200823795 九、發明說明： 【發明所屬之技術領域】 本發明係關於一種利用二維條碼之幾何不變性數位浮水 印產生、隱藏與偵測方法。 【先前技術】 •現有之幾何不變性數位浮水印技術主要有兩大類’（1)利 用重新同步信號（re-synchronization signals)，此同步信號可 _ 為一同步模版（synchronization template)或自我同步信號 (self-synchronization signals)，此類方法主要在於先找出已 知特定位置之模版或可能之信號尖峰點（peaks)，.再利用這些 資訊校正影像，最後以校正邊之影像執行數位浮水印偵測； (2)利用幾何不變性特徵（geometric invariant features)或幾何 不變十生轉換域（geometric invariant transform domain)，此類 方法主要在於找出重要之特徵點（robust salient points)或在 ® 幾何不變性轉換域如Fourier-Mellin transform domain執行數 位浮水印偵測。 上述（1)利用重新同步信號的偵測方法，缺點為其通常是 搭配展頻（spread spectrum)數体浮水印隱藏與彳貞測技術，不 僅可隱藏訊息量較少（1 bit或64〜128 bits)且強烈依賴幾何失 真校正之精確度，但通常這類校正仍存在誤差，造成校正影 像與原始數位浮水印做關聯性比對（correlation matching)偵 測時因位置校正無法達到.100%之準確度，大幅降低偵測率 200823795 以致無法可靠的抵擋幾何攻擊；而（2)利用幾何不變性特徵的 偵測方法，缺點則是無法找到穩定的重要特徵點，也無法找 到完美的幾何不變性轉換域，例如在F〇urier_MeUin transom d〇main隱藏數位浮水印之方法其在做逆轉換 (mverse Founer-Mellin transf0rm)時會破壞轉換域影像信號 且同時也弱化⑽藏之部份數位浮水印信號，造成影像品質 劣化與數位浮水印偵測之困難度。 由此可見’上述習用方式仍有諸多缺失，實非_良善之 設計者，而亟待加以改良。 本案發明人鑑於上述習用方式所衍生的各項缺點，乃 思加以改良創新，並經多年苦心孤詣潛心研究後，終於成 研發完成本件利用二維條碼之幾何不變性數位浮水印方法 【發明内容】200823795 IX. Description of the Invention: [Technical Field] The present invention relates to a method for generating, hiding and detecting a geometrically invariant digital floating ink using a two-dimensional barcode. [Prior Art] • The existing geometric invariant digital watermarking technology mainly has two major categories '(1) using re-synchronization signals, which can be a synchronization template or a self-synchronizing signal. (self-synchronization signals), such methods are mainly to first find the template or possible signal peaks of known specific locations, and then use this information to correct the image, and finally perform digital watermark detection with the image of the corrected edge. (2) Using geometric invariant features or geometric invariant transform domains, such methods are mainly to find important salient points or in the geometry The invariant conversion domain, such as the Fourier-Mellin transform domain, performs digital watermark detection. The above (1) utilizes the detection method of the resynchronization signal, and the disadvantage is that it is usually matched with the spread spectrum digital watermark hiding and detecting technology, which can hide not only a small amount of information (1 bit or 64 to 128). Bits) and strongly rely on the accuracy of geometric distortion correction, but usually there is still error in this type of correction, which causes the correction image to be correlated with the original digital watermark. Correlation matching cannot be achieved due to position correction. Accuracy, greatly reducing the detection rate of 200823795, can not reliably resist geometric attacks; and (2) using the geometric invariance feature detection method, the disadvantage is that it can not find stable important feature points, and can not find perfect geometric invariance The conversion domain, for example, the method of hiding the digital watermark in F〇urier_MeUin transom d〇main, which destroys the conversion domain image signal and also weakens (10) the partial digital watermark signal when performing inverse transformation (mverse Founer-Mellin transf0rm) , resulting in image quality degradation and digital watermark detection difficulties. It can be seen that there are still many shortcomings in the above-mentioned methods of use, which are not the designers of goodness, but need to be improved. In view of the shortcomings derived from the above-mentioned conventional methods, the inventor of the present invention has improved and innovated, and after years of painstaking research, finally developed into a geometric invariant digital watermarking method using two-dimensional barcodes. [Summary]
