玖、發明說明： 【發明所屬之技術領城】 發明領域 本發明係有關於文件的水印技術，雖非排它的，但特 別係指在一文件中來隱設光學水印的系統。 【先前技術3 發明背景 數位水印係為古老的潛隱顯訊技術之一種現代化形 式，而被用來將資訊隱藏在其它的資訊(影像、文件、音頻 和視訊貢料等）中。一水印即在數位化或列印媒體中之被隱 藏的訊息。由於數位化媒體係與該資料檔本身的内容整合 在一起，故並不需要另外的儲存。 水印有兩麵型：可見构和不可脉印。可見水印係為可看見 的半透明影像，其會重4在—主要影壯。其通料、由财該主要影 像之權利的機構之封印或標誌所構成。一可見水印可使該主要影像能 被看到’但亦能清楚地示出該文件係屬於某機構的財產。一不可見的 水印係為1疊的信號(媒體），其不能被翻，但可被運算顺出。 數位水印曾被用來作為該數位資料之所有者與配銷者 的辨認手段。獨特的代碼會被埋人—數位㈣巾，而該等 代碼可被供用於各種用途，例如日期/時間標示，來源識 別，内容辨認、處理狀況，所有權辨識等等。 光學水印與傳統的水印之不同處在於，該水印係使用 光學及視覺方法例如影印㈣取。此乃能以視覺來確認一 文件係為原本或已被影印或掃描者。 200406717 C發明内容3 發明概要 本發明之第一較佳態樣係提供一種可列印的水印影 像’其可被隱没於供列印的文件中。該影像包含一影像背 5景具有第一顏色的平均強度；及一指示區亦具有該第一顏 色的平均強度。至少该指示區的強度係為至少兩種不同強 度之頻振配置造成的結果。該水印影像的解析度，該頻振 配置，及該至少二強度係被選成，能使該影像背景與該指 示區的顏色和強度，會對肉眼呈顯為一致的，但在以一如 1〇低通濾波器地操作之複製裝置來處理該列印影像之後，該 指示區會變成令肉眼明顯可見的。 該影像背景的平均強度亦可為二或更多不同強度之頻 振配置（dithered arrangement)造成的結果。該指示區之二或 更多的強度I請，會大於該影像背景之二或更多的強度範 圍忒衫像月景與該指示區可為均一顏色。該水印影像可 包含-可見的標記區，其係與該均一顏色為不同的顏色， 或具有與該影像背景和該指示區不同的區域平均強度。 該水印影像可包含至少三個區，各區皆具有一選自一 組至少兩種顏色之第一顏色。該至少三個區可包含一第一 2〇區約。為該水印表面積的15%，一中間區約為該水印表面積 、 及第一區約為該水印表面積的15%。該第一和第 二區可為相同顏色。該中間區介設其間，而可為不同顏色。 在各區之間的邊界可由一變化圖案來凸顯。 於本發明之一第二較佳態樣中，乃提供一種能產生一 6 200406717 可列印之水印影像來結合於-可列印文件令的方法“亥方 法包含以下步驟：選擇-影像背景顏色和強度；以軸色 及-等於該所《度的平均強度來填滿該水印影像的背景 區，並使各像元的強度落在一背景強度範圍内。 5 一指示區會被以該所擇顏色與一等於該所擇強度之平 均強度來填滿，但使錢福強度落在_比上述第^圍 更大的範圍内，而令該影像解析度，該頻振配置，及該二 或更多的強度被選成，能使該影像背景與該指示區的顏色 和強度會對肉眼呈顯為-致的，但在以一如低通渡波器地 10操作之複製裝置來處理該列印影像之後，該指示區會令肉 眼明顯可見的。 填滿該背景的步驟可包含施加一具有第一差之二強度 的頻振圖案。而填滿該指示區的步驟可包含施加一具有第 一差之二強度的頻振圖案，該第二差係大於第一差。該所 15擇的強度可小於RGB(239,239,239)。 旦其亦可包含以下步驟··接收一基本影像；及依據該背 ’7、區和私不區之間的顏色或強度差異，而來識別該基本影 像中之一背景區與一指示區。 /、亦可包g以下步驟··辨認該基本影像内之一可見的 20 l己區；及以-不同於該第一顏色和強度的第二顏色和強 度來填滿該可見的標記區。 其又可更包含以下步驟··接收一可見標記基本影像及 —指示基本影像；辨認該指示基本影像中的指示區及背景 區’亚將之填滿。一可見標記區亦會被辨認來作為該可見 7 標記基本影像中之-背景區。該可見標記基本影像與該被 填滿及處理後的指示基本影像乃可被結合，而使各對應像 凡在所述之-該影像令的像元顏色和強度，會超過在另一 影像中的像元顏色和強度。在該可見標記影像之背景區中 5的像元係對該已處理之指示基本影像中的像元具有較低的 超差。而在該可見標記基本影像中之可見標記區内的像元 會對该已處理之指示基本影像中的像元具有較高的超差。 在本發明之另-較佳態樣中，係提供一種能產生一可 列印的水印影像來結合於一可列印文件中的系統。該系統 10包含：一裝置可選擇一影像背景顏色和強度；一裝置能以 4顏色及f於所擇強度的平均強度來填滿該水印影像的 月景區，亚使各像元的強度落在一背景強度範圍内；及一 裝置能以該所擇顏色及一等於所擇強度的平均強度來填滿 私不區，但使各像元的強度落在一比前述強度更大的範 15圍内，而令該影像的解析度，該頻振配置，及該二或更多 強度被選成’能使該影像背景與該指示區的顏色和強度， 會對肉眼呈顯為-致的，但在以一如低通濾波器地操作之 複製I置來處理該印列影像之後，該指示區會變成令肉眼 明顯可見的。 用來填滿該背景的裝置乃包含可施加一具有第一差之 二強度的頻振圖案之裝置。而用來填滿該指示區的裝置則 包含可施加一具有第二差之二強度的頻振圖案之裝置，該 第二差係大於第一差。該所擇的強度可不小於 RGB(239,239,239)。 8 200406717 其亦可包含一裝置可接收一基本影像；及一裝置可依 據該背景區與指示區之間的顏色或強度差異，而來識別該 基本影像中之一背景區及一指不區。 其最後可包含一用來辨認該基本影像中之一可見標記 5 區的裝置；及一裝置能以一不同於第一顏色和強度的第二 顏色和強度來填滿該可見標記區。 其最後亦可包含一裝置用來接收一可見標記基本影像 及一指示基本影像；一裝置可辨認該指示基本影像中的指 示區和背景區，並將之填滿；一裝置可辨認該可見標記基 10 本影像中之一可見標記區與一背景區；及一裝置可結合該 可見標記基本影像和該被填滿並處理過的指示基本影像， 而使各對應像元在其一影像中的像元顏色和強度超過在另 一影像中的像元顏色和強度，且在該可見標記影像之背景 區中的像元會對在已處理之指示基本影像中的像元具有較 15 低的超差’而在該可見標記基本影像之可見標記區中的像 元對在已處理之指示基本影像中的像元具有較高的超差。 在本發明之倒數第二個態樣中，乃提供一種由上述方 法所產生的水印影像；及一文件設有由上述方法所製成的 水印影像。 20 在本發明的最後一個態樣中，乃提供一種電腦可用的 媒體，其包含一電腦程式碼可令一處理器來執行一或多項 功能，而使上述方法能被進行。 圖式簡單說明 為使本發明能被容易瞭解並付諸實施，現將藉由本發 9 200406717 明之非限制性的較佳實施例來說明，該說明係參照所附圖 式來描述，其中： 第1圖為一灰階水印的原始圖案； 第2圖為一彩色水印的原始公司名稱影像； 5 第3圖為一彩色水印的複製字樣影像； 第4圖為供成型一水印的圖案影像； 第5圖示出一複製標記之點的初始圖案； 第6圖示出初始的水印影像； 第7圖示出伸展之後的初始水印影像； 10 第8圖示出控制強度之後的水印； 第9圖示出雜訊被施加於整體背景之後的水印； 第10圖為本發明產生水印之較佳過程的方塊圖； 第11圖示出本發明的像元伸展範圍； 第12圖為第10圖之水印產生程序的流程圖； 15 第13圖為第10圖之水印埋設程序的流程圖； 第14圖示出一具有水印的文件； 第15圖示第14圖的文件在影印機的濃度太淡時的影印 結果；及 第16圖示出第14圖的文件在影印機的濃度太暗時的影 20 印結果。 【實方包方式]3 較佳實施例之詳細說明 本發明係有關於複製保護性列印或數位文件的技術。 本發明能使在一列印文件中的影像或標記被隱設成，當該 10 200406717 文件該列印時，該影像或標記不會被看到。而當該文件被 影印或掃描時，該影像或標記即變成可見的。 在數位或列印文件上的水印係為一種系統，其可控制 以光學水印方法來列印之文件的複製。該系統包含一防複 5 製光學水印和防複製背景。有兩種水印可被使用：一種灰 階水印，及一種彩色水印。 該系統包含一水印產生器，及一水印可被埋設來控制 以光學水印法列印之文件的複製。 在一複製控制程序中，該數位文件的所有人或被授權 10 使用者會使用一標誌（圖案影像），其會結合某些著作權字樣 而來顯示該文件的所有權。在影印該水印文件之後，該影 印文件將會示出明顯的字樣(例如”複製”），即表示該文件已 被複製、修整、或竄改等等。當與原始水印文件之受控的 列印結合，將能形成一種可供鑑定數位及紙張文件的可靠 15 糸統。 如第12圖所示，欲造成一灰階水印時，該水印產生器 會接收一作為輸入的影像1201。該水印產生器會在步驟 1202由第1圖所見的圖案影像來擷取各特徵部份。被該水印 產生器所擷取的特徵部份乃包括原始背景103，公司字樣 20 101，及複製字樣102等。該公司字樣101的顏色應被選成， 當影印之後，其會完全或部份地消失。為能達到如此，故 該公司字樣最好為白色。又最好在該圖案影像中，該公司 字樣101和複製字樣102係為相同顏色（白色）。該複製字樣的 切始水印圖案會被產生，而包含很多小點和背景互相交 11 200406717 疊。該等小點107全為相同尺寸，且在該水印圖案中之各小 - 點的間距會保持相同。此可見於第5圖中。 該灰階水印和彩色水印之小點的大小和顏色皆會被控 制成，在影印之後，該等小點可被清楚地重製及分辨。