044] As shown in the table, when the second comparison The higher the result, the lower the threshold value obtained after the calculation is calculated, that is, the higher the similarity of the pattern comparison, the lower the similarity of the feature point alignment is allowed, because the second ratio The result is negatively correlated with the threshold value, so the processing unit can dynamically adjust the threshold according to the second comparison result, in addition to overcoming the problem that the prior art fails to pass the identity authentication when the number of feature points acquired by the feature point comparison is insufficient. It is also possible to overcome the time-consuming problem caused by image alignment by comparing the alignment information obtained by feature point comparison, and further improve the figure by means of thumbnails. 099141103 Form No. A0101 Page 15 of 31 0992071533-0 201222433 Rate of sample alignment. [0045] Although the foregoing description of the identity verification device based on the feature of the present invention has also described the concept of the biometric-based identity verification method, the detailed description of the method is clear. The following is still a flow chart [0046] [0047] 2 (4), which is a flow chart of the biometric-based identity verification tr of the present invention. As shown in the figure, the biometric-based body t method of the present invention is applicable to a biometric-based identity verification device, which uses a storage unit, an image capture unit, a biometric operation unit, and a processing unit 1 The biometric-based identity verification method comprises the following steps: S10. The data unit stores the first biometric image and the corresponding first feature point information; [0049] [0049] su: using the image capturing unit pair to wait for the hip The second biometric image is captured by the token; S12: analyzing the plurality of feature points of the second biometric image by the biometric extraction unit and correspondingly providing the second feature point information; [0050] S13. Transmitting the processing unit ratio Determining the first feature point information and the second feature point information to obtain the first comparison result and the alignment information; [0051] S14: performing, by the processing unit, the first biometric image and the second biometric feature according to the alignment information The alignment of the image is further compared with the pattern of the two to obtain a second comparison result; and [0052] S15: using the processing unit to judge the result of the first comparison and the second comparison result by 099141103, the form number A01 01 Page 16 of 31 0992071533-0 201222433 Whether the verification target has passed the identity verification. [0053] Please refer to FIG. 6, which is another flow chart of the biometric-based identity verification method of the present invention. As shown in the figure, the biometric-based identity verification method of the present invention further includes the following steps: [0054] S20: using a processing unit to pass the first comparison result and the second comparison result through the operation formula according to a preset weight ratio Calculating a third comparison result; and [0055] S21: determining, by the processing unit, whether the target to be verified passes the identity verification according to the third comparison result and the threshold value. [0056] Please refer to FIG. 7, which is still another flowchart of the biometric-based identity verification method of the present invention. As shown in the figure, the biometric-based identity verification method of the present invention further includes the following steps: [0057] S30: using the processing unit to obtain a second ratio according to the similarity between the first biometric image and the second biometric image [0058] S31: calculating, by the processing unit, a threshold value according to the second alignment result through an operation formula, wherein the operation formula is configured to make the second comparison result correspond to the similarity degree