本發明之目的即在於提供—種可克服習用方法無法同E 達到高容量、影像保真度與幾何強勒性之三大要求之利用. 維條碼之幾何不變性數位浮水1卩方法。 發月之人目的係在於提供一種採用兩階段式 不變性數位浮水印方法，在第—階段採用^ 碼之編碼方絲編魏料切訊息，㈣第二階段* 引⑽數量蝴職)方㈣執行專⑽數料水印訊: 之肷入與偵測之利用-絡Μ 〇 % 利用-維條碼之幾何不變性數位浮水印方 200823795 法。 本發明之另一目的係在於提供一種利用該QR Code二 維條碼具有高容量、高密度與高容錯之特性，可應用於數位 >于水印訊息之編碼與解碼功用之利用二維條碼之幾何不變 性數位浮水印方法The object of the present invention is to provide a method for overcoming the three requirements of high capacity, image fidelity and geometric strongness which can not be achieved by the conventional method. The geometric invariance of the bar code is a digital floating method. The purpose of the person to send the moon is to provide a two-stage invariant digital watermarking method. In the first stage, the code of the coded square wire is used to edit the message, (4) the second stage * (10) the quantity of the work) Execute the special (10) digital watermarking signal: the use of the intrusion and detection - the network Μ% uses the geometric invariance of the dimensional bar code digital watermarking party 200823795 method. Another object of the present invention is to provide a QR code two-dimensional barcode with high capacity, high density and high fault tolerance, which can be applied to digital > the use of two-dimensional barcode geometry for encoding and decoding of watermark messages. Immutable digital watermarking method
本發明之又一目的係在於提供一種透過QIMM在高容量 訊心肷入舆偵測能力可有效的完成幾何不變性數位浮水印。 可達成上述發明目的之利用二維條碼之幾何不變性數位 斤水P方法’主要包括有：達成上述發明目的之利用二維條 碼之幾刘變性數位浮水印方法，主要為兩階段式之幾何不 交14數位汙水印隱藏與偵測’在第一階段利用二 維條碼來編碼與同步數位浮水印訊息，而在第二階段利用索 引值餘數里化（QIMm)方法來執行qr匸滅數位浮水印訊息 之欲入與偵測。首先制QRc〇de二維條碼之定位圖案 (finder pattern)與校正圖案（alignment來自我同步數 位浮水印訊息，此方法之轉是其料何失真校正之準確度 有較好之容忍度，再透㈣職在高容量訊息嵌人與偵測能 力因此可有效的完成幾何不變性數位浮水印。。 【實施方式】 请參閲圖一所示， 數位浮水印方法實施例 係本發明利用二維條碼之幾何不變性 圖，主要包括第一階段的QRc〇de編 200823795 子欠印#102¾產生與第二階段㈣引值餘數量化法 (Q^iM)對於QR,CGd^碼數位浮水印之喪人與偵測， 1係以QR Code與影像空間域為說明範例，本發明所 揭示之方法同樣可適於其它種類二維條碼（如：_⑽、 她細ix CGde、Vi議】CQde)之定位方式，也可操作於 Fourier轉換域。 為易於了解本發明方法，我們利用簡化之QRc〇de來做 以下之實例說明： 第一階段：此簡化之QRCode請參閱圖二A所示，主要 係利用QRCoW定位圖案與校正圖案來編碼數位浮水印訊 息㈣1，功能在於债測解碼數位浮水印時辅助幾何校正。 編碼過之QR C—編碼數位浮水印r m大小為^ moduies之二元影像(r實際大小為25 χ 25 m_ies，但為方 便定位圖案之找尋，在r四周各加【m〇dule稱之為z_ 之白色框卜如则㈣心為户則其總晝素為”…〜^ 啦仏，纟中包含定位圖案、校正圖案與數位浮水印訊息 紹01。定位圖案乃由3個分別位於左上角、右上角與左下: 具同樣大小為7x7.modules之二元圖案組成，其圖案設計之 概念為使得其排列方式不太可能以隨機方式在上述3個特定 位置外之其它地方出現類似圖案，其主要特性（圖二B所示^ 為黑點與白點組成方式為不管從那一方向掃描，其連續黑白 200823795 點b:w:b:w:b比例皆為m】， • ·3·1 ·1 ’其功能在於定位數位浮水 印之位置與方向。