該 5 水印的控制係藉調制，此將會詳細說明於後。該等小點不 可太小而致在影印之後不能被重製；但亦不可太大而令它 們能在水印中被肉眼看出。 該原始圖案的水印影像之尺寸（寬X高）對該文件影像 _ 的比例，對該等小點在影印之後的可見度是一項重要的因 10 素。該原始圖案影像的尺寸係被選成，當該水印伸展並埋 入該文件中時，該等小點會足夠地小，但在影印之後能被 重製及分辨。例如，若所擇的原始圖案影像係為560x650， 則用來埋設的原始文件尺寸係為3000點x3500點。其它的尺 寸亦可被使用，只要該等小點能夠清楚並可被重製。 15 伸展率（即該水印分別沿X和y方向伸展的倍數）對其控 制該等小點最後的間隔是為一重要的因素。若該水印由原 _ 始尺寸伸展4x4倍，且在控制強度之後僅留下一小點，則各 點之間距將成為原始小點尺寸的3倍。該等間距即足以在影 印之後來分辨該等小點。 20 該公司字樣103並不會在該產生過程1203中改變，且該 複印字樣102會保持該點圖案。該水印背景109則會被選 成，當該列印影像被複印之後，該背景會成為灰色。由該 水印產生器所擇的複製字樣背景108將會比整體的背景109 更淡一些。在步驟1204時，該水印產生器將會伸展該初始 12 200406717 的水印。該初始水印會分別沿x及y方向來伸展。在一例中， — 若一原始像元為一點，則在該水印以mxn的伸展率來擴伸 之後，該原始的小點將會變成mxn倍的大小。此係可在第 11A圖看出，其中僅有（0,0)的位置被填滿。在第lib圖中， 5 該像元已以4倍的伸展率來擴展，故在（0,0)、（0，1)、（0,2)、 (0,3)、（1，0)、（1，1)、（1，2)、（1，3)、（2,0)、（2，1)、（2,2)、（2,3)、 (3,0)、（3，1)、（3,2)、（3,3)等各位置皆會被填滿。 伸展該初始水印的公式係為： Φ (nxX0+(i-l)，mxY0+(j-l))。 10 其中（Χο,Υο)係為點的切始位置，m與η為擴伸的伸展 率，而i及j為循環變數。 其如何應用可見如下模擬代碼： 對（i=0 ; i<m ; i++) (j ~0 » j ^〇 y j ^而吕， 15 則新點位置為（nxXG+(i-l)，mxY()+(j-l))二 1 該擴伸的水印可見於第7圖中。 鲁 該水印嗣會在步驟1205中被調制。調制係被用來減少 該水印的強度，而使該等小點的可見度最小化。由原始點 至擴伸點的距離會被算出。若該距離超過一所擇的基準 20 值，則該像元會被控制。調制係可用來獲得一良好的視覺 效果。調制的距離係使用以下公式來計算： 距離=^J(x-X0y 其中（Χ〇，Υ〇)係為原始點位置，而（x，y)為伸展位置。 若用來控制一點的基準值係被選為：基準值^1，則每 13 一擴伸的大點將會僅有一點被留下。此可見於第11C圖中。 假使在（0,0)位置的點被擴伸4x4倍，則在第π圖中之 (〇,〇)、(0，1)、(〇,2)、(0,3)、（1，〇)、（1，1)、（1，2)、（1，3)、(2,〇)、 (2，1)、（2,2)、（2,3)、（3,0)、（3，1)、（3,2)、（3,3)各位置的距 離值’將分別為〇，1，2，3，1，1.414，2.236，3.1623，2.236, 2.828, 3.6056, 3,3·1623，3.6056, 4.2426等。若該基準值係選 為：3>基準值>2 ;則在控制強度之後，僅有在（丨，2)，（2，1)， (2,2)位置的點會被留下。所擇的基準值是否能妥當運作， 將取決於影印之後之各小點的可見度。例如，若影印之後， 該小點不能被重製，則該基準值將會被調整為一較小值(使 所形成的點更大一些）。該伸展的水印可被見於第8圖中。 该水印的背景嗣會在步驟1206中，藉選擇一類似於整體背 景109的背景，並將該背景任意地散佈在水印背景中，而來 被隨機化(randomized)如第9圖所示。所產生的水印嗣會被 埋入南解析度的文件中（1207)，以滿足所要獲得的視感效 果0 在該程序中的所有步驟可見於第10圖的方塊圖中。該 原始的圖案影像會被饋入水印產生器中（120)，且初始的水 P圖木會被產生（121)。由該初始的水印圖案會產生一初始 水印（122) ’並被伸展（123)。嗣該水印的強度會被控制 (124) ’且该水印會被隨機化（ι25)。 產生一彩色水印的程序亦類似於上述灰階水印，但其 不同處在於一彩色水印可選擇隨機化該背景，或使用一組 新的複製字樣、複製字樣背景、及整體背景等。在該水印中， 200406717 將可添加一些白線來提高該等複製字樣的可見度。 為產生一彩色水印，該水印產生器會接收二圖案影像 來作為輸入。該二圖案影像之一係被用來獲得第3圖所見的 複製字樣資訊，而另一者會被用來獲得第2圖所見的公司字 5 樣資訊。該原始圖案影像的特徵部份將會被擷取。其擷取 特徵的過程係相同於灰階水印。 該初始的水印圖案將會被產生。其包含許多小點和背 景而會互相交疊。整個彩色背景將會分成三個部份：最左 部份、中間部份和最右部份。因此，該複製字樣和複製字 10 樣背景亦會被分成三個部份。不同的彩色值將會被使用於 此三部份。各影像的三個分割部份將會被重組，且一改變 的圖案會被產生於該等彩色值之間。在一較佳實施例中， 該等影像會被分成三個垂直部份：左邊約15%，中間約 70%，及右邊約15%。該等影像的左、右部份最好使用相同 15 的色彩。該改變的圖案影像亦可使用Adobe Photo Shop漸變 工具來產生。其可做到令一改變圖案能被產生，而肉眼不 能容易地看出隱藏的字樣。該初始水印將會參考所提供的 複製字樣與公司字樣來被產生。只有複製字樣會保持該等 小點。該公司字樣將會保持與在圖案影像中相同的顏色。 20 為能供作調制之用，該初始水印會分別沿X及y方向伸 展以獲得較大之點，如同在灰階水印中之操作。該彩色水 印將會被調制以獲得一良好的視感效果，此亦如同灰階水 印。其可選擇將該水印背景隨機化，或者形成一組新的複 製字樣、整體背景、及複製字樣背景而來為之。 15 200406717 若係選擇隨機化，則一類似的彩色背景將會被選出， 然後該彩色背景會被隨機化而來使其均勻地散佈。 若係選擇施加一組新的複製字樣、整體背景、及複製 字樣背景，則該等複製字樣、整體背景、及複製字樣背景 5 之圖案影像將會被施加於該彩色水印上，來達到良好的視 覺效果。 一複製字樣背景會被用來使該複製字樣在影印之後更 為明顯。所使用的複製字樣背景會比整體背景更淡，因此 在影印之後，該複製字樣背景將會淡化，而使該複製字樣 10 變得更為凸顯。 例如若使用粉紅色，則8%含量的粉紅色會被用於該複 製字樣背景，88%含量的粉紅色會被用於該複製字樣，而 42%含量的粉紅色會被用於整體背景。使用相同的顏色會 令肉眼較難以看出該複製字樣。 15 水印的埋設會在空間區域中來進行，而如第13圖所 示。原始文件的像元值會被直接修正來埋設該水印。若該 水印必須保持某種形狀，則一形狀影像會被用來作為基準 影像，而來將該水印成型為所需的形狀。 欲埋設一灰階水印時（第13圖），要被標記水印的原始文 20 件會被轉變成適當的影像格式1301，例如一gif.檔。在步驟 1302中，供埋設該水印的位置會被蒐尋。具有較少字樣的 區域係供標記水印的理想位置，因為其在影印之後能夠達 到較佳的可見度。 一旦決定該水印要被埋設在何處之後，則一水印會在发明 Description of the invention: [Technical Leadership to which the invention belongs] Field of the Invention The present invention relates to the watermarking technology of documents. Although it is not exclusive, it particularly refers to a system for hiding optical watermarks in a document. [Prior Art 3 Background of the Invention Digital watermarking is a modern form of ancient latent display technology, and is used to hide information in other information (images, documents, audio, video materials, etc.). A watermark is a hidden message in digitized or printed media. Since digital media is integrated with the content of the data file itself, no additional storage is required. Watermarking has two sides: visible structure and impulsive printing. The visible watermark is a visible semi-transparent image, which will weigh 4 in-the main shadow is strong. Its currency consists of