480454 A7 B7 五、發明說明（/) . 詳細說明 本發明關於一種高明度的立體式指紋讀取系統。 經濟部智慧財產局員工消費合作社印製 指紋是一個人最好的身份證明，每個人的指紋各不相 同，且與生倶來，一生都不會改變，且隨身攜帶，不會遺 失，且很難僞造。檢警確認不明人員身份，主要即利用指 紋。j最機密的機構，其門口的通行處的身份檢驗，也多有 用指紋辨識者。而觀諸傳統各種身份辨識的方式，都沒有 像指紋如此便利可靠者。***容易僞刻，簽字可模仿，且 較不容易用機器來辨認，通關密語與密碼容易忘記，晶片 卡或磁卡會忘了帶或遺失或遭人盜用…。可以預見，未來 人們追求更大的便利與舒適，指紋辨識必然被廣泛使用， 以取代或輔助一般常用的身份辨識手段。例如：在銀行提 款，忘了帶印鑑或金融卡，也可在櫃枱讓指紋辨識機讀取 自己指紋，與銀行建檔的客戶指紋比對，即可提款。或在 網路交易場合中，利用指紋讀取機讀取指紋，經由網路傳 送至交易主辦單位確認身份以確保交易的安全。上述的二 個例子，第一種情形是封閉式的，第二種情形是開放式的 。封閉式的情形，指紋辨識機除了前級的指紋讀取外，更 包含後級的記憶體（指紋庫）與處理單元，把讀入的指紋與 指紋庫中建檔的指紋比對。開放式的情形，只需一個指紋 讀取機，它經網路與主網站連接，指紋庫與處理單元只在 主網站有。這兩種情形，用途都很廣，尤其今日網路與電 子商務(E-Commerce)蓬勃發展，更使指紋辨識的商機潛力 無窮。 4 本紙張尺玉適用中國國家標準〇：NS)A4規格（210 X 297公釐） ' 經濟部智慧財產局員工消費合作社印製 480454 A7 _ B7 五、發明說明（>) 尤其是近年來，影像處理技術以及「面影像感測器」 (area image sensor)有長足進步，可處理很複雜的影像。而 指紋的辨識，在今日進步的影像處理技術能力所能做的事 情中，算是非常簡單的一種，因此可利用低廉的成本將它 普遍化。 習知的指紋辨識裝置，大致上原理如第十圖所示，爲 了使具有指紋的指面壓平成一平面，使指紋均勻地在一平 面上被讀取，因此，指紋讀取機(L)有一水平的透明放置板 (ΑΡ)，使指面平貼在其上。〔註：在實際上的指紋讀取機 ，光源都從放置板(ΑΡ)四周用同一種光均勻照明，但此處 爲了便利說明，只用一邊及兩邊照明爲例說明。先說明只 用一邊照明時的效果。〕如圖所示，有一光源(Β)從手指側 面斜斜向上使光線沿著平行ΥΖ面的方向照到放置板(ΑΡ)上 的指面。被照明後，指面上的指紋影像向下經一透鏡系統 (LS)成像到下方的面影像感測器(IS)(例如一個CCD影像感 測器或CMOS影像感測器），該面影像感測器(IS)上有許多 矩陣式的圖元小格(pixel)，各小格依本身收到的光強度而 產生一對應的電壓信號，然後經一道習知的「二元化轉換 」（亦即把電壓高於一個臨限値的圖元全當作「1」，而低 於臨限値者全當作「0」，例如利用一比較器)把這個指紋 變成一個由0與1組成的矩陣圖形’並送到一後段的影像 比較級並與儲存在指紋庫(FB)中的一個指紋樣品(FM)(亦呈 〇與1的矩陣圖形方式儲存)比較。如果讀入的指紋和儲存 之指紋樣品符合，則影像比較級送出一確認信號’起動後 5 本紙張尺度適用中國國家標準（CNS)A4規格（210 X 297公釐） (請先閱讀背面之注意事項再本頁) . •線. 480454 A7 B7 經濟部智慧財產局員工消費合作社印製 五、發明說明（巧） . 方的最終控元件(例如門閂的電磁閥），否則就送出一錯誤 信號，引發相關的防衛措施（例如將系統封鎖、發出警報) 〇 但是這種方式的指紋讀取有一些缺點。茲再配合圖式 詳細說明。請看第十一圖。指紋係由***的稜脊(R)與凹的 山谷$T)構成。理論上，稜脊(例如yi)明亮，山谷(例如y3) 黑暗。而且我們希望明亮黑暗的區域之間的過渡區儘可能 窄，如此，亮與暗之間有明顯的界限，所得的指紋影像才 會線條淸晰不模糊。但實際上，如第十一圖及十二圖A所 示，在山腹(例如y2，y4)區是不大亮也不大暗的灰色地帶， 模擬兩可。因此當第十二圖Y方向亮度分佈的類比曲線一 旦做了「二元化轉換」，則這種灰色地帶會因所選的臨限 値VS1、VS2大小不同，而在結果上有很大差異。第十二^圖 B爲用電壓乂^爲臨限値的二元化曲線，其中山腹y2，y4被 當作0，而第十二圖C用VS2作臨限値，y2，y4被當作1。 這表示，這種傳統指紋讀出機的臨限値容許誤差(tolerance) 很小。因此在上述開放式的網路連線情形，不大適用。尤 .其是在信號壓縮時，上述的模擬兩可的點y2，y4會造成輸 出之圖元變動及不穩，使壓縮效率變差。而且因爲讀取時 的誤差甚大，常需重複確認，會大量佔用網路連線時間。 只能用在同一部指紋讀取機(亦即上述封閉式的獨立系統情 形）。而且最初指紋建檔時，所用之讀取機與後來的讀取機 有差異，也會造成問題。但即使用同一部指紋讀取機，並 不能對所有人都適用，例如，有色人種的手指反射光一般 6 張尺度適用中國國家標準（CNS)A4規格（210 X 297公釐） (請先閱讀背面之注意事項再480454 A7 B7 V. Description of the invention (/). Detailed description The present invention relates to a high-luminance three-dimensional fingerprint reading system. The fingerprint printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs is the best proof of identity for a person. Each person's fingerprint is different and will not change during life. counterfeit. Prosecutors confirmed the identity of unknown persons, mainly using fingerprints. Most of the most confidential institutions, the identification of the pass at the doorway also uses fingerprint identification. And the traditional methods of identification are not as convenient and reliable as fingerprints. The seal is easy to be engraved, the signature can be imitated, and it is less easy to identify with the machine. The passphrase and password are easy to forget. The chip card or magnetic card will be forgotten or lost or stolen. It is foreseeable that in the future, people will pursue greater convenience and comfort, and fingerprint identification will be widely used to replace or assist the commonly used identification methods. For example, when withdrawing money at a bank, forgetting to bring a seal or a financial card, you can also allow the fingerprint reader to read your fingerprint at the counter, and compare it with the customer's fingerprint filed by the bank to withdraw money. Or in the case of online transactions, use a fingerprint reader to read the fingerprint and send it to the transaction organizer via the network to confirm the identity to ensure the security of the transaction. In the above two examples, the first case is closed and the second case is open. In the closed case, in addition to the fingerprint reading of the previous stage, the fingerprint recognition machine also includes the memory (fingerprint database) and the processing unit of the subsequent stage, and compares the read fingerprint with the fingerprint created in the fingerprint database. In the open situation, only a fingerprint reader is needed, which is connected to the main website via the network, and the fingerprint database and processing unit are only available on the main website. In both cases, the uses are very wide, especially the booming development of the Internet and e-commerce today, and the potential of fingerprint identification is endless. 