G 門檻值呈負相關；以及 [0059] S32 :藉由處理單元根據第一比對結果與門檻值，判斷待 驗證標的是否通過身分驗證。 [0060] 請參閱第8圖，其係為本發明之基於生物特徵之身分驗證 方法之比對特定區域範圍的圖像之流程圖。如圖所示， 本發明之基於生物特徵之身分驗證方法於處理單元進行 圖樣比對之步驟中，更包含下列步驟： [0061] S40 :利用處理單元根據對位資訊對位第一生物特徵圖像 099141103 表單編號A0101 第17頁/共31頁 0992071533-0 201222433 與第二生物特徵圖像； [0062] S41 :以特徵擷取單元根據第二特徵點資訊辨識出第二生 物特徵圖像之一奇異點（Singular Point);以及 [0063] S42 :利用處理單元以奇異點為中心向四周展開一特定區 域範圍，並以圖樣比對的方式比對特定區域範圍内之第 一生物特徵圖像與第二生物特徵圖像，而得到第二比對 結果。 [0064] 請參閱第9圖，其係為本發明之基於生物特徵之身分驗證 方法之以縮圖方式進行圖樣比對的流程圖。如圖所示， 本發明之基於生物特徵之身分驗證方法於處理單元進行 圖樣比對之步驟中，更包含下列步驟： [0065] S50 :利用處理單元根據對位資訊對位第一生物特徵圖像 與第二生物特徵圖像；以及 [0066] S51 :以處理單元藉由縮圖的方式進行第一生物特徵圖像 與第二生物特徵圖像的圖樣比對，而得到第二比對結果 〇 [0067] 上述的各步驟之詳細實施方式已於前面的基於生物特徵 之身分驗證裝置的敘述中詳細說明過了，在此便不再贅 述。 [0068] 綜上所述，本發明所提出之基於生物特徵之身分驗證裝 置及其方法可藉由比對資料量較小的特徵點資訊以得出 對位資訊，並根據此對位資訊進行圖像的對位，藉此解 決了先前技術中兩圖像對位所造成的耗時問題；又無論 099141103 表單編號ΑΟΙΟΙ 第18頁/共31頁 0992071533-0 201222433 tb對特定區域範圍内的圖像或以縮圖的方式進行圖樣比 對皆可提高比對的速率，減低系統運算的資源消耗， t達到節省儲存空間的目的。 [0069] Ο [0070] 此*外’根據以圖樣比對法所得出的比對結果調整門檻值 可避免單—比對法所造成身份無法辨識的問題，尤其 對於使用者習慣改變或環境改變，造成所擷取出的特徵 點過少而難以驗證等問題，更可透過本發明之基於生物 特徵之身分驗證裝置及其方法順利地通過身分驗證，且 了達到極高的準確率。 .' - ' : Ο 值得一提的是，本發明之發明人亦、關注身分驗證的安全 ^生與便利性問題^ _般而言，熟悉此項技藝者係以錯誤 拒絕率（FRR，False Rejection Rate)及錯誤接收率 (FM，False Acceptance Rate )作‘辨識系統的辨 識能力指標。錯誤拒絕率（FRR)代表應該辨識通過的卻沒 通過’另一種錯誤接收率（FAR)代表應該不能通過辨識的 部通過了。習知技術係僅用單一比對法及單一門檻值作 為身分通過與否的標準，有時為了考慮使用者的方便性 ’便降低門檻值，雖能讓使用者較易通過驗證，卻也增 加錯誤接收率（FAR)，讓不該通過的也通過，反而造成安 全上的漏洞；另一方面，對於高度管制的區域，為考慮 其安全性，便提高門檻值，雖能顧及安全性，卻容易讓 應該允許通過的使用者有較高無法通過驗證的機會，即 較高的錯誤拒絕率（FRR)，反而造成使用者的不便，由此 可知’習知技術難以兼顧安全性與便利性。 [0071] 099141103 然而，根據本發明所提出之基於生物特徵之身分驗證裝 表單編號A0101 第19頁/共31頁 0992071533-0 201222433 置及其方法可藉由圖樣比對法所得出的比對結果調整門 檻值，當圖樣比對法所得出的比對結果高，則降低特徵 點比對法的門檻值；反之，則提高特徵點比對法的門檻 值，並根據經驗法則動態調整該門檻值，使能兼顧錯誤 拒絕率（FRR)與錯誤接收率（FAR)。如此一來，本發明之 基於生物特徵之身分驗證裝置及其方法不僅可提高身份 驗證的安全性，更可兼顧使用者的使用便利性，克服了 先前技術僅能顧及錯誤拒絕率（FRR)或錯誤接收率（FAR) 的其中之一，進而達到同時改善錯誤拒絕率（FRR)及錯誤 接收率（FAR)的目的。 [0072] 以上所述僅為舉例性，而非為限制性者。任何未脫離本 發明之精神與範疇，而對其進行之等效修改或變更，均 應包含於後附之申請專利範圍中。 【圖式簡單說明】 [0073] 第1圖係為本發明之基於生物特徵之身分驗證裝置的第一 實施例的方塊圖。 第2圖係為本發明之基於生物特徵之身分驗證裝置的第一 實施例的示意圖。 第3圖係為本發明之基於生物特徵之身分驗證裝置的第二 實施例的方塊圖。 第4圖係為本發明之基於生物特徵之身分驗證裝置的第二 實施例的示意圖。 第5圖係為本發明之基於生物特徵之身分驗證方法的流程 圖。 第6圖係為本發明之基於生物特徵之身分驗證方法的另一 099141103 表單編號A0101 第20頁/共31頁 0992071533-0 201222433 流程圖。 第7圖係為本發明之基於生物特徵之身分驗證方法的再一 流程圖。 第8圖係為本發明之基於生物特徵之身分驗證方法之比對 特定區域範圍的圖像之流程圖。 第9圖係為本發明之基於生物特徵之身分驗證方法之以縮 圖方式進行圖樣比對的流程圖。 【主要元件符號說明】 [0074] ❹ Ο 099141103 I :基於生物特徵之身分驗證裝置 10 :儲存單元 100 :第一生物特徵圖像 101 :第一特徵點資訊 II :影像擷取單元 110 :第二生物特徵圖像 111 :第二特徵點資訊 12 :生物特徵擷取單元 13 :處理單元 130 :第一比對結果 131 :對位資訊 132 :第二比對結果 133 :第三比對結果 134 :門檻值 2:待驗證標的 30 :面積式指紋感測器 31 :手指 32 :主機 表單編號Α0101 第21頁/共31頁 0992071533-0 201222433 33 :指紋圖像 330 :奇異點 331 :特定區域範圍 40 :掃掠式指紋感測器 41 :指紋圖像 S10-S15 > S20-S21 ' S30-S32 ' S40-S42 ' S50-S5 ： 步驟 099141103 表單編號A0101 第22頁/共31頁 0992071533-0The G threshold is negatively correlated; and [0059] S32: determining, by the processing unit, whether the object to be verified passes the identity verification according to the first comparison result and the threshold value. [0060] Please refer to FIG. 8 , which is a flow chart of the biometric-based identity verification method of the present invention for comparing images of a specific region range. As shown in the figure, the biometric-based identity verification method of the present invention includes the following steps in the step of pattern matching by the processing unit: [0061] S40: using the processing unit to align the first biometric map according to the alignment