the seal or logo of the institution that owns the rights to the main image. A visible watermark allows the primary image to be seen 'but also clearly shows that the document is the property of an institution. An invisible watermark is a stack of signals (media), which cannot be reversed, but can be calculated out. Digital watermarks have been used as a means of identifying the owner and distributor of the digital material. Unique codes are buried—digital wipes, and these codes can be used for various purposes, such as date / time identification, source identification, content identification, processing status, ownership identification, and so on. Optical watermarking differs from traditional watermarking in that it uses optical and visual methods such as photocopying. This is to visually confirm that a document is original or has been photocopied or scanned. 200406717 C Summary of the Invention 3 Summary of the Invention A first preferred aspect of the present invention is to provide a printable watermark image 'which can be hidden in a document for printing. The image includes an image background with an average intensity of the first color; and an indication area also has the average intensity of the first color. At least the intensity of the indication area is the result of at least two different frequency vibration configurations. The resolution of the watermark image, the frequency vibration configuration, and the at least two intensity systems are selected so that the color and intensity of the image background and the indication area will be consistent with the naked eye, but in the same manner as After the copying device operated by the low-pass filter processes the printed image, the indication area will become apparent to the naked eye. The average intensity of the image background can also be the result of a dithered arrangement of two or more different intensities. The intensity of two or more of the indication area, please, will be greater than the intensity of two or more of the background of the image. The surrounding shirt and the indication area may be uniform in color. The watermark image may include a visible mark area, which is a different color from the uniform color, or has a region average intensity different from the image background and the indication area. The watermark image may include at least three regions, each region having a first color selected from a group of at least two colors. The at least three regions may include a first 20 region. Is 15% of the watermark surface area, an intermediate region is about 15% of the watermark surface area, and the first region is about 15% of the watermark surface area. The first and second regions may be the same color. The middle area is interposed therebetween and may be of different colors. The boundaries between the regions can be highlighted by a changing pattern. In a second preferred aspect of the present invention, a method capable of generating a 20042004717 printable watermark image to be combined with a printable document order is provided. The method includes the following steps: selecting-the background color of the image Fill the background area of the watermark image with the axis color and the average intensity equal to the degree, and make the intensity of each pixel fall within a background intensity range. 5 An indicator area will be used by the location Select the color and fill it with an average intensity equal to the selected intensity, but make the Qianfu intensity fall within a larger range than the above ^ range, so that the image resolution, the frequency vibration configuration, and the two Or more intensities are selected so that the color and intensity of the background of the image and the indication area will be consistent with the naked eye, but in a copying device that operates as a low-pass wave device 10 to process the After the image is printed, the indication area will be clearly visible to the naked eye. The step of filling the