TW532039B - Robust gray-level watermark embedding and retrieving technique - Google Patents

Robust gray-level watermark embedding and retrieving technique Download PDF

Info

Publication number
TW532039B
TW532039B TW90100412A TW90100412A TW532039B TW 532039 B TW532039 B TW 532039B TW 90100412 A TW90100412 A TW 90100412A TW 90100412 A TW90100412 A TW 90100412A TW 532039 B TW532039 B TW 532039B
Authority
TW
Taiwan
Prior art keywords
watermark
image
digital
gray
quantization
Prior art date
Application number
TW90100412A
Other languages
Chinese (zh)
Inventor
Jen-Cheng Jang
Tung-Shiau Chen
Chun-Hung Hsieh
Chuei-Shiung Tsai
Ming-Huang Chen
Original Assignee
Jen-Cheng Jang
Tung-Shiau Chen
Chun-Hung Hsieh
Chuei-Shiung Tsai
Ming-Huang Chen
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Jen-Cheng Jang, Tung-Shiau Chen, Chun-Hung Hsieh, Chuei-Shiung Tsai, Ming-Huang Chen filed Critical Jen-Cheng Jang
Priority to TW90100412A priority Critical patent/TW532039B/en
Application granted granted Critical
Publication of TW532039B publication Critical patent/TW532039B/en

Links

Landscapes

  • Editing Of Facsimile Originals (AREA)
  • Image Processing (AREA)

Abstract

The present invention relates to information security, and proposes a novel digital gray-level watermark system by applying the discrete cosine transform and digital quantization technique to embed the gray scale watermark into the digital image effectively, so as to achieve the purpose of protecting the intellectual property of image. The gray scale watermark is processed by discrete cosine transform in the present invention, so that the important information of watermark is concentrated on few frequency coefficients, which not only can strengthen the resistance of the watermark to the intentional/casual damage to the image, but also can enhance the image quality of the watermark brought through image attack after embedding the watermark. A quantization table is also designed to reduce the amount of data of gray scale watermark, so that the parameter value passing through digital quantization after discrete cosine transform can be close to the frequency value at the low frequency position in the digital image. Finally, the digital watermark is embedded in the way of replacing one-by-one. The digital watermark embedded into the digital image provides a high resistance to the image attack in the present invention, and can resist the attack of JPEG distortion compression and damaging signal processing.

Description

532039532039

五、發明説明( 近年來,隨著網際$明背景> 者被iri:, ,權的問題,尤其備受目失前像之智慧財產 上最廣為使用的影像保障技術之t。子水p技術是網際網路 印亦即所謂的浮水 P=pS)i浮水印取出程序(Exiting 訂 程序f浮水印藏入原始影像中並而ίί新5: !ΐ ί ^Tked^ ° ^ τ««^»Α^ίί!ΐίΐ ^疋讀-個私密的鍵(Private κ當^^用 Random Number Generator) ^#^(Seed)^ m勺位置’並向公正的第三者註冊時n〉于 ί著 1ίίΐίΤ,,Ι卩在晴上遞i發 者1乍%旋異或f5義時,合法的擁有者提出 經濟部智慧財產局員工消費合作社印t 從浮水印影像中取回所藏入^水印,二由 Α正的弟二者來驗證擁有權。 丹、、二由 具備?·=ίί討可以歸納H好的數位浮水印技術應 (2 )不易被偵測(Un-detection):對於非法佶用去一 欲利用其他技術(如:統計分析方法)^象中 本、,氏張尺度適用中國國家榡準(c ^規格(2ι〇χ 297公釐 532039 A7 B7 立、發明^説明(1) -- 浮水印藏入位置是相當的困難。 (3) :對於非法使用者而言,欲 ⑷巧,(Robust·):藏入影像中的浮水印須 μ、一般失真性壓縮及破壞性數位訊號處理之能^。_ (51回洋水印不需藉由原始影像辅助(ExtraeJ^ithout :由於不須多保留一份原始影像,因 此可減少儲存體的浪費。 ⑴即」寺處理(Real_Time Pr〇cessing):基於實用性 取出浮水印的動作必須能做到快速且有效率< H rF ’手水P技術可區為空間域(Spatial domain)盥頻率 i、iie^Ty domain)兩種。其中,空間域的方法 ίίίί 抵抗訊號處理的破壞;頻率域的方 ίίϊίΐί置5運鼻’但較能抵抗訊號處理的破壞,且 可分為ion)。一般數位浮水印 (Gray-Level Wate贿k )。黑心火==: 5:fi,二維影像;而灰階浮水印主 ί ί、ίΐ 影像。灰階浮水印除了視覺效果的呈現較優 經濟部智慧財產局員工消費合作社印^ 印卜人,且因其資料量大,較能抵抗外來之破壞 …、戶 13^^斗量過大,造成其藏入處理亦較為困難。 學者於1999年1月,提出了 一個利用離二 =換的數位⑦水印技術,其制的數位浮 [ 理’接著將轉換後的原始影像之中低以ΪίΪ; =影像中低頻參數值的位置。印;二字㈣ 用原始影像與藏有數位浮水印的影像兩利 210X 297公釐) ( cnsTmw^I---- 532039 五 A7 、發明説明(V. Description of the invention (In recent years, with the Internet background, people have been iri :, the problem of rights, especially the most widely used image security technology on the intellectual property before the eye-catching image. Zizui The p technology is the Internet printing, which is the so-called floating water P = pS. The floating watermark removal procedure (Exiting ordering procedure) f The watermark is hidden in the original image and new 5:! ΐ ί ^ Tked ^ ° ^ τ «« ^ »Α ^ ίί! Ϊ́ίΐ ^ 疋 Read-a private key (Private κ when ^^ using Random Number Generator) ^ # ^ (Seed) ^ m spoon position 'and register with a fair third party With 1 ί ΐ Τ Τ, Ι 卩 submitted the sender on the sunny day 1% or X5 meaning, the legal owner proposed that the employee ’s consumer cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs print the seal from the watermark image to retrieve the hidden ^ watermark Two, A's younger brother, to verify ownership. Dan ,, two, have? · = Ί Discuss the digital watermarking technology that can be summarized H should be (2) Not easy to detect (Un-detection): For illegal佶 Use other technologies (such as statistical analysis methods) ^ Like the Chinese version, the scale is applicable to the Chinese national standard (c ^ specifications (2 ι〇χ 297 mm 532039 A7 B7 Establishment and invention ^ Explanation (1)-It is quite difficult to hide the watermark in place. (3): For illegal users, want to be smart, (Robust ·): Tibetan The watermark in the image must be μ, the ability of general distortion compression and destructive digital signal processing ^. (51 times ocean watermark does not need to be assisted by the original image (ExtraeJ ^ ithout: because there is no need to retain an original image) Therefore, the waste of the storage body can be reduced. “Real processing” (Real_Time Pr〇cessing): The action of taking out watermarks based on practicability must be fast and efficient < H rF 'hand water P technology can be used as space Spatial domain (i, iie ^ Ty domain). Among them, the method of the spatial domain ίίίί resists the damage of signal processing; the frequency domain square is ίίϊίΐί5, but it is more resistant to the damage of signal processing, and It can be divided into ion). General digital watermark (Gray-Level Wate). Black heart fire ==: 5: fi, two-dimensional image; grayscale watermark master ί, ίΐ image. Grayscale watermark except visual The effect is better. Employees of Intellectual Property Bureau of the Ministry of Economy The Zuo Yin ^ Yin Bu people, and because of the large amount of data, more able to resist external damage ..., the amount of household 13 ^ ^ is too large, making it difficult to hide and handle. Scholars in January 1999, proposed a Using the digital ⑦ watermarking technique of Li == change, the digital floating system produced by it is then used to lower the original image after conversion; = the position of low-frequency parameter values in the image.印; 二字 ㈣ Use the original image and the image with the digital watermark to benefit 210X 297mm) (cnsTmw ^ I ---- 532039 5 A7 、 Explanation of the invention (