4 This paper rule jade applies Chinese national standard 0: NS) A4 specification (210 X 297 mm) '' Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 480454 A7 _ B7 V. Description of the invention (>) Especially in recent years, The image processing technology and "area image sensor" have made great progress and can process very complex images. Fingerprint recognition is a very simple thing that can be done by today's advanced image processing technology capabilities, so it can be universalized at a low cost. The conventional fingerprint recognition device is roughly as shown in the tenth figure. In order to flatten the finger surface with a fingerprint so that the fingerprint is evenly read on a plane, the fingerprint reader (L) There is a horizontal transparent placement plate (AP) with the finger surface flat against it. [Note: In actual fingerprint readers, the light source is uniformly illuminated with the same light from around the placement plate (AP). However, for convenience, only one side and two sides are used as an example. First, the effect when only one side is illuminated will be explained. ] As shown in the figure, there is a light source (B) obliquely upward from the side of the finger so that the light shines on the finger surface on the placement plate (AP) in a direction parallel to the ΥZ plane. After being illuminated, the fingerprint image on the finger surface is imaged downward by a lens system (LS) to the lower surface image sensor (IS) (such as a CCD image sensor or CMOS image sensor). The sensor (IS) has many matrix-like pixel cells (pixels). Each cell generates a corresponding voltage signal according to the light intensity received by it, and then undergoes a conventional "binary conversion" (That is, all the pixels whose voltage is higher than a threshold value are regarded as “1”, and those lower than the threshold value are regarded as “0”, for example, a comparator is used) to change this fingerprint into a 0 and 1 The composed matrix pattern is sent to a subsequent image comparison stage and compared with a fingerprint sample (FM) stored in the fingerprint database (FB) (also stored as a matrix pattern of 0 and 1). If the read fingerprint and the stored fingerprint sample match, the image comparison level sends a confirmation signal 'After starting 5 paper sizes apply to Chinese National Standard (CNS) A4 specifications (210 X 297 mm) (Please read the notes on the back first Matters are on this page). • Line. 480454 A7 B7 Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs 5. Description of the Invention (Clever). The final control element (such as the solenoid valve of the door latch), otherwise it will send an error signal, Causes related defense measures (such as blocking the system, issuing an alarm). However, this method of fingerprint reading has some disadvantages. It is described in detail in conjunction with the drawings. Please see the eleventh figure. The fingerprint consists of a raised ridge (R) and a concave valley $ T). In theory, the ridges (e.g. yi) are bright and the valleys (e.g. y3) are dark. Moreover, we hope that the transition area between bright and dark areas is as narrow as possible. In this way, there is a clear boundary between light and dark so that the fingerprint image obtained will have clear lines and no blur. But in fact, as shown in Figures 11 and 12A, in the mountainside (for example, y2, y4), the gray areas are not too bright or dark, which is a ambiguity. Therefore, once the analog curve of the luminance distribution in the Y direction in Figure 12 is "binary transformed", the gray area will be very different in the results due to the different thresholds VS1 and VS2 selected. . The twelfth figure B is a binary curve using the voltage 临 ^ as the threshold 値, in which the mountain belly y2, y4 is regarded as 0, and the twelfth figure C uses VS2 as the threshold 値, and y2, y4 is regarded as 1. This means that the threshold tolerance of such a conventional fingerprint reader is small. Therefore, the above-mentioned open network connection situation is not applicable. In particular, when the signal is compressed, the above-mentioned points y2 and y4 of the analog ambiguity will cause the output picture element to change and become unstable, making the compression efficiency worse. And because the error in reading is very large, repeated confirmation is often required, which will take up a lot of network connection time. It can only be used on the same fingerprint reader (that is, in the case of the closed independent system mentioned above). In addition, when the fingerprint is first created, the reader used is different from the later readers, which also causes problems. But even if you use the same fingerprint reader, it is not applicable to everyone. For example, the reflection of light from fingers of colored people is generally 6 standards applicable to China National Standard (CNS) A4 (210 X 297 mm) (please first Read the notes on the back again
· --線- 經濟部智慧財產局員工消費合作社印製 480454 A7 B7 五、發明說明（p) . 比白人少得多，因此可能出現有色人種指紋全是0，或白 人指紋全是1的無意義的矩陣圖形。理想的情形，是要明 暗界限分明，過渡區儘量窄，曲線坡度儘量陡、亦即光度 對Y分佈的圖形儘量接近矩形波狀，如此不論臨限値高一 些或低一些，二元化後的圖形部差不多，亦即，臨限値要 有很$的容許誤差，如此就可適用在開放式的網路中。 由第十二圖A可看出，曲線在山谷左邊(y2處)較陡， 山谷右邊(y4處)較緩，這表示，裝在左邊的光源對山谷左邊 的山腹的檢出能力比對山谷右邊山腹的檢出能力強，也就 是山谷左邊能得到明顯的界限，右邊則否。由以上的簡單 分析，我們由此可知道在單邊的照明的單獨效果爲背光坡 線條淸楚，向光坡線條模糊，但在實際的指紋讀取機爲了 得到整體的印象，乃是由四邊或至少兩邊同時照明的，因 此其實際結果乃是左右二邊照明相加的效果。如第十三圖 所示，在右邊也有一個光源(B2)，雖然它對山谷右邊山腹 (y4處)的死角有較強的檢出能力，但左右同時照明，則全部 都是「向光坡」，因此背光坡的明顯界限就完全失去了。 請看第十三圖，如在左、右各裝一光源(B1)(B2)，則其光度 與Y的關係如第十四圖的實線K1(B1光源)虛線K2(B2光 源)所示。但二者同時照明，則二曲線相加，得到如第十四 圖B的鋸齒曲線，不但兩邊緣都不明確，山峰、山谷的落 差減少了，全暗區沒有了，因此明暗對比較不淸晰，因此 成像後的信號幅度變化會較小。