位於， 於右下方之校正圖案大小為5 χ 5 modules之二元圖案电杰，甘 '成，、圖案設計之概念類似於定位圖 案，其主要特性（圖二C、劣财 所不）為黑點與白點組成方式為不管 從那一方向掃描’其連續黑白點b:w:b:w:b比例皆為 1 · 1 · 1其功月b係在於結合另3個^位圖案找出4個中心Another object of the present invention is to provide a geometric invariant digital watermarking that can effectively perform high-capacity signal-injection detection capability through QIMM. The geometric invariance of the two-dimensional bar code can be achieved by the method of the above invention. The method mainly includes: a plurality of Liu-denature digital watermarking methods using the two-dimensional bar code to achieve the above object, mainly for the two-stage geometry. Handling 14 digits of smear printing and detection 'in the first stage using 2D barcodes to encode and synchronize digital watermark messages, and in the second stage using index value remainder digitization (QIMm) method to perform qr annihilation digital watermarking The desire to enter and detect the message. First, the QRc〇de two-dimensional bar code finder pattern and correction pattern (alignment from my synchronous digital watermarking message, this method is better tolerate the accuracy of the distortion correction, and then through (4) The high-capacity message embedding and detection capability can effectively complete the geometric invariant digital watermarking. [Embodiment] Referring to Figure 1, the digital watermarking method embodiment is a second-dimensional bar code. The geometric invariance map mainly includes the first stage QRc〇de 200823795 sub-imprint #1023⁄4 generation and the second stage (four) index residual quantization method (Q^iM) for QR, CGd^ code digit watermarking funeral And the detection, 1 is based on the QR Code and the image space domain as an illustrative example, and the method disclosed by the present invention is also applicable to the positioning manner of other kinds of two-dimensional barcodes (such as: _(10), her fine ix CGde, Vi discussion] CQde) For easy understanding of the method of the present invention, we use the simplified QRc〇de to illustrate the following examples: Phase 1: This simplified QRCode is shown in Figure 2A, mainly for profit The QRCoW positioning pattern and the correction pattern are used to encode the digital watermark message (4). The function is to assist the geometric correction when decoding the digital watermark. The encoded QR C-coded digital watermark rm is the binary image of the moduies (r actual size) It is 25 χ 25 m_ies, but in order to facilitate the search for the positioning pattern, add [m〇dule called z_ white box to the other side of the r] (4) the heart is the household, then its total quality is "...~^ 啦, The 包含 contains positioning patterns, correction patterns and digital watermarking messages. 01. The positioning pattern consists of three binary patterns in the upper left corner, upper right corner and lower left: the same size is 7x7.modules, the concept of pattern design In order to make it possible to arrange a similar pattern in a random manner elsewhere in the above three specific positions, the main characteristics (the figure shown in Figure 2B is that the black and white dots are formed regardless of the direction from which they are scanned. , its continuous black and white 200823795 point b:w:b:w:b ratio is m], • ·3·1 ·1 'The function is to locate the position and direction of the digital watermark. Located at the bottom right of the correction pattern size for 5 χ 5 modules binary pattern electric Jie, Gan 'cheng, the concept of pattern design is similar to the positioning pattern, its main characteristics (Figure 2 C, bad money is not) for the black and white points composed of no matter One direction scan 'its continuous black and white point b:w:b:w:b ratio is 1 · 1 · 1 Its power month b is based on the other 3 ^ pattern to find 4 centers