經濟部智慧財產局員工消費合作社印^ i訊號處/的破切|^,力:般性失錢縮及破壞 水印取回時需原影像。 早此力部不如灰階浮水印且浮 轉換的數位浮tjSp:二 =3出、:-種利用離散餘弦 Decomposition )將灰階、n / ^紋 位元分解(Bit P-);接著,由i文;忠 元層作為藏入的資訊(Ma4Tsf中挑>出右干個較重要的位Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs, ^ iSignal / Broken | ^, Force: General loss of money, shrinkage and destruction. The original image is required to retrieve the watermark. Earlier, the force was not as good as the gray-scale floating watermark and floating-converted digital floating tjSp: two = 3 out,:-a method that uses discrete cosine Decomposition to decompose the gray-scale and n / ^ grain bits (Bit P-); then, by I text; the loyalty layer as the hidden information (Ma4Tsf pick > out of the right to do a more important bit

Si位i水3的影像,得先前藏入^用量2525 失i性 仍需原影像。鱗理的破壞,然而,浮水印取回時 旦^^使用數位量化方法減少其欲藏入的浮 !新的數位灰階浮水印系統。本發明首先‘灰ί t在ΐ數的參數上,如此即可強化數位浮水 力^繼而為提升藏人數位浮水印後的影像;^^ 卩1^過數位Λ化後的錢缝與數位^ 像中低頻位置的頻率值相近,最後再以一對— 成數位浮水印的藏入。根據實驗結果顯示,本笋明所提1ί 方法除了在抵抗影像切割有較佳的成果外,另X 浐 JPEG失真性壓縮及破壞性訊號處理的破壞。 -几 本紙張尺度適標準(CNS) A4規^ττ^χ2ϋ (請先閱讀背面之注意事項再填寫本頁)The image of the Si water 3 in the Si position was obtained in the previous dose of 2525. The original image is still needed. The destruction of the scales, however, when the watermark is retrieved ^^ uses digital quantization methods to reduce the floats it wants to hide! The new digital grayscale watermark system. In the present invention, firstly, the parameters of the number are grayed, so that the digital water-flooding can be strengthened ^, and then the image after the watermark of the Tibetan number is enhanced; ^^ 卩 1 ^ money stitches and digits after digitalized ^ The frequency values in the low and medium positions of the image are similar, and finally they are hidden as a pair — a digital watermark. According to the experimental results, in addition to the better results of the image cutting method proposed by Ben Shuming, X 浐 JPEG distortion compression and destructive signal processing damage. -Several paper size standards (CNS) A4 rules ^ ττ ^ χ2ϋ (Please read the precautions on the back before filling this page)

532039 A7 B7 五 發明説明( 〈發明目的〉 本發明提出運用離散餘弦轉換與數的 =將用於數位浮水印系 二ί f極咼比例的影像裁剪的破壞之下,本發明所取回 的看起來仍是清晰而且是可 的強韋f性明顯地優於-般黑白 么明所取回的洋水印之影像品質亦有較佳的效果。+ 亜本的浮水印方法能夠滿足目前數位浮水印系統所 要f的备、件。因此,本發明的數位浮水印方法.陬 電子商務發展上最受重視關題_智# ^、、,;、, 最佳的解決方案。 义曰心对產%,提出 (請先閲讀背面之注意事項再填寫本頁)532039 A7 B7 Five invention descriptions (<Objective of the Invention> The present invention proposes to use discrete cosine transform and number = under the destruction of the image of the digital floating watermarking system 2 f f pole ratio destruction, the results retrieved by the present invention It is still clear and practicable, and its strong Wei f is obviously better than the image quality of the ocean watermark retrieved by the black-and-white black-and-white moming. It also has a better effect. + The original watermarking method can meet the current digital watermarking. The spare parts and components required by the system. Therefore, the digital watermarking method of the present invention is the most important issue in the development of e-commerce. The best solution is the best solution. (Please read the notes on the back before filling this page)

尺 張 紙 _^_本 經濟部智慧財產局員工消費合作社印製Rule paper _ ^ _ This printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs

W2039 五 A7 B7 浮^=月^^個植基於數位量化方法的數位 ίΗΞ-, X {x(i,j), 0^x(ij)^255} ⑴ W,(M),〇gP,q腹],〇iiU55} 忠 成、4=Γί位量化方‘少5的藏上量而成功地達 ίί 來’我們將詳盡地說明 、浮水印藏入處理演算法 X的找到—個規則將w的内容對應到 步驟1 :原始影像DCT轉換 處W圖 巧始影像X蝴成不重疊的8χ8像素區塊,然後, 步驟 水印影像有較高的_性而能夠抵抗-般 方法由$的每—個區塊頻域係數内 之範圍:圖以用來藏入^所選擇的申頻係數 、7 DCT轉換後的影像Μ从···》_)&gt;&lt;()}且每-個 被轉換的8x8區德交《 -\ ++1^ , 所決 (請先閲讀背面之注意事項再填寫本頁) 訂 經濟部智慧財產局員工消費合作社印製 532039 A7 年 B7 五 X Μ (Ν/8)χ(Ν/8) (3) 經濟部智慧財產局員工消費合作社印製 ,了增加安全性,本方法選取一個秘密 為隨機數產生器(PRNG)的種子,並對 =中頻係數。令每一個區塊^被選以= 係數組成5且5=(、“,_,...“同時,令所有 的!i組成一個MxM的影像imf。 步驟3:浮水印像素位置的隨機置換 ^了增加藏入的浮水印之安全性,W在藏 ,須將其所有像素先加以打散。本方法 U 堂w的母一個像素產生一個置換位置 2 巧可以朗影像中的另-個截然不同m 原來的浮水印W被置換成為Wr。 置7 步驟4:浮水印的DCT轉換和數位量化處理 的每一個像素是由一個位元組所構成,使 t Ρ貧料量太大而不能直接藏入原始影像ΐ成本發 明透過數位量化處理來減少藏入的資料$ 丄以使JfJ影像變形,因此本發明必 ϋί臧人的資料與原始影像的資 枓之間可此緊在。基於上述這些理由,一個可^ ^化表之,必然是從和5^的# : y付到的。言先,把wr分成不重疊的4χ4區 母一個區塊個別被DCT轉換。令&lt; = DCT(W 為’每一^固DCT^轉換區塊中的係數都有不同&amp; 例如DC係數是最重要的而其他係數的重要性 -齒形次序依序減低,因此在量化處理時要們J雙 每-個DCT轉換區塊,數中對應並得;〗曰^ 本紙張尺度適用中國國家標準(CNS ) Α4規格(210X297公--- (請先閱讀背面之注意事項再填寫本頁)W2039 Five A7 B7 floats ^ = month ^^ digits based on digital quantization method, X {x (i, j), 0 ^ x (ij) ^ 255} ⑴ W, (M), 〇gP, q ]], 〇iiU55} Zhongcheng, 4 = Γ quantized Fang 'successful amount of less than 5 and successfully reached ίί' We will explain in detail, the finding of the watermark hiding processing algorithm X-a rule will be w The content corresponds to step 1: the original image DCT conversion location W image is a non-overlapping 8 × 8 pixel block, and then, the step watermark image has a high degree of resistance and can resist-the general method consists of $ each The range within the block frequency domain coefficients: the map is used to hide the selected frequency coefficients, 7 DCT converted images M from ... '' _) &gt; &lt; ()) and each one is converted The 8x8 district of Germany, "-\ ++ 1 ^, decided (please read the precautions on the back before filling out this page) Ordered by the Intellectual Property Bureau Employee Consumer Cooperative of the Ministry of Economic Affairs 532039 A7 B7 Five X Μ (N / 8 ) χ (Ν / 8) (3) Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs to increase security. This method selects a seed that is a secret random number generator (PRNG), and sets = IF coefficient. Let each block ^ be selected with = coefficients composed of 5 and 5 = (, ", _, ..." At the same time, let all! I form an MxM image imf. Step 3: Random displacement of the watermark pixel position ^ Increase the security of the hidden watermark. W is in Tibet, and all its pixels must be broken up first. In this method, one pixel of the parent of U Tangw creates a replacement position. 2 It can clear another one in the image. The original watermark W of different m is replaced by Wr. Set 7 Step 4: Each pixel of the DCT conversion and digital quantization processing of the watermark is composed of a byte, which makes t ρ lean and cannot be directly used. The cost of hiding the original image is to reduce the hidden data through digital quantization to transform the JfJ image. Therefore, the present invention must closely link the resources of the Tibetan people and the resources of the original image. Based on the above The reason is that a ^^^ table must be paid from #: y and 5 ^. First, divide wr into non-overlapping 4 × 4 blocks, and each block is converted by DCT. Let &lt; = DCT ( W is' the coefficients in each DCT transform block are different &amp; for example DC The number is the most important and the importance of other coefficients-the order of the tooth shape is reduced in order, so we need to double the DCT conversion block during the quantization process, and the corresponding number is obtained;〗 ^ This paper scale is applicable to China National Standard (CNS) Α4 specification (210X297 male --- (Please read the precautions on the back before filling this page)

• ,Γ-, 五、發 明説明(, 表Tw 〇 A7 B7 ϊ三顯5對應到—個4x4區塊的4x4比重表•, Γ-, V. Explanation of the invention (, Table Tw 〇 A7 B7 ϊThree display 5 corresponds to a 4x4 weight table of a 4x4 block

Tw。接著,ά—跑现的4x4比重 Μ〜1Μί 4式(4)計异^斤有係數的平均值^i?〇(4)Tw. Next, ά—4x4 specific gravity of running out Μ ~ 1Μί The formula (4) calculates the average value of different coefficients ^ i? 〇 (4)

XmfXmf

ίΪ用等式(5)產生可調整量化表Q崎有的4x4資 Q(a,b) Μ μτ τ' Wr(a + 4xi,b + 4xj)歹 ΜΜ · 入mf χ — χ — xTw(a,b) (5) 請 閱 讀 背 面 5,Use equation (5) to generate an adjustable quantization table Q: 4x4 resources Q (a, b) Μ μτ τ 'Wr (a + 4xi, b + 4xj) 歹 ΜΜ · Enter mf χ — χ — xTw (a B) (5) Please read back 5,