background may include applying a frequency vibration pattern having a first difference of two intensities. The step of filling the indication area may include applying a Frequency-Vibration Pattern The second difference is greater than the first difference. The selected intensity may be less than RGB (239,239,239). It may also include the following steps: receiving a basic image; and according to the back '7, zone and private zone The difference in color or intensity between them to identify a background area and an indication area in the basic image. / 、 You can also include the following steps: · Identify one of the 20 l visible areas in the basic image; -A second color and intensity different from the first color and intensity to fill the visible marking area. It may further include the following steps:-receiving a visible marked basic image and-indicating the basic image; identifying the indicating basic image The indication area and the background area in the "Ya" will be filled. A visible mark area will also be identified as the-background area in the visible 7 mark basic image. The visible mark basic image and the filled and processed Indicates that the base image can be combined so that each corresponding image is mentioned-the color and intensity of the pixels in the image will exceed the color and intensity of the pixels in another image. The background of the visible mark image 5 pixels in the zone The pixels in the processed indication base image have a lower tolerance. The pixels in the visible mark area in the visible marker base image will have a lower contrast than the pixels in the processed indication base image. High super difference. In another preferred aspect of the present invention, a system capable of generating a printable watermark image to be combined in a printable document is provided. The system 10 includes: a device selectable An image background color and intensity; a device capable of filling the monthly scenic area of the watermark image with an average intensity of 4 colors and f at a selected intensity, so that the intensity of each pixel falls within a background intensity range; and a device The private area can be filled with the selected color and an average intensity equal to the selected intensity, but the intensity of each pixel falls within a range of 15 greater than the aforementioned intensity, so that the resolution of the image The frequency vibration configuration and the two or more intensities are selected to 'can make the color and intensity of the image background and the indication area appear to the naked eye, but in the same way as a low-pass filter Local copy operation to process the print image The indication area will become so visible to the naked eye. The device used to fill the background includes a device capable of applying a frequency vibration pattern having a first difference of two intensities. The device for filling the indication area includes a device capable of applying a frequency vibration pattern having a second intensity of two intensities, the second difference being greater than the first difference. The selected intensity can be no less than RGB (239,239,239). 8 200406717 It may also include a device that can receive a basic image; and a device that can identify a background area and a fingerless area in the basic image based on the color or intensity difference between the background area and the indication area. Finally, it may include a device for recognizing a visible mark 5 area in the basic image; and a device capable of filling the visible mark area with a second color and intensity different from the first color and intensity. Finally, it can also include a device for receiving a visible basic image and a pointing basic image; a device can recognize and fill the indication area and the background area in the pointing basic image; a device can recognize the visible mark A visible mark area and a background area in the base image; and a device can