點做為解碼數位浮水印之幾何校正參考點。扣*㈣赠、 定位圖案、校正®案還有介於定位圖案、校正圖案與數位浮 水印payload間估i m〇dule之白色框之外，數位浮水印 payload為384 bits，此數位浮水印訊息从ι〇ι可事先經 3DES或AES濟算法加密處理以增加安全性。 第二階段：綸碼好QR Code數位浮水印後，我們利用索 引值餘數量化（QIMM)法來執行QR c〇de數位浮水印之隱藏 與偵測。 一、數位浮水印之隱藏方法描述如下： 1·依照數位浮水印酽102之大小，選取待隱藏影像12〇1 之中間區塊來做為主体信號（h〇st signal)，令其為 义={々，&，···，〜} ’ 其中 η 為 27 XJ9X 27 xp，而；7 為 module 大小， 每一之元素Z用來隱藏1個bit之訊息，隱藏息後z改變為 Z。{〜七，…，〜}。對每一 Z之元素除以量化值5 (quantization step size)取得商數，再取最接近此商數之整數值為索引值 200823795 200 ,然後依照0X)除以2之餘數值免（m〇dulo)以及擬隱藏之 第ζ·個bit訊息所改變\值為；c;。假如k=m則：c;改變為， 否則以⑹(\)-1)*5或（2(A) + 1)W與A之差值較接近者為X;之改變 值，以數學式表示如下： (a) Take Q(xt) = Round{xt /δ). (b) k = Q{xi)modlThe point is used as a geometric correction reference point for decoding the digital watermark. Buckle* (4) Gift, Positioning Pattern, Correction® case, and the white box of the estimated im〇dule between the positioning pattern, the correction pattern and the digital watermark payload, the digital watermark payload is 384 bits, and the digital watermark message is from 〇 〇 can be encrypted in advance by 3DES or AES algorithm to increase security. The second stage: After the QR Code digital watermarking, we use the index value remainder quantization (QIMM) method to perform the hiding and detection of the QR c〇de digital watermark. The hiding method of the digital watermark is described as follows: 1. According to the size of the digital watermark 酽102, the middle block of the image to be hidden 12〇1 is selected as the main signal (h〇st signal), so that it is meaning= {々,&,···,~} ' where η is 27 XJ9X 27 xp, and; 7 is the size of the module, each element Z is used to hide the message of 1 bit, and z is changed to Z after hiding. {~Seven,…,~}. Divide the element of each Z by the quantization step size to obtain the quotient, and then take the nearest integer value of the quotient to the index value 200823795 200, and then divide by 0X) by the value of 2 (m〇 Dulo) and the hidden value of the bite message to be hidden \ value; c;. If k=m then: c; change to, otherwise (6) (\)-1) * 5 or (2 (A) + 1) W and A are closer to X; the change value is mathematically Expressed as follows: (a) Take Q(xt) = Round{xt /δ). (b) k = Q{xi)modl
If k = m then (1) else x； = s(x,;m) = argmm(P(Xi)~x )5 --------------(2) ηχ·) v y 其中上述Eqe (2)中之户⑹為必或（妳)+ 1)*在 2·待取得X後’將原影像/之27 X 27modules每一像素以 對應之允元素值取代，即得到隱藏過之影像/w202。 二、數位浮水印之偵測方法描述如下： 假設隱藏過之影像/w202經幾何攻擊（以旋轉300。且放大 I·2倍為例)後之影像為/:301(圖三A所示）； 1·從/U由取QR Code數位浮水印二元影像％401(圖三B 所杀）：對每一 ={“···,X;}之元素，其中r為256 X 256，除 以同樣之量化值5取得商數，再取最接近此商數之整數值為 索引值2(0，2«)除以2之餘數值即％第ζ·個bit訊息m*值， 以數學式表示如下： ' 200823795 (b) τη = Q{x]) mod 2. 2·找出3個定位圖案（finder pattern)之中心點402:在逐 4亍與逐列找所有付合連縯黑白點b:w:b:w:b比例皆為 1:1:3: k 1之線段，取其交叉點即為候選之定位圖案中心點， 再從這些候選點中決定出最可能之3個定位圖案中心點並取 得旋轉角度Θ，如圖三c所示。 3·經叙轉歸正後再找出校正圖案pattern)之中 心點：依照步驟2取得之沒旋轉％取得歸正後之影像4〇3， 在L右下方找出校正圖案之中心點，如圖三D所示。 4·正規化QR Code解碼數位浮水印『，4〇4 :以步驟2與3 取侍之仏4個參考點（Χ·，Λ)以及其在『，（或『）之已知對應點座 標位置‘\)(如圖四a中之4個標示點所示），其中 透過投影轉換（projective transf〇m)依下式做正規化： x^au±bv^£ c/u + ev + f su-^hv^v 一 "577^77 將4組對應之座標點分別代入上式取得8個參數值口〜办， 即可取得r，在以目對位置的二元值，崎解碼數位浮水印^ (如圖四A所示）與原編碼數位浮水印r(如圖四b所示）可 知，解碼效果相當好（在此例中，雖然之左下角定位圖案有 1個bit解碼錯誤，但數位浮水印訊息·㈠斤在之區域完全 正確）。 5.依續取得r㈣384 bits之數位浮水印訊息…〇5。 