I 4 4 ^ /、中Q(a,b)疋位於q中第a列第&amp;行的資料項· 經濟部智慧財產局員工消費合作社印製 ’/^,冗分散在中的每—個4x4區塊 仏匕一二:區塊進行量化操作。然後,產生 步驟5:浮水印藏入所選的係數中 在文mf中的每個係數都由宄q中的一個始仫齡$,,浮水印藏入的丄 容夾擇士另一個秘密鑰匙Κ3作為腦G的種 U被代替^有係數被替代的序列。令後面 步驟6:改寫原始影像被選的中頻係數 [紙張尺度適用fiiiijTcNS成非重疊的^ 訂 B7 發明説明丄?) &quot;~~ 區塊由大小8x8像素所構成,然後 勺轉換個別地應用於每一個區塊以声得:反 即,Xw=IDCT(iw^ 又于 Xw,亦 經濟部智慧財產局員工消費合作社印製 532039 五 w v^丄v八wy 影像的著作權擁有者將可調整量化表 ‘ 11 f中=註冊並獲得加人時間郵戳的可調G 2 ί 裁者並進行取回浮水印處理來驗證是否ίίίϊ 二、浮水印取回處理演算法 仏仲ί 發生爭議時,主張擁有權所有者財六 ΐ 所ί存的秘麵匙Kl,KA和時間郵= 出從加^水㈣影像中取 1四顯不洋水印取回處理的流程圖。 步驟1:藏有浮水印的影像之DCT轉換 印的气像Xw,首先將其分割成不重疊 象素區塊;再對每—個區塊個別執行Dc = 令Xw = DCT(Xw)。 寻供 步骤2:量化的係數之取回 所有旦過置換操作,可恢復原浮水印έ ^ 有里化的係數原始順序並令恢復的量化浮水二 步驟3:反轉量化操作和反DCT轉換 _一所轉可調整i化表Q,割 尺度適用 準(CNS ) 格 ______ (請先閱讀背面之注意事項再填寫本頁;&gt;I 4 4 ^ /, China Q (a, b) 疋 data item located in column a & line q of q · Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs' / ^, each scattered in the 4x4 block 12: Blocks are quantified. Then, step 5 is generated: the watermark is hidden in the selected coefficients. Each coefficient in the text mf is started by one of 宄 q, and the watermark hidden in the folder is a secret key KK3. The species U, which is the brain G, is replaced with a sequence in which the coefficient is replaced. Order later Step 6: Rewrite the selected IF coefficients of the original image [paper size applies fiiiijTcNS to non-overlapping ^ Order B7 Description of the invention? ) &Quot; ~~ The block is made up of 8x8 pixels, and then the spoon transform is applied to each block individually to sound: On the contrary, Xw = IDCT (iw ^ also in Xw, also the staff of the Intellectual Property Bureau of the Ministry of Economic Affairs, employee consumption Cooperative printed 532039 Five wv ^ 丄 v eight wy The copyright owner of the image will be able to adjust the quantization table '11 f 中 = adjustable G 2 registered and get the time stamp added to the arbitrator and perform a watermark retrieval process to verify Whether ίίίί 2. Watermark retrieval processing algorithm 仏 仲 ί When a dispute occurs, it is claimed that the secret key Kl, KA and time stamp saved by the owner of the property, Cai Liuyi, are taken from the image of Jiashui The flow chart of the process of retrieving the four-mark non-foreign watermarks. Step 1: The aerial image Xw of the DCT-converted image of the image with the watermark is first divided into non-overlapping pixel blocks; Execute Dc = Let Xw = DCT (Xw). Prospecting Step 2: Retrieve the quantized coefficients. After all replacement operations, the original watermark can be restored. ^ There are two steps in the original order of the coefficients and the restored quantized water. 3: Inverse quantization operation and inverse DCT conversion _ one turn can adjust the iization table Q, Scale quasi applicable (CNS) cell ______ (Please read the Notes on the back page and then fill; &gt;

532039 五、發明謀明 經濟部智慧財產局員工消費合作社印製 A7 B7 古執彳^反量化操作和反轉DCT轉換。首先對你 的所有每一個4x4區塊完成反量化操作而取得g。 接 '對於★中的每一個4x4區塊利用反轉DCT “换 以獲得對應的空間領域影像W;。 ^吳 步驟4:浮水印的反相的置換 最後’針對取回的浮水印W;透過PRNGO^)來執行 相的置換而恢復原來浮水印W'。經由仲裁者註冊的 W和取回的浮水印w,比較,能夠判定原始影像的擁 有權之歸屬。 本紙張尺度適用中國國家榡準(CNS ) A*規格(2丨〇&gt;&lt; 297公釐) (請先閑讀背面之注意事¾再填寫本頁}532039 V. Inventing and Printing Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs A7 B7 Ancient execution ^ Inverse quantization operation and inverse DCT conversion. First complete inverse quantization on all of your 4x4 blocks to get g. Then, for each 4x4 block in ★, use the inverted DCT "to obtain the corresponding spatial domain image W; ^ Wu Step 4: Inverse permutation of the floating watermark finally" for the retrieved watermark W; through PRNGO ^) to perform the phase replacement to restore the original watermark W '. After comparing the registered W and the retrieved watermark w, the arbiter can determine the ownership of the original image. This paper scale is applicable to China's national standards (CNS) A * specifications (2 丨 〇 &gt; &lt; 297 mm) (Please read the notes on the back ¾ before filling this page}

532039 2fl A7 B7 五 '發明説明((。) -- &lt;實驗結果與分析&gt; 本發明所選用的實驗平台為pentium ΠΙ 縣 Wind_ 2_ 场 且私式撰寫工具為Java 1·3。此外,本發明中亦 ί里lmp_輸影像處 巧取回,位浮水印的結果供參考外,亦分 ^ r ^ PSNR(Peak Sl§nal to Noise Ratios) M 巧CCNormahzed Correlation}比值作為評估標準,以評估藏乂 浮水印後的數位影像與取回的數位浮水印之效益。 NC分別定義於式子⑹與⑻中·· 1烈皿及 ps—.SldB, ⑹ 其中MSE代表均值平方錯誤(Mean Square Error),對於一個 NxN影像而言,其MSEi定義如(7)所示, 、 MSE:532039 2fl A7 B7 Five 'invention description ((.)-&Lt; Experimental results and analysis &gt; The experimental platform used in the present invention is Wind_ 2_ field of pentium ΠΙ county and the private writing tool is Java 1.3. In addition, this In the invention, the lmp_ input image is retrieved cleverly, and the result of the floating watermark is for reference. It is also divided into ^ r ^ PSNR (Peak Sl§nal to Noise Ratios) M CCCC Normahzed Correlation} ratio as an evaluation standard to evaluate The digital image of the Tibetan Mastiff watermark and the benefits of the retrieved digital watermark. NC is defined in the formula ⑹ and ⑻ in the formula · 1 strong dish and ps-.SldB, ⑹ where MSE stands for Mean Square Error ), For an NxN image, its MSEi definition is shown in (7),, MSE:

N 1 i=N-lj=N-l Σ Σ (x(ij)-xw(ij))2 i=0 i=〇 ⑺ 其中x(i,j)和Xw(i,j)分別表示原影像和浮水印影像的第(y)個 像素值。 ’ 經濟部智慧財產局員工消費合作社印製N 1 i = N-lj = Nl Σ Σ (x (ij) -xw (ij)) 2 i = 0 i = 〇⑺ where x (i, j) and Xw (i, j) respectively represent the original image and the floating image The (y) th pixel value of the watermark image. ’Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs

NC M-l M-1 Σ Zw(i,j)w'(i,j) i=0 j=〇 2 ⑻ 其中,針對NxN浮水印影像,W(i,j)和W(i,j)分別表示浮水 印和取回之浮水印影像分別的第(i,j)個像素值。 在本發明的實驗,使用一張有512x512像素的灰階影像 Lena當作原始影像,而另一張有128x128像素灰階影像冬 為^水印影像(分別顯示在圖五⑻和(b))。經對本發明的浮^ I藏^入_呈受處理以後’浮水印影像的影像品質值PSNR^ 本紙張尺度適用中國國家標準(CNS ) A4规抵公格 (請先閲讀背面之注意事項再填寫本頁)NC Ml M-1 Σ Zw (i, j) w '(i, j) i = 0 j = 〇2 ⑻ For the NxN watermark image, W (i, j) and W (i, j) represent The (i, j) th pixel value of the watermark and the retrieved watermark image respectively. In the experiment of the present invention, one grayscale image Lena with 512x512 pixels is used as the original image, and the other grayscale image with 128x128 pixels is a watermark image in winter (shown in Figure 5 and (b), respectively). The image quality value PSNR of the floating watermark image after the floating ^ I hide ^ into the processed_presentation of the present invention ^ This paper size is applicable to China National Standard (CNS) A4 regulations (please read the precautions on the back before filling in (This page)

明説明((丨) 一::…3-- 外·65 dB(顯示在圖五(c)),對於 言,提供了很好的浮水印影像:見糸統的辨視能力济 基本要求。除此之外,、足目前 回的浮水印與原始浮水印之間^^^于水印取回程序所承 值衡量的結果為0.99,其值相告了者,為相似。以NC 當浮水印影像如受到焱壞攻纂夂1丨J員示在圖五(d》。 J生的信號處理,本發明的影像4 么G 3真壓縮和破f 仍然能強章刃的取回藏入浮水^尺印方法經由下列實驗證實 一、抵抗攻擊後的效果 央冲ΐΪί水印隨機置換和隨機的選擇屌妒旦彡偾Φ栺在勣 有應用隨機置換也沒有使用隨播、,予水印藏入時,既沒Explanation ((丨) 1 :: ... 3--outer · 65 dB (shown in Figure 5 (c)), for speech, provides a good watermark image: see the basic requirements of the discrimination ability of the system. In addition, between the current watermark and the original watermark ^^^ The result measured by the watermark retrieval program is 0.99, and the value is reported to those who are similar. Use NC as the watermark. If the image is damaged, the image is shown in Figure 5 (d). J's signal processing, the image of the present invention, G 3, true compression and breaking f can still be retrieved and hidden in the floating water. ^ The footprint method is confirmed by the following experiments: 1. The effect after resistance to attack. Ϊ́Ϊ Watermark random replacement and random selection. 屌 彡 偾 彡 偾 栺 栺 栺 When random replacement is used, no companion is used. When the watermark is hidden, Neither

??有Λ用隨機代替,結果顯示所T回的V 置換ϋ顧示右始沒有應用隨機代替但使用隨;f 時,有^用ί真現象發生。圖六(e)*藏入過弄 0.25, 〇^7S:^ ^ 回的浮水印仍狹於由人足久P J昃攻4 ’貫驗結果顯示所耳 水印々類的眼睛識別;即示使用本發明翻 艺下’仍可取回可_ 示Hsu和取,f水印的NC值。圖八顯 裁f攻擊下的相同的大小浮水印後各種比例 Hsu和術,的方°2更$較貫驗結果證明了本發明的強拿刃性比 I- -i I 11- ! , I- -i * --裝---- (請先閱讀背面之注意事項再填寫本頁) f li 1— ·?? There is a random replacement of Λ, and the results show that the replacement of V by T times. No random replacement is applied but the random f is used. Figure 6 (e) * The hidden watermark of 0.25, 〇 ^ 7S: ^ ^ is still narrower than that of the person's foot PJ attack 4 'The result of the test shows that the watermark of the ear is recognized by the eye; According to the present invention, the NC value of Hsu and f, f watermark can still be retrieved. Fig. 8 shows the same size watermark under various attacks under the f attack, and the square H2 is more square. The more consistent results prove that the strong grip of the present invention is better than I- -i I 11-!, I --i * --install ---- (Please read the precautions on the back before filling this page) f li 1— ·