combine the visible mark basic image and the filled and processed indication basic image, so that each corresponding pixel in one of its images The color and intensity of the pixels exceed the color and intensity of the pixels in another image, and the pixels in the background area of the visible marker image will have a lower than 15 pixels over the pixels in the processed indication base image The difference is that the pixels in the visible mark area of the visible marked basic image have a high deviation from the pixels in the processed indicated basic image. In the penultimate aspect of the present invention, a watermark image produced by the above method is provided; and a file is provided with a watermark image produced by the above method. 20 In a final aspect of the present invention, a computer-usable medium is provided that includes a computer program code that enables a processor to perform one or more functions, thereby enabling the above method to be performed. BRIEF DESCRIPTION OF THE DRAWINGS In order that the present invention can be easily understood and put into practice, it will now be described by means of a non-limiting preferred embodiment described in this publication 9 200406717, which is described with reference to the attached drawings, where: Figure 1 is the original image of a grayscale watermark; Figure 2 is the original company name image of a color watermark; 5 Figure 3 is a copy type image of a color watermark; Figure 4 is a pattern image for forming a watermark; Figure 5 shows the initial pattern of a point where the mark is copied; Figure 6 shows the initial watermark image; Figure 7 shows the initial watermark image after stretching; 10 Figure 8 shows the watermark after controlling the intensity; Figure 9 FIG. 10 shows a watermark after noise is applied to the overall background. FIG. 10 is a block diagram of a preferred process for generating a watermark according to the present invention. FIG. 11 is a picture showing a pixel stretching range of the present invention. Flowchart of watermark generation procedure; 15 Figure 13 is a flowchart of the watermark embedding procedure of Figure 10; Figure 14 shows a document with a watermark; Figure 15 shows the density of the document of Figure 14 is too light in the photocopier The time of photocopying; and Fig. 16 shows the result of copying the document of Fig. 14 when the density of the photocopier is too dark. [Real package method] 3 Detailed description of the preferred embodiment The present invention relates to a technique for copy-protected printing or digital files. The present invention enables the image or mark in a printed document to be concealed so that when the 10 200406717 document is printed, the image or mark will not be seen. When the document is photocopied or scanned, the image or mark becomes visible. Watermarking on digital or printed documents is a system that controls the copying of documents printed by optical watermarking. The system includes an anti-duplication 5-optical watermark and anti-duplication background. Two watermarks can be used: a grayscale watermark, and a color watermark. The system includes a watermark generator and a watermark can be embedded to control the reproduction of documents printed by optical watermarking. In a copy control program, the owner or authorized user of the digital file 10 will use a logo (image of a pattern) that displays the ownership of the file in combination with certain copyright words. After copying the watermark document, the copied document will show obvious words (such as "copy"), which means that the document has been copied, trimmed, or tampered, and so on. When combined with the controlled printing of the original watermarked document, a reliable system for identifying digital and paper documents can be formed. As shown in Figure 12, when a grayscale watermark is to be created, the watermark generator will receive an image 1201 as an input. In step 1202, the watermark generator captures