使得指紋檢出的困難度增 加，且由於峰谷落差的「振幅」變小，故臨限値的容許誤 7· Line-Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 480454 A7 B7 V. Description of the invention (p). It is much less than white people, so it may appear that colored people's fingerprints are all 0, or white people's fingerprints are all 1 Meaningless matrix graphics. Ideally, the boundary between light and dark should be clear, the transition area should be as narrow as possible, and the slope of the curve should be as steep as possible, that is, the graph of light distribution versus Y should be as close to a rectangular wave shape as possible, so whether the threshold is higher or lower, the binary The graphics department is similar, that is, the threshold must have a very large tolerance, so it can be used in open networks. It can be seen from the twelfth figure A that the curve is steeper on the left side (y2) of the valley and slower on the right side (y4) of the valley, which indicates that the light source installed on the left has a better detection ability on the left side of the valley than on the valley. The detection ability on the right side of the mountain is strong, that is, the left side of the valley can get a clear boundary, and the right side is not. From the above simple analysis, we can know from this that the single effect of lighting on one side is the backlit slope lines, and the lines toward the light slope are blurred. However, in order to obtain the overall impression, the actual fingerprint reader is composed of four sides. Or at least two sides are illuminated at the same time, so the actual result is the effect of adding the left and right sides of the lighting. As shown in the thirteenth figure, there is also a light source (B2) on the right side. Although it has a strong detection ability for the dead corner of the mountainside (y4) on the right side of the valley, all the lights are left and right at the same time. "Therefore, the obvious boundary of the backlight slope is completely lost. Please see the thirteenth figure. If a light source (B1) (B2) is installed on the left and right, the relationship between the photometric and Y is as shown by the solid line K1 (B1 light source) and the dotted line K2 (B2 light source) in FIG. Show. But when the two lights are illuminated at the same time, the two curves are added to obtain a zigzag curve as shown in Figure 14B. Not only are the two edges unclear, the gap between the peaks and valleys is reduced, and the total dark area is gone. Clear, so the signal amplitude change after imaging will be smaller. The difficulty of fingerprint detection is increased, and the “amplitude” of the peak-to-valley gap becomes smaller, so the allowable error of the threshold is 7
本紙張尺度適用中國國家標準（CNS)A4規格（210 X 297公釐） 480454 A7 B7 五、發明說明（<) . 差的幅度也變得更小。故精確度也較差，而檢出的穩定度 較差。 因此，爲了維持單邊照射背光坡有明顯界限的優點， 必須避免K1與K2重疊。爲此目的，人們或許會想到將左 、右光源(B1)(B2)打亮的時間錯開，分別先後二次取像，例 如先打亮(B1)得到第一次取像的對應曲線K1，再打亮(B2) 得到第二次取像的對應曲線K2，然後將二者各取其背光坡 的資料，則上述界限模糊，對比偏低的情形就可解決。但 這種作法又有其他缺點。第一、要先後成像兩次，故成像 時間大大增加會佔用網路連線時間，第二、第一次成像後 ，所得資料(例如K1)須暫存在記憶體中，以在第二次成像 後與K2比較。如此就需要額外的暫存記億體，其容量要 涵蓋一整個矩陣圖面，成本很可觀，第三、由於左、右光 源(B1)(B2)的影像資料(ΚΙ)、（K2)係在不同時間取得，如果 第一次取得與第二次取像之間手指有任何移動，則測試的 精確度就會大幅下降。 因此本發明的主要目的在於提供一種高淸晰度指紋讀 取機，能利用設在手指左右兩側的二組光源同時作一次成 像，以達成準確的讀取效果，而無上述的缺點。 依本發明，這種目的的達成之道，係利用一種彩色的 面影像感測器(例如彩色CCD或彩色CMOS影像感測器)來 接收指紋影像，而所用之二組光源則爲二個不同色之原色 光源（紅、綠、藍光源）的任何二種，分別設在手指的左右 兩側，且宜爲紅色與綠色LED。 8 _ 本紙張尺度適用中國國家標準（CNS)A4規格（210 X 297公釐） (請先閱讀背面之注意事項再本頁) 訂· 經濟部智慧財產局員工消費合作社印製 480454 A7 B7 五、發明說明（L) · 彩色面影像感測器的每一個圖元包含三個原色子圖元 ，分別只接收紅光、綠光與藍光。爲此目的’每一個子圖 元都有相關的濾色片，例如紅色子圖元就設有過濾藍光及 綠光的濾色片，只讓紅光通過。因此如果我們用紅色led 作第十三圖中的光源(B1)，用綠色LED作第十三圖中的光 源(B2)，則二個LED可以同時打光。由於紅色LED的紅光 只會被彩色面影像感測器的紅色子圖元吸收，而綠色LED 的綠光只會被其綠色子圖元吸收，故二者造成的影像信號 分別由不同端子輸出，不會重疊也不會造成明暗對比降低 ，邊緣模糊的情事。故從紅色子圖元的端子與綠色子圖兀 的輸出端可分別得到如第十四圖A所示的不重疊而完整的 K1與K2曲線。 如上述，每個人膚色深淺相差可以很大(尤其是有色人 種與白人），且充血的泛紅手指與貧血的蒼白手指對紅光、 綠光的反射也有不同的影響，爲了要對個別情形的差異作 補償，依本發明的另一特點，係提供一種每次打光時總曝 光量自動調整的驅動電路。當測有色人種指紋時，由於其 反射率較低，可自動將總曝光量增加，使面影像感測器仍 能收到一樣多的光，使其輸出信號維持在最佳狀況。 又，現代的影印、攝影製版及光刻蝕等技術十分發達 ，且可達相當高的解析度，因此人們有時不慎，自己的指 紋可能被人利用上述技術僞造濫用。爲了防止不法之徒使 用僞造的平面指紋圖形騙過指紋讀取機，故依本發明另一 特點，係在提供一種指紋的防僞手段，其原理在於利用本 9 (請先閱讀背面之注意事項再d寫本頁) if -·線. 經濟部智慧財產局員工消費合作社印製 本紙張尺度適用中國國家標準（CNS)A4規格（210 '297公釐） 經濟部智慧財產局員工消費合作社印製 480454 A7 __ B7 五、發明說明（y ) · 發明的這種雙光源取像是一種「立體」圖形（亦即從左右 二邊所取像之信號略有差異），而一般較廉價的指紋僞造 都只有平面效果（亦即左右二邊取像之信號完全相同）， 因此只要做一道平面圖形的排除，即可有效防止大多數的 僞造指紋。即使用光刻飽製成立體的指紋模型，用本發明 仍可辨出（此點將於下文說明）。 最後，本發明還關於一種利用前述原理讀取指紋的方 法。 本發明之重要內容及基本原理已於上文詳述。關於其 進一步細節再配合圖式中之實施例作說明： 第一圖A係本發明指紋讀取機的示意圖； 第一圖B係本發明另一改良實施例的光路徑示意圖， 第二圖係本發明二組LED排列方式的底視示意圖， 第三圖係一僞造之平面指紋影印樣品在被讀取時的余斗 向底視圖， 第四圖A、B係第三圖中讀取所得之紅、綠信號圖， 第四圖C係用光刻蝕僞造之立體指紋構造模型， 第五圖係本發明指紋讀取機的方塊線路圖， 第六圖A、B係面影像感測器之紅色及綠色信號vR、 Vg圖， 第七圖A〜D爲該二信號分別以高、低臨限値二元化 的信號圖形，第七圖E爲該二元化之信號再經及閘Ih，處 理後的最終指紋影像信號己， 第八圖係本發明方法的流程圖， 10 本紙張尺度適用中國國家標準（CNS)A4規格（21〇 X 297公釐） (請先閱讀背面之注意事項再本頁) fl -線· 480454 A7 B7 五、發明說明（次） 第九圖A與B分別本發明的讀取機在封閉資訊系統及 開放網路系統的應用情形， 第十圖爲一種習用指紋辨識機的示意圖， 弟十~圖係一種假想單邊投光在指紋上的不意圖， 第十二圖A爲第^^一圖所得之信號在橫方向上的走勢 圖’第十二圖B，C爲第十二圖A的信號作二元化後所得 之信號， 第十三圖係習知雙邊用同色光源投光在指紋上的情形 第十四圖A係第十三圖中二同色光源所分別產生之信 號 Κι，K2，This paper size applies the Chinese National Standard (CNS) A4 specification (210 X 297 mm) 480454 A7 B7 V. Description of the invention (<). The margin of difference has become smaller. Therefore, the accuracy is also poor, and the stability of detection is poor. Therefore, in order to maintain the advantages of unilaterally illuminated backlight slopes, K1 and K2 must be avoided from