In 丨:.=I In HI . •^! 經濟部智慧財產局員工消費合作社印製 .Ϊ --1 ! - . 532039 五 、發明説明(/:! A7 B7 經濟部智慧財產局員工消費合作社印說 水印的影像受到·壓縮攻擊後,對取 JPFP圖示在各種壓縮比例下的實驗結果◦表-%、明a 日寺相^比例^^,NC值也隨之變小 表一、二和四貫驗結果,可歸納出在一船旦彡德&amp;古 ,,下,灰_浮水印比這二元浮水印在更^;:像失真 ―、抵抗^種破壞性信號的影像攻擊之效果 你&gt;7驗將探討當浮水印的影像分別經由p模糊化&amp; 銳化破_,對取回浮水印的影響。 ^婦化和 (一) 模糊化攻擊實驗 a 示藏有浮水印的Lena在—次模糊化攻擊取 次模糊化攻擊後取回的浮水印。表五顯示當模 ^,數攻#下本發明仍可取回可辯識的浮水印。 、 (二) 銳化攻擊實驗 f 贫別顯示浮水印影像Lena在一次銳化攻擊I 5次銳 ΐϊίΐϊ回的浮水印、结果。由於*影像銳化過程士,透過 f ΐίϊ—像ΐ度有增加趨勢,使得灰階的像素值在 ίίί正ί4有升南現象。因此,會造成Nc值出現大於1 象。表六說明在合理次數銳化攻擊後仍可取回可 洋水印。 ^張尺度適财_^T^_S ) Μ規格(2ΗΓΧ297公釐 (請先閲讀背面之注意事項再填寫本頁)In 丨:. = I In HI. • ^! Printed by the Employees 'Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs. Ϊ --1!-. 532039 5. Invention Description (/ :! A7 B7 Printed by the Employees' Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs. It is said that after the watermark image is subjected to compression attacks, the experimental results of JPFP graphs at various compression ratios are taken. Table-%, Minga Ri Temple Phase ^ ratio ^^, and the NC value also decreases. Tables 1, 2 and The four-pass test result can be summarized in a ship's watermark and the ancient, lower, and gray watermarks are more than these binary watermarks: like distortion-resisting image attacks of ^ destructive signals The effect of your &gt; 7 test will explore the effect of when the watermark image is blurred and sharpened by p, respectively, on the retrieval of the watermark. ^ Women and (1) Blur Attack Experiment a shows that the watermark is hidden Lena retrieved the watermark after the fuzzification attack and the fuzzification attack. Table 5 shows that the present invention can still retrieve the identifiable watermark under the mode ^ , 数 攻 #. (2) Sharpening attack Experiment f: Do not show the watermark image of Lena during a sharpening attack I sharpen the watermark and result of 5 times. Due to * In the process of image sharpening, through f ϊίϊ—the degree of image has an increasing trend, so that the gray-scale pixel value has a rising phenomenon in ίίίίί4. Therefore, the Nc value will be greater than 1 phenomenon. Table 6 shows that the reasonable number of times Watermarks can still be retrieved after a sharpening attack. ^ 张 码 量 可 财 _ ^ T ^ _S) M size (2ΗΓ × 297 mm (please read the precautions on the back before filling this page)

532039 經濟部智慧財產局員工消費合作社印製 圖式532039 Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs

^ 丄々 &lt; 圖式說明〉、 之數位浮水印藏入流程圖 圖-一 •太發明夕Q wO rrr ·,, I於n ,一 q匀、1τ脚往圚0之8X8區塊中低頻位置示意圖。月之過遽表Tw(flltertable)。 本f明之數位浮水印取回流程圖。NC:〇g ⑹取回之數位浮水印, 圖六水印的Lena影像遭切割1/2處理,(b) 』任何處理,⑷僅做_藏入之i圖七.處理與亂數藏入,取回之數位ί ΪΪ Lena影像遭切割處理之實驗(a)切iP (b)切割5〇% ’⑹切割75%,取回之數位浮 圖八.·'影像切割之實驗結果比較。圖九.^婁^^印的Lena.影像經jpEG失真壓縮處理之 ⑻屋&amp;比6 ’(b)壓縮比8.1,(C)壓縮比以丨,⑹ i=、、、侣比16&gt;,⑹壓縮比19,取回之數位浮水印。 回十·藏有數+位浮水印的Lena影像經模糊化處理之實驗 -人模糊化處理,(b)五次模糊化處理,取回之數位浮 水印。 ^ 固十·藏入數位浮水印的Lena影像經銳化處理之實驗⑻ 一次銳化處理,(b)五次銳化處理,取回之數位浮水印〇 圖三 圖四 圖五^ 丄 々 &lt; Schematic description>, digital watermark hiding flow chart-I • Tai Qin Xi Q wO rrr · ,, I in n, one q uniform, 1τ foot to 圚 0 8X8 block mid-low frequency Location diagram. Tw (flltertable) in the moon. This figure shows the digital watermark retrieval flowchart. NC: 〇g ⑹Retrieved digital watermark, Fig. 6 The watermarked Lena image is cut and processed by 1/2, (b) 『Any processing, 做 only do _ hiding i Figure 7. Processing and random number hiding, Digitally retrieved ί na Lena images were cut. Experiments (a) cut iP (b) cut 50% '⑹ cut 75%, retrieved digital floating image VIII. ·' Comparison of experimental results of image cut. Figure IX. Lou ^^ printed by Lena. The image was subjected to jpEG distortion compression processing and the ratio 6 '(b) compression ratio 8.1, (C) the compression ratio with 丨, ⑹ i = ,,, and ratio 16> , ⑹ compression ratio of 19, to retrieve the digital watermark. Back to ten · Experiment of blurring Lena image with digital + bit watermark-Human blurring process, (b) Five times blurring process, retrieve the digital watermark. ^ Gu Shi · Experiment on sharpening of Lena image with digital watermark embedded⑻ One sharpening process, (b) Five sharpening processes, retrieved digital watermarks. Figure 3 Figure 4 Figure 5

A4規格(210X297公釐) (請先閎讀背面之注意事項再行繪製) -訂-A4 size (210X297 mm) (Please read the notes on the back before drawing) -Order-

Claims (1)