each feature part from the pattern image seen in FIG. The feature parts captured by the watermark generator include the original background 103, the company's words 20 101, and the copied words 102. The color of the company's typeface 101 should be selected so that it will disappear completely or partially after photocopying. In order to achieve this, the company's lettering should preferably be white. It is also preferable that in the pattern image, the company's typeface 101 and the copied typeface 102 are the same color (white). The cut-off watermark pattern of the copied typeface will be generated, and it contains many small dots and backgrounds that overlap with each other. The small dots 107 are all the same size, and the pitch of the small dots in the watermark pattern will remain the same. This can be seen in Figure 5. The size and color of the dots of the grayscale watermark and color watermark will be controlled. After photocopying, these dots can be clearly reproduced and distinguished. The control of the 5 watermark is modulated, which will be explained in detail later. These dots should not be too small to be reproduced after photocopying, but they should not be so large that they can be seen by the naked eye in the watermark. The ratio of the size (width x height) of the watermark image of the original pattern to the document image _ is an important factor for the visibility of these small dots after photocopying. The size of the original pattern image is selected so that when the watermark is stretched and embedded in the document, the small dots will be sufficiently small, but can be reproduced and resolved after photocopying. For example, if the selected original pattern image is 560x650, the size of the original file used for embedding is 3000 points x3500 points. Other sizes can be used as long as the dots are clear and can be reproduced. 15 The elongation (that is, the multiples of the watermark stretching in the X and y directions, respectively) is an important factor in controlling the final interval of these small dots. If the watermark is stretched 4x4 times from the original size and only a small dot is left after controlling the intensity, the distance between the dots will be 3 times the original small dot size. These spacings are sufficient to distinguish the small dots after copying. 20 The company's typeface 103 will not change during the generation process 1203, and the copy typeface 102 will retain the dot pattern. The watermark background 109 is selected. After the print image is copied, the background becomes gray. The copy type background 108 selected by the watermark generator will be lighter than the overall background 109. At step 1204, the watermark generator will stretch the initial 12 200406717 watermark. The initial watermark is stretched in the x and y directions, respectively. In one example, — if an original pixel is a point, after the watermark is expanded by the mxn stretch rate, the original small point will become mxn times the size. This system can be seen in Figure 11A, where only (0,0) is filled. In the lib graph, the pixel 5 has been expanded by 4 times the stretch rate, so it is in (0,0), (0,1), (0,2), (0,3), (1,0 ), (1,1), (1,2), (1,3), (2,0), (2,1), (2,2), (2,3), (3,0), (3,1), (3,2), (3,3) and other positions will be filled. The formula for stretching this initial watermark is: Φ (nxX0 + (i-1), mxY0 + (j-1)). 