information. Like 099141103 Form No. A0101 Page 17 of 31 0992071533-0 201222433 and the second biometric image; [0062] S41: the feature capturing unit identifies one of the second biometric images according to the second feature point information Singular Point; and [0063] S42: using a processing unit to expand a specific area range around the singular point, and compare the first biometric image in a specific area by pattern matching The second biometric image is obtained to obtain a second alignment result. [0064] Please refer to FIG. 9 , which is a flow chart of the biometric-based identity verification method of the present invention in a thumbnail manner for pattern comparison. As shown in the figure, the biometric-based identity verification method of the present invention includes the following steps in the step of pattern matching by the processing unit: [0065] S50: using the processing unit to align the first biometric map according to the alignment information. Image and second biometric image; and [0066] S51: performing a second alignment result by comparing the pattern of the first biometric image with the second biometric image by means of a thumbnail of the processing unit详细[0067] The detailed implementation of each of the above steps has been described in detail in the foregoing description of the biometric-based identity verification device, and will not be described herein. [0068] In summary, the biometric-based identity verification device and the method thereof according to the present invention can obtain alignment information by comparing feature point information with a small amount of data, and perform mapping according to the alignment information. The alignment of the image, thereby solving the time-consuming problem caused by the alignment of the two images in the prior art; and regardless of the 099141103 form number ΑΟΙΟΙ page 18 / total 31 page 0992071533-0 201222433 tb image for a specific area Or the pattern comparison in the form of thumbnails can increase the rate of comparison, reduce the resource consumption of the system operation, and achieve the purpose of saving storage space. [0070] [0070] This * outside 'adjusting the threshold according to the comparison result obtained by the pattern comparison method can avoid the problem that the identity caused by the single-alignment method is unrecognizable, especially for the user's habit change or environmental change. The problem that the extracted feature points are too small and difficult to verify, and the biometric-based identity verification device and the method thereof of the present invention can pass the identity verification smoothly and achieve extremely high accuracy. .' - ' : Ο It is worth mentioning that the inventor of the present invention also pays attention to the security and convenience of identity verification. ^ Generally speaking, those who are familiar with the technology are rejecting the error rate (FRR, False). Rejection Rate) and False Acceptance Rate (FM) are used to identify the identification ability of the system. The false rejection rate (FRR) represents the passage that should be identified but not passed. Another type of false acceptance rate (FAR) indicates that the department should not pass the identification. The prior art technique uses only a single comparison method and a single threshold as a criterion for passing or not, and sometimes it is to reduce the threshold value in consideration of user convenience, although it allows the user to pass the verification more easily, but also increases The false acceptance rate (FAR), which should not be passed, causes security loopholes. On the other hand, for highly regulated areas, in order to consider the security, the threshold is increased, although security