532039 步驟, 經濟部智慧財產局員工消費合作社印製 其中灰 A8 B8 C8 D8 六、申請專利範圍 〈申請專利範圍&gt; 1H用離散餘弦轉換的原理與數位量化方法來藏入灰 P白=士印於欲,護的灰階原始影像之中的新型數位浮水 』於法,包含藏入及取回灰階浮水印兩大部份;其特徵 灰階浮水印的藏入方法 步驟一·針對欲保護的灰階原始影像選擇一張 作權資訊的灰階浮水印; 〃 •浮水印的藏入演算法以產生—張藏入 净水印的洋水印影像; 步驟三· j巧生財浮水印的浮水特#數位量化 表向公正的公正的認證中心註冊。然後,合法 ^有者就可以將浮水印影像在纟_上公佈及傳 灰階浮水印的取回之方法 步驟一 .ΐϋίΡ印影像、註冊過的數位量化表以及灰 步驟二.sssr以取得藏入於浮水 第1項所述—種新型數位浮水印方法,並 水印概保護的灰階原始影 整的量化▲來藏二1 印的 黑你,L m A 1_、~ ^ ..張藏入灰1¾、手水 本紙張尺度適用中國(2ΐ〇χ297公釐·· (請先閱讀背面之注意事項再填寫本頁)Step 532039, printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs, which is gray A8 B8 C8 D8. 6. Patent application scope <Scope of patent application> 1H uses the principle of discrete cosine transformation and digital quantization method to hide gray P white = Shiyin Yu Yu, a new type of digital floating water in the original gray-scale image of protection "Yu Fa includes two parts: hiding and retrieving the gray-scale watermark; its feature, the first step of hiding the gray-scale watermark. Select a gray-scale watermark for the original grayscale image; 〃 • The watermarking algorithm to generate-a watermarked image of the ocean watermark hidden in the watermark; Step 3 · Qiaoshengcai floating watermark's floating water feature # The digital quantification table is registered with an impartial certification center. Then, the lawful person can publish the watermark image on 纟 _ and transfer the gray-scale watermark to the method. Step 1. Print the image, the registered digital quantization table, and step 2. .sssr to obtain Introduced in Item 1 of Floating Water-A new type of digital watermarking method, and quantization of grayscale original shadowing that is protected by watermarking. ▲ Let's hide the 1st black ink, L m A 1_, ~ ^ .. Gray 1¾, the size of this paper is suitable for China (2ΐχχ297mm ... (Please read the precautions on the back before filling in this page) if 申請專利範園 [灰階浮水印的藏入之演算法] 輸入·灰卩自原始影像X與灰階 NxN個像素組成,灰階、享. ’八中原始影像X由 輪出:藏有縣印影像Xw素組成; ' Q與比重表Tw ; '句K2, K3、數位量化表 步驟1 : 始影像X成不重疊的8x8區塊,並且把每一 步驟2 :ϊϊΐίί,DCT轉軸時取得; 種子,並針ίίΪ—Kl作為隨機數產生器(PRNG)的 隨機=⑽擇 ,令所綱娜數組成 訂 Μ X Μ (Ν /8)χ(ν/8) 步驟當膽g的種子,同時執行 士的母—個像產生—個置換位置用 半炉4.八4 'ίΐ?所有?素並產生置換後的結果Wr; 經濟部智慧財產局員工消費合作社印製 乂〃·刀^告r成為不重疊的4x4區塊,且個別jQCT轉換 令心DCTW,然後,指定DCT轉 換區鬼中的母一個係數對應一個比重值並得到一個 4x4比重士 Tw’接著’由等式(4)計算文时所有係數 的平均,並利用等式(5)產生—個可調整量化 J Q :最後,透過量化表Q對每一個4χ4區塊進行 i化操作,令產生的數位量化後的浮水印為也 1If applied for a patent Fan Yuan [Greyscale floating watermark hiding algorithm] Input · Grayscale is composed of the original image X and grayscale NxN pixels, grayscale, enjoyable. 'The eight original images X are rotated out: hidden County print image Xw element composition; 'Q and specific gravity table Tw;' sentence K2, K3, digital quantization table Step 1: Start image X into non-overlapping 8x8 blocks, and each step 2: ϊϊΐίί, DCT obtained when the axis is rotated Seeds, and ίίΪ—Kl as a random number generator (PRNG) of random = ⑽choice, so that the number of constituents constitutes the seed of the order Μ X Μ (N / 8) χ (ν / 8), when the gall g Simultaneous execution of the master's mother-an image generation-a replacement position with a half furnace 4. 8 4 'ίΐ? All? The result of the replacement is Wr; the consumer cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs prints a 刀 · blade ^ report r into a non-overlapping 4x4 block, and individual jQCT conversions make DCTW, and then specify the DCT conversion area. A coefficient of the parent corresponds to a specific gravity value and gets a 4x4 specific gravity Tw 'then' calculates the average of all coefficients in the text from equation (4), and uses equation (5) to generate an adjustable quantified JQ: Finally, through The quantization table Q performs an iization operation on each 4 × 4 block, so that the digitally quantized watermark is also 1 本紙張尺度適财家;fe準(CNS ) Α4_ ( 2ΐ()χ29»Ι) ^2039 A8 B8 C8 D8 申請專利範圍 Μ -1 Μ -ι— Σ Σ x f π η Xmf = 5j) Μ χ Jy[ (4) Μ Μ --1--1 4 4 Q(a,b) = RE^^hh + 4^ γ ^ ν Μ , Τ 4 xTw(a5b) 其中Q(a,b)疋位吻中第a列第b行的資料項 (5) 缝濟部智慧財產局員工消費合作社印製 Μ 擇^ 一個秘密鑰匙&amp;作為pRNG的種子來產生 1中的1有係數被替代的序列並根據此序列將免q :的=個量化後係數取代在戈mf中的對應的一個係 步驟被代替後的結果為九^ ; 〜2 Wmf,中的每—個係數改寫到相對應的一個文之係 步驟7.得到浮水印藏人後的影像(; ,λ ·=剔w成為不重疊的8χ8區塊,並且以反DCT轉換 母一個區塊並獲得藏有w的灰階影像Xw ; [灰階浮水印的取回演算法] 輪入識有浮水印的影像Xw、秘密输匙Κι, κ2, Κ3與數位量 化表Q ; 輪出:取回的灰階浮水印W ; 步驟1:ίΓ=ΐϊ水印的影像Xw,首先將其分割成不重疊 .上8 8 £塊,再對每一個區塊個別執行DCT轉換; 令 XW=DCT(XW); 寸、 乂驟戸臟机)從、的每一個8x8區塊内 的1化係數,接下來,透過pRNG(K3)對所有得^的 (請先閲讀背面之注意事項再填寫本頁)The size of this paper is suitable for wealth collectors; fe quasi (CNS) Α4_ (2ΐ () χ29 »Ι) ^ 2039 A8 B8 C8 D8 The scope of patent application Μ -1 Μ -ι— Σ Σ xf π η Xmf = 5j) Μ χ Jy [ (4) Μ Μ --1--1 4 4 Q (a, b) = RE ^^ hh + 4 ^ γ ^ ν Μ, Τ 4 xTw (a5b) where Q (a, b) is the highest Data item in column a, line b. (5) Printed by the Consumers ’Cooperative of the Intellectual Property Bureau of the Ministry of Secrecy. Select a secret key &amp; as the seed of pRNG to generate a sequence of 1 in which coefficients are replaced and according to this sequence Replacing q: = quantized coefficients in the corresponding one of the steps in the mf is replaced with a result of nine ^; ~ 2 Wmf, each of the coefficients is rewritten to the corresponding one of the steps in the text 7. Obtain the image after the watermark is hidden (;, λ · = tick w becomes a non-overlapping 8 × 8 block, and transform a parent block with inverse DCT and obtain a grayscale image Xw with w; Algorithm for retrieving watermark] Turn in the image Xw with the watermark, the secret key κ, κ2, Κ3, and the digital quantization table Q; Round out: the retrieved grayscale watermark W; Step 1: ίΓ = ΐϊwatermark Image Xw, first divide it Make no overlap. On 8 8 £ blocks, perform DCT conversion on each block individually; Let XW = DCT (XW); Inch, 乂 戸 戸 戸 戸) from each of the 8x8 blocks, the unity coefficient Next, use pRNG (K3) to get all the ^ (please read the precautions on the back before filling this page) 532039 A8 B8 C8 D8 六、申請專利範圍 量化係數執行反相的置換,用來恢復原浮水印的所 有量化的係數之原始順序並令恢復的量化浮水印為 Wmf ; 步驟3:利用可調整量化表Q對古mf執行反量化操作和反 DCT轉換以獲得對應的空間領域影像W;; 步驟4:針對W;而言,透過PRNGO^)執行反相置換並得到 取回的浮水印W、 (請先閱讀背面之注意事項再填寫本頁) 、v 一口 Γ 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(210X 297公釐)532039 A8 B8 C8 D8 VI. Apply patented range quantization coefficients to perform inverse permutation to restore the original order of all quantized coefficients of the original watermark and make the recovered quantized watermarks to Wmf; Step 3: Use adjustable quantization table Q performs inverse quantization and inverse DCT conversion on the ancient mf to obtain the corresponding spatial domain image W; Step 4: For W; in terms of PRNGO ^), perform inverse permutation and obtain the retrieved watermark W, (please Read the notes on the back before filling in this page), v sip Γ The paper printed by the Consumers' Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs applies the Chinese National Standard (CNS) A4 specification (210X 297 mm)
TW90100412A 2001-01-05 2001-01-05 Robust gray-level watermark embedding and retrieving technique TW532039B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
TW90100412A TW532039B (en) 2001-01-05 2001-01-05 Robust gray-level watermark embedding and retrieving technique