overlapping. For this purpose, people may think of staggering the lighting time of the left and right light sources (B1) (B2) and taking two images one after the other, for example, first lighting (B1) to obtain the corresponding curve K1 of the first image. Highlight (B2) again to obtain the corresponding curve K2 of the second image acquisition, and then take the data of the backlight slope for each of them. The above boundary is blurred and the situation with low contrast can be resolved. However, this approach has other disadvantages. First, the imaging must be performed twice, so the imaging time will increase greatly, which will take up the network connection time. Second, after the first imaging, the obtained data (such as K1) must be temporarily stored in the memory for the second imaging. Compared with K2. In this way, an additional temporary memory of 100 million volumes is needed, and its capacity should cover an entire matrix surface, and the cost is considerable. Third, due to the left and right light sources (B1) (B2), the image data (ΚΙ), (K2) are Acquired at different times, if there is any finger movement between the first acquisition and the second acquisition, the accuracy of the test will be greatly reduced. Therefore, the main object of the present invention is to provide a high-definition fingerprint reader, which can use two sets of light sources located on the left and right sides of a finger to make an image at the same time to achieve an accurate reading effect without the above-mentioned disadvantages. According to the present invention, this purpose is achieved by using a color surface image sensor (such as a color CCD or color CMOS image sensor) to receive fingerprint images, and the two sets of light sources used are two different Any of the two primary color light sources (red, green, and blue light sources) is located on the left and right sides of the finger, and should be red and green LEDs. 8 _ This paper size is in accordance with Chinese National Standard (CNS) A4 (210 X 297 mm) (please read the precautions on the back before this page) Order · Printed by the Intellectual Property Bureau of the Ministry of Economic Affairs and Consumer Cooperatives 480454 A7 B7 V. Description of the Invention (L) · Each picture element of the color surface image sensor contains three primary color sub-picture elements, and only receives red light, green light and blue light respectively. For this purpose, each sub-picture element has an associated color filter. For example, the red sub-picture element is provided with a filter for filtering blue and green light, and only red light is allowed to pass through. Therefore, if we use a red LED as the light source (B1) in the thirteenth figure and a green LED as the light source (B2) in the thirteenth figure, the two LEDs can light at the same time. Because the red light of the red LED will only be absorbed by the red sub-picture element of the color surface image sensor, and the green light of the green LED will only be absorbed by its green sub-picture element, the image signals caused by the two will be output by different terminals respectively. , Will not overlap or cause a decrease in light and dark contrast, blurred edges. Therefore, the non-overlapping and complete K1 and K2 curves shown in Figure 14A can be obtained from the terminals of the red sub-picture element and the output of the green sub-picture element, respectively. As mentioned above, the skin tone of each person can be very different (especially colored people and white people), and congested red fingers and anaemic pale fingers also have different effects on the reflection of red and green light. According to another feature of the present invention, a driving circuit for automatically adjusting the total exposure amount each time lighting is provided is provided. When the colored human fingerprint is measured, the total exposure can be automatically increased due to its low reflectivity, so that the area image sensor can still receive as much light, and its output signal can be maintained in the best condition. In addition, modern technologies such as photocopying, photoengraving, and photolithography are very developed and can reach a very high resolution. Therefore, people are sometimes not careful, and their fingerprints can be forged and abused by people using the above-mentioned technologies. In order to prevent unscrupulous persons from using a forged flat fingerprint pattern to deceive the fingerprint reader, another feature of the present invention is to provide a fingerprint anti-counterfeiting method. The principle is to use this 9 (please read the precautions on the back before d write this page) if-· line. Printed on the paper by the Consumer Property Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs. Printed on a paper of Chinese National Standard (CNS) A4 (210 '297 mm). A7 __ B7 V. Description of the invention (y) · The invention of this dual light source image acquisition is a "stereoscopic" pattern (that is, the signals taken from the left and right sides are slightly different), and generally cheaper fingerprint forgeries are Only the