11 200823795 實驗結果：If k = m then (1) else x; = s(x,;m) = argmm(P(Xi)~x )5 --------------(2) ηχ·) Vy where the above Eqe (2) household (6) is must or (妳) + 1) * after 2·after X is obtained, 'the original image / 27 X 27modules is replaced by the corresponding allowed element value, that is, Hidden image / w202. Second, the digital watermark detection method is described as follows: Assume that the hidden image/w202 is geometrically attacked (by rotating 300. and magnifying I·2 times as an example), the image is /: 301 (shown in Figure 3A) 1·From /U by QR Code digital watermark binary image %401 (Figure 3B killed): For each ={"···, X;} element, where r is 256 X 256, except Obtain the quotient with the same quantized value of 5, and then take the nearest integer value of the quotient as the index value 2 (0, 2«) divided by the value of 2, that is, the % ζ·bit message m* value, in mathematics The expression is as follows: ' 200823795 (b) τη = Q{x]) mod 2. 2· Find the center point 402 of the three finder patterns: find all the cohesive black and white in 4逐 and column by column. Point b:w:b:w:b ratio is 1:1:3: k 1 line segment, take the intersection point as the candidate positioning pattern center point, and then determine the most likely 3 from these candidate points Position the center point of the pattern and obtain the rotation angle Θ, as shown in Figure 3c. 3. After the correction is corrected, find the center point of the correction pattern pattern: Obtain the corrected image according to the non-rotation % obtained in step 2. 4〇3, at the bottom right of L Find the center point of the correction pattern, as shown in Figure 3D. 4. Normalize the QR Code to decode the digital watermark ", 4〇4: Take the steps 2 and 3 to pick up the 4 reference points (Χ·,Λ) And its known corresponding point coordinate position '\) (as shown in the four marked points in Figure 4a), which is normalized by projection transformation (projective transf〇m) according to the following formula : x^au±bv^£ c/u + ev + f su-^hv^v A "577^77 Substituting the corresponding coordinate points of the four groups into the above formula to obtain 8 parameter values~, you can get r, in the binary value of the target position, the Suga decoding digital watermark ^ (as shown in Figure 4A) and the original coded digital watermark r (as shown in Figure 4b), the decoding effect is quite good (here) In the example, although the bottom left corner positioning pattern has 1 bit decoding error, the digital watermark message ((1) is completely correct in the area) 5. Continue to obtain the r (four) 384 bits digital watermark message...〇5. 11 200823795 Experiment result:
以256 X 256 “Lena”影像為待隱藏影像，量化值占為1〇， 以module size p分別為4、5與6之三種QR Code編碼數位 浮水印102嵌入影像，隱藏過數位浮水印之影像2〇2其pSNRThe 256 X 256 "Lena" image is used as the image to be hidden, and the quantized value is 1 〇. The three QR Code coded digital watermarks 102 with module size p of 4, 5 and 6 are embedded in the image, and the image of the digital watermark is hidden. 2〇2 its pSNR