10 where (χο, Υο) is the starting position of the point, m and η are the elongation rates of the extension, and i and j are the cyclic variables. How it is applied can be seen in the following simulation code: for (i = 0; i <m; i ++) (j ~ 0 »j ^ 〇yj ^ and Lu, 15 the new point position is (nxXG + (il), mxY () + ( jl)) 2 1 The expanded watermark can be seen in Fig. 7. The watermark 鲁 is modulated in step 1205. The modulation system is used to reduce the intensity of the watermark and minimize the visibility of the small dots. The distance from the original point to the extension point will be calculated. If the distance exceeds a selected reference value of 20, the pixel will be controlled. The modulation system can be used to obtain a good visual effect. The modulation distance is used The following formula is used to calculate: Distance = ^ J (x-X0y where (χ〇, Υ〇) is the original point position, and (x, y) is the extended position. If the reference value used to control a point is selected as: If the reference value is ^ 1, only one point will be left for every 13 large points of extension. This can be seen in Figure 11C. If the point at the (0,0) position is extended 4x4 times, it will be at (0, 〇), (0,1), (0,2), (0,3), (1,0), (1,1), (1,2), (1,3) , (2, 〇), (2,1), (2,2), (2,3), (3,0 ), (3,1), (3,2), (3,3) The distance values of each position will be 0, 1, 2, 3, 1, 1.414, 2.236, 3.1623, 2.236, 2.828, 3.6056, 3, 3 · 1623, 3.6056, 4.2426, etc. If the reference value is selected as: 3 > reference value >2; after controlling the intensity, only the (丨, 2), (2, 1), (2 2) The dot at the position will be left. Whether the selected reference value can work properly will depend on the visibility of the small dots after photocopying. For example, if the small dots cannot be reproduced after photocopying, the reference point The value will be adjusted to a smaller value (making the dots larger). The stretched watermark can be seen in Figure 8. The background of the watermark will be in step 1206, by selecting a similar to the whole The background of background 109, and this background is randomly scattered in the watermark background, and it is randomized as shown in Figure 9. The generated watermark will be buried in the South Resolution file (1207) To meet the visual effect you want to get 0 All steps in this program can be seen in the block diagram in Figure 10. The original pattern image will be fed into the watermark Generator (120), and the initial water P figure will be generated (121). From this initial watermark pattern, an initial watermark (122) 'will be generated and extended (123). The intensity of the watermark will Control (124) 'and the watermark will be randomized (ι25). The process of generating a color watermark is similar to the above grayscale watermark, but the difference is that a color watermark can choose to randomize the background, or use a new set of Copy text, copy text background, and overall background. In this watermark, 200406717 will be able to add some white lines to improve the visibility of such copied text. To generate a color watermark, the watermark generator receives two pattern images as input. One of the two pattern images is used to obtain the copy type information as seen in Figure 3, while the other is used to obtain the company type 5 information as shown in Figure 2. The characteristic part of the original pattern image will be captured. The process of capturing features is the same as grayscale watermarking. This initial watermark pattern will be generated. It contains many dots and backgrounds that overlap each other. The entire colored background will be divided into three parts: the leftmost part, the middle part, and the rightmost part. Therefore, the copy text and copy text 10 background will be divided into three parts. Different color values will be used in these three parts. The three divided parts of each image will be recombined, and a changed pattern will be generated between the color values. In a preferred embodiment, the images are divided into three vertical portions: about 15% on the left, about 70% in the middle, and about 15% on the right. The left and right parts of these images should preferably use the same 15 colors. The changed pattern image can also be generated using the Adobe Photo Shop gradient tool. It can be done so that a change pattern can be produced, and the hidden words cannot be easily seen by the naked eye. The initial watermark will be generated with reference to the provided copy text and company text. Only the copy will keep these small dots. The company lettering will remain the same color as in the pattern image. 