plane effect (that is, the signals taken from the left and right sides are exactly the same), so as long as a plane figure is excluded, most forgery fingerprints can be effectively prevented. Even if a three-dimensional fingerprint model is made using photolithography, it can still be identified with the present invention (this point will be described later). Finally, the invention also relates to a method for reading fingerprints using the aforementioned principle. The important content and basic principles of the present invention have been described in detail above. Further details will be described in conjunction with the embodiments in the drawings: The first diagram A is a schematic diagram of a fingerprint reader of the present invention; the first diagram B is a light path diagram of another modified embodiment of the present invention, and the second diagram is A bottom view of the arrangement of two sets of LEDs in the present invention. The third figure is a bottom view of a counterfeit flat fingerprint photocopy sample when it is read. The fourth figures A and B are obtained by reading the third figure. The red and green signal diagrams, the fourth diagram C is a three-dimensional fingerprint structure model forged by photolithography, the fifth diagram is a block circuit diagram of the fingerprint reader of the present invention, and the sixth diagrams A and B are the surface image sensors. Red and green signals vR, Vg. The seventh diagrams A ~ D are the signal patterns of the two signals with high and low thresholds, respectively. The seventh diagram E is the binary signal re-pass and gate Ih. The final fingerprint image signal after processing has been processed. The eighth figure is a flowchart of the method of the present invention. 10 This paper size applies the Chinese National Standard (CNS) A4 specification (21 × 297 mm). (Please read the notes on the back first (Further page) fl-line · 480454 A7 B7 V. Description of the invention (times) The ninth figures A and B respectively show the application of the reader of the present invention in a closed information system and an open network system. The tenth figure is a schematic diagram of a conventional fingerprint recognition machine, and the tenth figure is a hypothetical unilateral light-emitting device. Unintended on the fingerprint, the twelfth picture A is the trend of the signal obtained in the horizontal direction of the first picture ^^ twelfth picture B, C is the signal obtained by binarizing the signal of the twelfth picture A The signal, the thirteenth picture is the case where the two-sided light source with the same color is used to project light on the fingerprint. The fourteenth picture A is the signal generated by the two same-color light sources in the thirteenth picture, K2,
第十四圖Β係第十四圖Α中二信號加成的結果Kj+IG ， 第一圖A係一個本發明的指紋讀取機(L，）。一如第十 之指紋讀取機(L)，它有一透明放置板(AP)及一透鏡系統 (LS) °與習知者不同者，它的面影像感測器是一種彩色面 影像感測器(FIS)，且它具有二組不同原色的光源：一個紅 色LED(Br)及一個綠色LED(Bg)，分別設在放置板(AP)兩側 下方。 關於光源(BR)(B。），此處要再補充說明一點：由於人的 指紋一般不是環形就是渦形，換言之，在指頭末節的遠體 自由端（指尖）與近體關節端的指紋並非縱向，而是趨近橫 向的(如第二圖所示）。爲了使LED光線大致都垂直於指紋 稜線和山谷’以得到最明確的界線，故左右各設有三紅色 11 本紙張尺度適用中國國家標準（CNS)A4規格（210 X 297公釐） (請先閱讀背面之注意事項再本頁) 言 Γ 經濟部智慧財產局員工消費合作社印製 480454 A7 B7 五、發明說明（7) , (Br1)(Br2)(Br3)或綠色(Bch)(B(32)(BC3)，分別負責由下方斜斜照 明指紋(FA)的前部、中部及後部，一如前面第十三圖所述 情形，左邊來的紅光對山谷的左邊緣較易分辨，而右邊來 的綠光對山谷右邊緣較易分辨，因此採用紅光的左邊緣及 綠光的右邊緣（其方法將於後述）可得到淸晰的立體指紋 的資料。指紋凹凸面的解析度因此可大大提高。 又，爲了使指紋(FA)各部位的亮度均勻，故所用LED 宜爲不聚焦的透明封裝的點光源，可在稍遠處得到均勻的 照光強度。 要防止不法之徒利用影印或印刷得到的指紋來朦混， 十分簡單。由於指紋是立體的構造，因此正如第十四圖A 曲線IG，K2所示，左邊的紅LED產生紅色的影像部分(K1) 的左緣淸晰，右緣較緩坡變化，而右邊的緣LED產生的綠 光的影像部分(K2)則是右緣淸晰，左緣較緩坡變化。換言 之，真正立體的指紋，其紅色的影像部分與綠色的影像部 分，應如第十四圖中A的ΚΙ、K2，兩者不大一樣。但是 如果用平面的指紋影印樣本(VF)來矇混(見第三圖），則由於 其山脊⑴與山谷⑴沒有立體構造，從左邊右邊照來，亮度 完全一樣，如第四圖A與B中所示，因此產生的紅色影像 (K〇與綠色影像(K。)大致一樣。只要把Kr與K。作比較（例 如用一個二元比較器），如果大致相同，就知道是平面印 刷的假指紋。 爲了避免不必要的耗電及延長LED的壽命’故將此指 紋讀取機設計成使led只在讀取指紋時才會短暫地啓亮一 12 ^紙張尺度適用中國國家標準（CNS)A4規格（210 x 297公釐） "~ (請先閱讀背面之注咅？事項再本頁) 訂-_ -_線· 經濟部智慧財產局員工消費合作社印製 480454 A7 B7 五、發明說明（/c) . 瞬間（例如約5msec〜10msec)。則每次讀取時影像感測器 (FIS)的總曝光量就與總點亮時間成正比。因此可藉控制 LED點亮時間得到最佳之影像信號品質。 如前述，爲了使彩色面影像感測器(FIS)不論手指的色 澤如何都能保持每次取像時恆定的總接收量，因此我們可 藉控制LED的發光時間長短來配合。當從指紋反射的光線 較弱時（例如有色人種的手指只有白人一半的反射率）可將 LED照明時間加倍。爲此目的，該紅色LED(BR1)(BR〇(BR3) 及綠色(Bch)(Bc32)(B〇3)可各接一單擊器(MM)，該二單擊器接 受一個尖峰檢出器(SD)的信號，當影像感測器(FIS)接收到 的紅光及綠光越弱，則其輸出VR，V。也越弱，則尖峰測器 送到各單擊器的信號較強，使單擊器(MM)每次觸發期間較 長。 第五圖顯示本發明指紋讀取機的一方塊電路圖。如前 所述，電路元件主要有發光元件的各三個紅綠 LED(B^)(BR2)BR3)(Bcn)(Bcn)(B〇3)、接收元件(彩色面影像感測 器：FIS)、二元化單元(BE)、防僞單元(VFS)、光量調節單元 (LR)。此外，一如傳統的CCD裝置，有一時序產生器分別 接到該CCD及後級(NS)使二者同步。 三個紅色LED宜用串聯，三個綠色LED亦然，如此 線路較簡單且電流完全一致。 如前所述，面影像感測器的每一個圖元(FIS)中，其紅 色子圖元的信號Vr，以及其綠色子圖元的信號V。各送到 一比較器(KR)、（K。)與一臨限値VSR、Vs。比較而作二元化， 13 本紙張尺度適用中國國家標準（CNS)A4規格（210 X 297公釐） (請先閱讀背面之注意事項再Wk本頁) · •線. 經濟部智慧財產局員工消費合作社印製 480454 A7 B7 五、發明說明（//) . 變成0或1的紅色與綠色矩陣圖形資料（h)(h)。該二信號 (「0(7。)再送到一及閘(IL)使指紋邊緣更明確，（這點後面還 要說明）。 如前所述，如果讀到平面的僞造指紋，則信號VR=V。 ，亦即，因此將己經一反相器送到一及閘(U2)—輸 入端，而h則送入(U2)另一輸入端，則及閘的輸出P2恆爲 0，因此可在後級(NS)做出對付措施，將系統封鎖住。 如果沒有僞造情事，則後級就將信號依傳統方式處理 、壓縮（並小量暫存作爲傳送緩衝），以將指紋送到封閉或 開放的系統作辨識或存檔。 如前所述，光量調節單元(LR)可對紅光及綠光LED各 設一單擊器(MM)，接到一電晶體的基極，如此單極器觸發 期間，該LED就導通而發光。如此，即可利用面影像感測 器(FIS)的輸出(VR)(V。)的回授而使其每次讀取時總曝光量保 持恆定。 如前所述，依本發明，使輸出信號(VR)(VC)二元化用的 臨限値Vsr、Vsa可有很大的容許誤差。換言之，作二元化 所選用之臨限値Vsr、Vs〇選高一些或進一些，結果Pi都是 一樣。請看第六圖A、B及第七圖A〜E。