值分別為 46.5 dB、44·5 dB 與 42·9 dB，圖五 a、B、C、D 所示為原始影像201與隱藏過數位浮水印之影像2〇2比較， 其影像保真度都相當好。由於我們將數位浮水印加在影像中 間區塊，因此無論是平移（translati〇n)或裁切（cr〇ppin幻，只 要此影像中間區塊仍存在，數位浮水印之偵測不受影響。為 了測試經旋轉（rotation)與縮放（scaling)之混合攻擊，將隱藏 過數位浮水印之影像202以不同旋轉角度（3〇。〜36〇。，步進 值為30°)，不同縮放比例（〇5〜15，步進後為〇ι)處理，得 到132張測試影像，另外以不同之數位浮水印訊息从ι〇ι重 覆-人上述模擬，針對每一不同module size皆得到13200 張（1〇〇 X 132)測試影像來測試其平均偵測率（detecti〇n ratio) ’偵測率定義為户令，其中义為摘測正確之訊息_ 數乂為所隱藏之全部訊息bits數，p比率值愈高表示數位 浮水印之偵測強靭性愈好。 為進一步區分旋轉或縮放各別之強靭性，收集了兩種統 a十值’其一為固定縮放比例計算在不同旋轉角度之平均偵測 率，另一為固定旋轉角度計算在不同縮放比例之平均偵測 12 200823795 率；結果分别如圖六A、B所示，對於〜為心5 與6 =平均⑭測率分別為Q㈣、g.988 _ q州，讀大，. 將數位浮水印訊息透過糾錯瑪（BCH或Reed s〇i_n) 先予以編，，則可再提高平均_率。 κ I7之偵測強靭性愈好'如減少部份可隱藏訊息量但The values are 46.5 dB, 44·5 dB and 42·9 dB respectively. Figure 5 a, B, C, and D show that the original image 201 is compared with the image of the hidden digital watermark 2〇2, and the image fidelity is Quite good. Since we add a digital floating watermark to the middle block of the image, whether it is translati〇n or cropped (cr〇ppin illusion, as long as the middle block of the image still exists, the detection of the digital watermark is not affected. Test a rotation and scaling hybrid attack, which will hide the image of the digital watermark 202 at different rotation angles (3〇.~36〇., step value is 30°), different scaling ratios (〇 5~15, after stepping, it is processed by 〇ι), and 132 test images are obtained. In addition, different digital watermark messages are repeated from ι〇ι--the above simulation, and 13200 sheets are obtained for each different module size (1 〇〇X 132) Test the image to test the average detection rate (detecti〇n ratio) 'The detection rate is defined as the household order, where the meaning is the correct measurement message _ number is the number of all hidden bits of the message, p The higher the ratio value, the better the detection of the strength of the digital watermark. To further distinguish the different toughness of rotation or scaling, two kinds of unified ten values are collected, one of which is the average of different rotation angles calculated by fixed scaling. Detection rate The other is the average rotation detection rate of the different rotation scales for the fixed rotation angle 12 200823795; the results are shown in Figure 6A, B, respectively, for ~5 and 6 = average 14 rate is Q (four), g.988 _ q state, read big,. The digital watermark message is edited first by error correction (BCH or Reed s〇i_n), then the average _ rate can be increased. κ I7 detection of the better the toughness 'such as the reduction Share the amount of information but
發月所提供之利用二維條碼之幾何不變性數位浮水印 方法，與其他習用技術.相互比較時，更具有下列之優點： 1·本發明係在於提供_種該QR CGde :維條碼之定 =圖案與校正®案來自_步數位浮水印訊息，好處是其對 成何失真权正之準確度有較好之容忍度。 2·本發明係在於提供—種透過Q][MM有較高之訊息嵌入 與偵測能力等功效。 本么月係在於提供一種結合前述兩項可有效的完成幾 何不變性數位浮水印方法，在可隱藏訊息量、影像保真度與 強轫性二方面之綜合評比優於現有之方法。 上列詳細說明係針對本發明之一可行實施例之具體說 明，惟該實施例並非用以限制本發明之專利範圍，凡未脫離 本發明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。 每上所述’本案不但在技術思想上確屬創新，並能較習 用物品增進上述多項功效，應已充分符合新穎性及進步性之 13 200823795 法定發明專利要件，爰依法提出中請，懇請貴局核准本件 發明專利申請案，以勵發明，至感德便。 【圖式簡單說明】 明參閱以下有關本發明一較佳實施例之詳細說明及其附 圖，將可進-步瞭解本發明之技術内容及其目的功效；有關 該實施例之附圖為： 圖為本發明利用二維條碼之幾何不變性數位浮水印方 法之系統流程圖，係本發明之代表圖； " B C為該利用二維條碼之幾何不變性數位浮水 印方法所使用簡化之QRC()de#定位圖案之特性說明圖； …圖三A、B、C、D為該利用二維條碼之幾何不變性數位 洋水印方法偵測影像與數位浮水印之幾何校正過程圖； 、圖A B為該利用二維條碼之幾何不變性數位浮水印 、、Q Code解碼數位浮水印與原始編碼數位浮比 較圖； 之影像比較 為该科用二維條碼之幾何不變性數位 浮水印方法之原始影像與隱藏過數位浮水印 圖；以及 二維條碼之幾何不變性數位浮水印 圖六A、B為該利用 方法平均偵測率比較圖。 