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
TW90100412A TW532039B (en) 2001-01-05 2001-01-05 Robust gray-level watermark embedding and retrieving technique

Publications (1)

Publication Number Publication Date
TW532039B true TW532039B (en) 2003-05-11

Family

ID=28787702

Family Applications (1)

Application Number Title Priority Date Filing Date
TW90100412A TW532039B (en) 2001-01-05 2001-01-05 Robust gray-level watermark embedding and retrieving technique

Country Status (1)

Country Link
TW (1) TW532039B (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8144923B2 (en) 2005-12-05 2012-03-27 Thomson Licensing Watermarking encoded content
US9875365B2 (en) 2008-05-29 2018-01-23 Hewlett-Packard Development Company, L.P. Providing authenticated communications to a replaceable printer component
TWI820916B (en) * 2022-09-16 2023-11-01 友達光電股份有限公司 Watermark embedding method and display device

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8144923B2 (en) 2005-12-05 2012-03-27 Thomson Licensing Watermarking encoded content
US9875365B2 (en) 2008-05-29 2018-01-23 Hewlett-Packard Development Company, L.P. Providing authenticated communications to a replaceable printer component
TWI820916B (en) * 2022-09-16 2023-11-01 友達光電股份有限公司 Watermark embedding method and display device

Similar Documents

Publication Publication Date Title
Makbol et al. A new robust and secure digital image watermarking scheme based on the integer wavelet transform and singular value decomposition
Sverdlov et al. Robust DCT-SVD domain image watermarking for copyright protection: embedding data in all frequencies
Bhatnagar et al. A new robust reference watermarking scheme based on DWT-SVD
Lee et al. A public verifiable copy protection technique for still images
Cheddad et al. A secure and improved self-embedding algorithm to combat digital document forgery
Wu et al. Digital watermarking based on chaotic map and reference register
Luo et al. Blind image watermarking based on discrete fractional random transform and subsampling
Wang et al. A secure dual-color image watermarking scheme based 2D DWT, SVD and Chaotic map
Zhang et al. Image watermarking scheme based on Arnold transform and DWT-DCT-SVD
Patra et al. Crt-based fragile self-recovery watermarking scheme for image authentication and recovery
Singh et al. A high capacity image adaptive watermarking scheme with radial harmonic Fourier moments
Chang et al. Preserving robustness and removability for digital watermarks using subsampling and difference correlation
Mohan et al. On the use of Schur decomposition for copyright protection of digital images
CN102930498B (en) A kind of digital watermark embedding based on Lifting Wavelet and extracting method
Huang et al. A robust watermarking technique for copyright protection using discrete wavelet transform
Ye et al. A multiple-level DCT based robust DWT-SVD watermark method
TW532039B (en) Robust gray-level watermark embedding and retrieving technique
Jha et al. Dynamic stochastic resonance-based grayscale logo extraction in hybrid SVD-DCT domain
Rao et al. A novel image zero-watermarking scheme based on DWT-BN-SVD
Singh et al. A robust blind watermarking framework based on Dn structure
Ma et al. A novel blind grayscale watermark algorithm based on SVD
Chen et al. A robust wavelet-based watermarking scheme using quantization and human visual system model
Wang et al. Review on digital image watermarking based on singular value decomposition
Chang et al. A multipurpose wavelet-based image watermarking
ZEEBAREE et al. COPYRIGHT PROTECTION SYSTEM BASED WATERMARKING

Legal Events

Date Code Title Description
GD4A Issue of patent certificate for granted invention patent