第六圖A與B分 別爲信號Vr、與V。可，它們和第十四圖a圖中的曲線L 、K2相似。假設對Vr分別使用一高臨限値Vsiu及一低臨限 値Vsu，則得第七圖A、B的二元化合號PR1、PR2。而對V。 用一高臨限値Vcn及一低臨限値Vc2，則得到第七圖c，D 的 KG1，，我們採（PR1，，）(PR2，KCJ1)(PR1，KG2，）（‘ 14 }紙張尺度適用中國國家標準（CNS)A4規格（21G x 297公爱） (請先閱讀背面之注意事項再Wk本頁) 訂： i線- 經濟部智慧財產局員工消費合作社印制农 480454 A7 B7 五、發明說明（/l) ’ K C2)任一種組合通過及聞U 1 ’所得結果K 1都是一樣(第七 圖E)。換言之，Vr的較佳左緣與Vb的較佳右緣都會被留 下來。這說明了本發明在臨限値的設定上，有很大的容許 誤差。 茲再用第八圖的流程圖說明本發明的操作過程。當指 紋讀取機被一指尖放到其放置板上時就被啓動。左右二排 紅綠LED —齊投光。首先在單元LR中檢查：從指紋上反 射到FIS的光線是否足夠，如不夠，則增加單擊器(MM)的 觸發期間，以增加FIS的曝光。 其次在單元(VFS)中檢查指紋是否僞造，如有僞造，則 可發出一防僞信號到後級NS，採取相關措施(例如通知駐 警或將系統封鎖）。 如確認指紋非僞造，再將指紋讀取。在封閉系統的場 合就送去與系統中儲存的指紋樣品比對。在開放系統的場 合，則經網路送到網站的指紋庫比對，再由對方送回確認 的信號。 經濟部智慧財產局員工消費合作社印製 本發明只做成指紋讀取機(L’），而非指紋辨識機，因 此不需要後級的一些繁複的記憶體，處理機等元件。因此 可以做成像滑鼠一樣大小當做一種電腦或網站的周邊裝置 使用。由於構造簡單，成本低廉，故可廣泛使用，或當作 電腦的附屬元件一樣使用及販賣。它可以利用一適當介面 和電腦或任何封閉的資訊系統(CS)連接。如第九圖，也可 經一個網路介面（包含發射/接收電路)(NI)經一開放的網路 (N)與網站上的指紋庫(FB)連線。（第十圖）。 15 本紙張尺度適用中國國家標準（CNS)A4規格（210 X 297公釐） " — ' 一 480454 A7 B7 經濟部智慧財產局員工消費合作社印製 五、發明說明) . 如上述’本發明在實用上希望做成像滑鼠一般扁平形 狀，因此第一圖A中那種垂直光路徑的設計，會使構造高 度(Z方向)偏高。因此，可依第一圖B，在光途徑中加一面 鏡，使光路徑轉一角度(例如90° ，但不一定要爲此値）， 如此，可使高度大大減少。 最後一點，本發明的防僞偵測(VFS)，不僅可有效識別 一般平面的僞造指紋影印圖形，即使把該影印的指紋圖形 拿去作照像光刻蝕，做成一個相對應之類似指紋表面的凹 凸構造，仍可用本發明檢出，因爲光刻蝕技術產生之構造 係如第四圖C，兩邊近乎垂直甚至超過90度，因此左、右 光源照射時，其信號走熱仍和平面指紋圖形相似，亦即Kr 與K。大致相同。而人的指紋，在山腹處係呈如第十一、十 四圖的斜坡形’适種微構造很難用一般的加工技巧造成。 又，本發明並不限於只使用雙色光，也可同時使用第 三色光（例如藍色LED)來，其所設位置可在無紅綠LED 之處。例如紅、綠、藍光源各隔120。設在放置板(AP)下方 周圍。由於不同色光不會互相干擾與重疊，因此有許多可 能的排列方式，茲不贅述，只要一個色光源（例如紅色 LED)的180°的正對面左右一角度範圍(例如左右各60° ) 內沒有同色光源和它照射的區域重疊而使它的對比變弱即 可。有這種三色光源的場合，彩色CCD的三個子圖元的輸 出全被用上。 本發明所讀出的指紋界限分明，幾無過渡灰色地帶。 我們知道，灰色的圖元在傳輸上遠比黑、白的圖元費時。 16 (請先閱讀背面之注意事項再 --- 本頁：Fig. 14B is the result of adding two signals Kj + IG in Fig. 14A, and Fig. 1A is a fingerprint reader (L,) of the present invention. Like the tenth fingerprint reader (L), it has a transparent placement plate (AP) and a lens system (LS). Unlike the known, its face image sensor is a color face image sensor. Device (FIS), and it has two groups of light sources of different primary colors: a red LED (Br) and a green LED (Bg), which are respectively arranged below the sides of the placement board (AP). Regarding the light source (BR) (B.), I would like to add one more point here: Because human fingerprints are generally either circular or vortex, in other words, the fingerprints on the free end (finger tip) of the distal end of the finger and the joint end of the near body are not Vertical, but closer to the horizontal (as shown in the second figure). In order to make the LED light almost perpendicular to the fingerprint edge and the valley, to get the clearest boundary, there are three reds on the left and right. 11 This paper size applies the Chinese National Standard (CNS) A4 (210 X 297 mm) (Please read first (Notes on the back page are on this page.) Γ Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 480454 A7 B7 V. Invention Description (7), (Br1) (Br2) (Br3) or Green (Bch) (B (32) (BC3), responsible for obliquely illuminating the front, middle, and rear of the fingerprint (FA) from below, as in the case of the thirteenth figure above, the red light from the left is easier to distinguish the left edge of the valley, and the right The incoming green light is easier to distinguish the right edge of the valley, so using the left edge of red light and the right edge of green light (the method will be described later) can obtain clear three-dimensional fingerprint data. The resolution of the concave and convex surface of the fingerprint can therefore be It is greatly improved. In addition, in order to make the brightness of each part of the fingerprint (FA) uniform, the LED used should be an unfocused transparent packaged point light source, which can obtain uniform light intensity at a distance. It is necessary to prevent criminals from using photocopying or Printed fingerprints come Because the fingerprint is a three-dimensional structure, as shown in the curve IG and K2 in Figure 14A, the red LED on the left produces a red edge of the red image part (K1), and the right edge changes more slowly. The image part (K2) of the green light produced by the right edge LED is clear on the right edge, and the left edge changes more slowly. In other words, for a real three-dimensional fingerprint, the red image part and the green image part should be like the tenth. The KI and K2 of A in the four pictures are not the same. However, if a flat fingerprint photocopy sample (VF) is used to confuse them (see the third picture), since the ridge ridge and valley ridge have no three-dimensional structure, from the left to the right As a result, the brightness is exactly the same, as shown in the fourth pictures A and B, so the red image (K0) is roughly the same as the green image (K.). Just compare Kr and K. For example, use a binary Comparator), if they are roughly the same, you know that it is a fake fingerprint printed on a plane. In order to avoid unnecessary power consumption and extend the life of the LED ', this fingerprint reader is designed so that the LED will only be short when reading the fingerprint Di Qiliang 1 12 ^ Paper scale Printed in China National Standard (CNS) A4 (210 x 297 mm) " ~ (Please read the note on the back? Matters on this