can be considered. It is easy for the user who should be allowed to pass to have a higher chance of failing to pass the verification, that is, a higher error rejection rate (FRR), which in turn causes inconvenience to the user, and thus it is known that the conventional technology is difficult to balance security and convenience. [0071] 099141103 However, according to the present invention, the biometric-based identity verification package form number A0101, page 19/total 31 page 0992071533-0 201222433, and the method thereof can be compared by the pattern comparison method. Adjusting the threshold value, when the comparison result obtained by the pattern comparison method is high, the threshold value of the feature point comparison method is lowered; otherwise, the threshold value of the feature point comparison method is increased, and the threshold value is dynamically adjusted according to the rule of thumb. Enables both the error rejection rate (FRR) and the error acceptance rate (FAR). In this way, the biometric-based identity verification device and the method thereof of the present invention can not only improve the security of identity verification, but also take into account the user's convenience of use, and overcome the prior art to only consider the false rejection rate (FRR) or One of the false acceptance rates (FAR), which in turn achieves both the error rejection rate (FRR) and the false reception rate (FAR). The above description is by way of example only and not as a limitation. Any equivalent modifications or alterations to the spirit and scope of the invention are intended to be included in the scope of the appended claims. BRIEF DESCRIPTION OF THE DRAWINGS [0073] Fig. 1 is a block diagram showing a first embodiment of the biometric-based identity verification apparatus of the present invention. Fig. 2 is a schematic view showing a first embodiment of the biometric-based identity verification apparatus of the present invention. Figure 3 is a block diagram showing a second embodiment of the biometric-based identity verification device of the present invention. Fig. 4 is a schematic view showing a second embodiment of the biometric-based identity verification device of the present invention. Figure 5 is a flow chart of the biometric-based identity verification method of the present invention. Figure 6 is another example of the biometric-based identity verification method of the present invention. 099141103 Form No. A0101 Page 20 of 31 0992071533-0 201222433 Flowchart. Fig. 7 is still another flow chart of the biometric-based identity verification method of the present invention. Figure 8 is a flow chart showing the comparison of images of a specific area based on the biometric-based identity verification method of the present invention. Fig. 9 is a flow chart showing a pattern comparison in a thumbnail manner based on the biometric-based identity verification method of the present invention. [Main component symbol description] [0074] 99 Ο 099141103 I : Biometric-based identity verification device 10: storage unit 100: first biometric image 101: first feature point information II: image capturing unit 110: second Biometric image 111: second feature point information 12: biometric extraction unit 13: processing unit 130: first alignment result 131: alignment information 132: second alignment result 133: third alignment result 134: Threshold value 2: to be verified target 30: area fingerprint sensor 31: finger 32: host form number Α 0101 page 21 / total 31 page 0992071533-0 201222433 33 : fingerprint image 330 : singular point 331 : specific area range 40 : Sweeping Fingerprint Sensor 41: Fingerprint Image S10-S15 > S20-S21 ' S30-S32 ' S40-S42 ' S50-S5 : Step 099141103 Form No. A0101 Page 22 of 31 0992071533-0