【主要元件符號說明】 14 200823795 101數位浮水印訊息 102 QR Code編碼數位浮水印 201原始影像 202隱藏過數位浮水印之影像 301經幾何攻擊之影像 401經QIMM抽取出之QRCode數位浮水印 402找出3個定位圖案之中心點 403經旋轉歸正後再找出校正圖案之中心點 404 QR Code解碼數位浮水印 405解碼數位浮水印訊息The geometric invariant digital watermarking method using the two-dimensional bar code provided by the moon has the following advantages when compared with other conventional techniques: 1. The present invention is to provide the QR CGde: the bar code setting The = Pattern and Correction® case comes from the _ step digital watermark message. The advantage is that it has a good tolerance for the accuracy of the distortion. 2. The present invention is to provide a kind of effect through the Q] [MM has a higher message embedding and detection capability. This month is to provide a method for effectively performing geometric invariant digital watermarking in combination with the above two methods. The comprehensive evaluation of hidden information volume, image fidelity and robustness is superior to the existing method. The detailed description of the preferred embodiments of the present invention is intended to be limited to the scope of the invention, and is not intended to limit the scope of the invention. The patent scope of this case. Each of the above cases is not only innovative in terms of technical thinking, but also able to enhance the above-mentioned multiple functions compared with the conventional articles. It should be fully in line with the novelty and progressiveness of the 13200823795 statutory invention patents. The bureau approved the application for the invention patent, in order to invent the invention, to the sense of virtue. BRIEF DESCRIPTION OF THE DRAWINGS The technical contents of the present invention and the effects of the objects of the present invention will be further understood by referring to the following detailed description of the preferred embodiments of the invention and the accompanying drawings. The figure is a system flow chart of a geometric invariant digital watermarking method using a two-dimensional barcode, which is a representative diagram of the present invention; " BC is a simplified QRC used by the geometric invariant digital watermarking method using a two-dimensional barcode () de# positioning pattern characteristic description map; ... Figure 3 A, B, C, D is the geometric invariance process of the image invariant digital watermarking method using the two-dimensional bar code detection image and digital watermarking; AB is the geometric invariant digital watermark using the two-dimensional barcode, Q code decoding digital watermark and the original code digital floating comparison map; the image comparison is the original of the geometric invariant digital watermarking method of the two-dimensional barcode The image and the hidden digital watermark image; and the geometric invariance digital watermark of the two-dimensional bar code. The sixth and fourth B are the comparison of the average detection rate of the utilization method. [Main component symbol description] 14 200823795 101 digital watermark message 102 QR Code coded digital watermark 201 original image 202 hidden digital watermark image 301 geometric attack image 401 QIMM extracted QRCode digital watermark 402 The center point 403 of the three positioning patterns is rotated and corrected to find the center point of the correction pattern 404 QR Code decoding digital watermark 405 decoding digital watermark message
1515