page first) Order -_ -_ Line · Printed by the Intellectual Property Bureau of the Ministry of Economic Affairs and Consumer Cooperatives 480454 A7 B7 V. Description of the invention (/ c). Instantaneous (for example, about 5msec ~ 10msec). The total exposure of the image sensor (FIS) is proportional to the total lighting time each time it is read. Therefore, it can be borrowed Control the LED lighting time to get the best image signal quality. As mentioned above, in order to keep the color surface image sensor (FIS) regardless of the color of the finger, it can maintain a constant total reception amount every time you take an image, so we can borrow Control the length of the LED light to match. When the light reflected from the fingerprint is weak (for example, a finger of a colored person has half the reflectivity of a white person), the LED lighting time can be doubled. For this purpose, the red LEDs (BR1) (BR0 (BR3) and green (Bch) (Bc32) (B〇3) can each be connected to a clicker (MM), and the two clickers accept a spike detection When the red and green light received by the image sensor (FIS) is weaker, it outputs VR, V. The weaker it is, the weaker the signal sent by the spike detector to each clicker is. Strong, making the clicker (MM) each triggering period longer. The fifth figure shows a block circuit diagram of the fingerprint reader of the present invention. As mentioned earlier, the circuit element mainly has three red and green LEDs (each of which is a light emitting element). B ^) (BR2) BR3) (Bcn) (Bcn) (B〇3), receiving element (color surface image sensor: FIS), binary unit (BE), anti-counterfeit unit (VFS), light quantity adjustment unit (LR). In addition, like a conventional CCD device, a timing generator is connected to the CCD and the rear stage (NS) to synchronize the two. Three red LEDs should be connected in series, as are three green LEDs. This way the circuit is simpler and the currents are exactly the same. As mentioned above, in each image element (FIS) of the area image sensor, the signal Vr of its red sub-picture element and the signal V of its green sub-picture element. Each is sent to a comparator (KR), (K.) and a threshold 値 VSR, Vs. For comparison and dualization, 13 paper sizes are applicable to the Chinese National Standard (CNS) A4 (210 X 297 mm) (please read the precautions on the back before going to this page) · • Line. Employees of the Intellectual Property Bureau of the Ministry of Economic Affairs Printed by the Consumer Cooperative 480454 A7 B7 V. Description of the invention (//). The red and green matrix graphic data (h) (h) that becomes 0 or 1. The two signals ("0 (7.)" Are then sent to the one and gate (IL) to make the edge of the fingerprint clearer (this will be explained later). As mentioned earlier, if a forged fingerprint on a plane is read, the signal VR = V. That is, therefore, an inverter is sent to an AND gate (U2) -input, and h is fed to the other input of (U2), and the output of the AND gate is always 0, so Countermeasures can be made at the backstage (NS) to block the system. If there is no forgery, the backstage processes and compresses the signal in the traditional way (and temporarily stores it as a transmission buffer) to send the fingerprint to Closed or open system for identification or archiving. As mentioned earlier, the light quantity adjustment unit (LR) can be equipped with a clicker (MM) for each of the red and green LEDs, connected to the base of a transistor. During the trigger of the polarizer, the LED is turned on to emit light. In this way, the feedback of the output (VR) (V.) of the surface image sensor (FIS) can be used to keep the total exposure constant at each reading As mentioned above, according to the present invention, the thresholds Vsr, Vsa for binarizing the output signal (VR) (VC) can have a large allowable error. In other words, as The thresholds used for the dualization: Vsr and Vs0 are selected higher or higher, and the results Pi are the same. See the sixth graphs A and B and the seventh graphs A to E. The sixth graphs A and B are respectively The signals Vr and V. However, they are similar to the curves L and K2 in the fourteenth figure a. If we use a high threshold Vsiu and a low threshold Vsu for Vr, respectively, we get the seventh chart A, The binary combination of B is PR1 and PR2. For V. Using a high threshold 値 Vcn and a low threshold 値 Vc2, we get the KG1 of the seventh graph c, D. We use (PR1 ,,) (PR2 , KCJ1) (PR1, KG2,) ('14) The paper size applies the Chinese National Standard (CNS) A4 specification (21G x 297 public love) (Please read the precautions on the back before Wk page) Order: i-line-economy The Ministry of Intellectual Property Bureau's Consumer Cooperative Printed Agriculture 480454 A7 B7 V. Description of the Invention (/ l) 'K C2) Any combination passed and U 1' the result K 1 is the same (seventh figure E). In other words, Both the better left edge of Vr and the better right edge of Vb will be left. This shows that the present invention has a large allowable error in the setting of the threshold value. The flow chart in FIG. 8 is used to explain Describe the operation process of the present invention. The fingerprint reader is activated when it is placed on its placement board by a fingertip. The left and right two rows of red and green LEDs — all cast light. First check in the unit LR: reflected from the fingerprint to Whether the light of the FIS is sufficient, if not, increase the trigger period of the clicker (MM) to increase the exposure of the FIS. Secondly, check whether the fingerprint is forged in the unit (VFS). If it is forged, an anti-counterfeit signal can be sent to After the NS, take relevant measures (such as notifying the police or blocking the system). If it is confirmed that the fingerprint is not forged, read the fingerprint. In the closed system, it is sent for comparison with the fingerprint samples stored in the system. In the case of an open system, the fingerprint database sent to the website via the network for comparison, and then the other party sends back a confirmation signal. Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economics The present invention is only made as a fingerprint reader (L '), not a fingerprint identification machine, so some complicated memory, processor and other components in the subsequent stage are not needed. Therefore, it can be used as a mouse or a peripheral device of a website. Because of its simple structure and low cost, it can be widely used or sold as an accessory to a computer. It can be connected to a computer or any closed information system (CS) using an appropriate interface. As shown in the ninth figure, it can also be connected to the fingerprint database (FB) on the website through an open network (N) through an Internet interface (including transmit / receive circuits) (NI). (Tenth picture). 15 This paper size applies the Chinese National Standard (CNS) A4 specification (210 X 297 mm) " — 'One 480454 A7 B7 Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 5. Description of the invention). It is practically desirable to make the imaging mouse generally flat in shape, so the design of the vertical light path in the first figure A will make the structural height (Z direction) higher. Therefore, according to the first figure B, a mirror can be added to the light path to turn the light path by an angle (for example, 90 °, but it is not necessary to do so). In this way, the height can be greatly reduced. Finally, the anti-counterfeit detection (VFS) of the present invention can not only effectively identify fake fingerprint photocopy patterns on general planes, but even if the photocopying fingerprint pattern is used for photolithographic etching, a corresponding fingerprint-like surface is made. The concave-convex structure can still be detected by the present invention, because the structure produced by the photolithography technology is as shown in Figure C, the two sides are almost vertical or even more than 90 degrees, so when the left and right light sources are illuminated, the signal heat is still flowing and the plane fingerprint The graphics are similar, namely Kr and K. Roughly the same. Human fingerprints are slope-shaped as shown in Figures 11 and 14 on the mountain belly. Suitable microstructures are difficult to create with ordinary processing techniques. In addition, the present invention is not limited to using only two-color light, and a third color light (for example, a blue LED) can also be used at the same time, and the set position can be in a place where there are no red and green LEDs. For example, the red, green, and blue light sources are separated by 120. Set around the bottom of the placement plate (AP). Since different colors of light do not interfere and overlap with each other, there are many possible arrangements, so I will not repeat them, as long as there is no left and right angle range (such as 60 ° left and right) of a color light source (such as a red LED) directly opposite to 180 ° The same-color light source and the area it illuminates can overlap to weaken its contrast. Where such a three-color light source is used, all three sub-pixel outputs of the color CCD are used. The fingerprints read by the present invention have clear boundaries and few transitional gray areas. We know that gray primitives take much longer to transmit than black and white primitives. 16 (Please read the notes on the back before --- this page:
言 Γ 良 本紙張尺度適用中國國家標準（CNS)A4規格（210 X 297公釐） 480454 A7 __B7_ 五、發明說明（丨A) · 因，此依本發明，指紋之隆凸邊緣淸晰明確，在二元化處理、 後，信號品質特佳，可大幅提高信號壓縮效率，故可大幅 減少傳輸時間，提高網路效率。 經濟部智慧財產局員工消費合作社印製 本紙張尺度適用中國國家標準（CNS)A4規格（210 X 297公釐）The paper size of the good paper is in accordance with the Chinese National Standard (CNS) A4 specification (210 X 297 mm) 480454 A7 __B7_ V. Description of the invention (丨 A) · Therefore, according to the present invention, the raised edges of the fingerprint are clear and clear. After the binary processing, the signal quality is extremely good, which can greatly improve the signal compression efficiency, so it can greatly reduce the transmission time and improve the network efficiency. Printed by the Employees' Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs This paper is sized to the Chinese National Standard (CNS) A4 (210 X 297 mm)