20 For the purpose of modulation, the initial watermark is extended in the X and y directions to obtain larger points, as in grayscale watermarking. The color watermark will be modulated to obtain a good visual effect, which is also like a grayscale watermark. It can choose to randomize the watermark background, or to form a new set of copy type, overall background, and copy type background. 15 200406717 If randomization is selected, a similar colored background will be selected, and then the colored background will be randomized to make it evenly spread. If you choose to apply a new set of copy type, overall background, and copy type background, the pattern images of the copy type, overall background, and copy type background 5 will be applied to the color watermark to achieve a good Visual effects. A copy type background is used to make the copy type more visible after photocopying. The background of the copied typeface will be lighter than the overall background, so after copying, the background of the copied typeface will fade, making the copied typeface 10 more prominent. For example, if pink is used, 8% pink will be used for the copy type background, 88% pink will be used for the copy type, and 42% pink will be used for the overall background. Using the same color makes it harder for the naked eye to see the copy. 15 The embedding of the watermark takes place in the space area, as shown in Figure 13. The cell values of the original file are directly modified to embed the watermark. If the watermark must maintain a certain shape, a shape image will be used as a reference image to shape the watermark into the desired shape. When you want to embed a grayscale watermark (Figure 13), the original 20 files to be marked with the watermark will be converted into an appropriate image format 1301, such as a gif. File. In step 1302, the location for embedding the watermark is searched. Areas with fewer typefaces are ideal locations for marking watermarks because they provide better visibility after photocopying. Once you decide where the watermark is to be buried, a watermark
16 200406717 步驟1303藉如灰階水印的像元調制來被埋設。假使該原如 文件的像元落在該水印像元範圍之外，則所造成的像元將 會與原始文件的像元值相同。若該原始文件的像元為白 色，則所造成的像元將會相同於該水印像元值。但是，若 5該原始文件的像元為黑色，則所造成的像元將會相同於該 原始文件像元值。最後，若該原始文件像元係介於白色和 黑色像元之間，則所造成的像元將會相同於該原始文件像 元值。 若该水印必須使用一形狀來成型，如第4圖所示，則該 10調制將會不同。若該原始文件像元落在該水印像元範圍之 外，則所造成的像元將會相同於原始文件的像元值。若今 原始文件像元為白色且落在該形狀外部，則所造成的像元 將為一白色像元。但是，若該原始文件像元為白色且落在 該形狀内部，則所造成的像元將會相同於該水印像元。若 15該原始文件像元為黑色且落在該形狀外部，則所造成的像 元將會相同於該原始文件像元。但是，若該原始文件像元 為黑色且落在該形狀内部，則所造成的像元將會相同於該 原始文件像元。若該原始文件像元既非白色亦非黑色，則 所k成的像元將會相同於該原始文件像元值。 '〇 $程序將會發生於該水印影像的每-個像元。在該水 印影像已被埋設之後，該文件將會被轉成一適當的影像格 式。例如，一jpeg或gif檔。 若對-彩色水印而言，在步驟1303中，水印將會被埋 入一文件的整體背景中。要埋設—水印，職需要標記水 17 200406717 印的原始文件將會被轉化成該適當的影像格式，例如if ， 檀。供埋設該水e卩的位置將會被找出。就彩色文件而言， 該整個背景將會被選作埋設區域，而可獲得良好的視覺效 果。 5 _色水印會被以像元_來埋設。若該原始文件像 元落在水印像元範圍外部，則所造成的像元會相同於原始 文件像元值。若a亥原始文件像元為白色像元，則所造成的 像7C將會相同於該水印像元值。若該原始文件像元為黑色 _ 像元則所^成的像元將會相同於該原始文件像元值。若 10該原始文件像元值大於水印像元，則所造成的像元將會相 同於水印像元值。若該原始文件像元既非白色亦非黑色， 貝J所ie成的像元將會相同於該原始文件像元值。 假使一彩色水印必須被成型，則彩色調制來埋設該水 印必須將该形狀納入考量。若該原始文件像元落在該水印 15像元範圍之外，則所造成的像元會與該原始像元相同。若 該原始文件像元值為白色且落在該形狀外部，則所造成的 _ 像兀將會相同於該原始文件像元。若該原始文件像元值為 白色並落在該形狀内部，則所造成的像元將會相同於該水 印像元值。若該原始文件像元值為黑色且落在該形狀之 20外，則所造成的像元將會相同於原始文件像元值。若該原 始文件像元值為黑色並落在該形狀内部，則所造成的像元 將會相同於邊原始文件像元值。若該原始文件像元值既非 黑色亦非白色並落在該形狀外部，則所造成的像元將會相 同於原始文件像元值。若該原始文件像元值既非黑色亦非 18 200406717 白色且落在該形狀内部，則所造成的像元將會相同於該原 始文件像元值。 在該水印影像被埋設之後，該影像將會被轉成一適當 的影像格式。例如，一jpeg或gif襠。 5 假使該影印機的濃度太淡或太暗，或其解析度相似於 列印機，則會較難以分辨該複製字樣。一列印機的解析度 典型為600〜1200dpi，而一影印機的解析度典型為150〜 600dpi(比此範圍更高的上眼亦可見於更高級的影印機）。為 克服此問題，一顏色檢測器將會被使用。如第14至16圖所 10 示，該顏色檢測包含一高灰階背景，及該複製性質的指示， 例如”無效”的字樣。該”無效”字樣在當被以非常暗的濃度 來影印時（第16圖）將會明顯易見；而在濃度非常輕淡時（第 15圖）將會甚為模糊。該顏色檢測器可用來辨識該影印機的 濃度。有兩種顏色檢測器可被使用：淡色檢測器及暗色檢 15 測器。當該影印機的濃度太淡時，該淡色檢測器將會變成 看不到，此乃表示該濃度的選擇不正確。當該影印機的濃 度太暗時，則該暗色檢測器會變得太暗而看不到，此亦表 示其濃度並未被正確地選擇。 在該標記水印的過程中，該水印的背景係為一重要因 20 素。若該背景太暗，則在影印之後不會淡化，因此該水印 效果會不如應有狀況一般地明顯可見。若該背景太淡，則 原始的標諸將不會在該水印中清楚可見。因此，必須在該 原始標誌的可見度與該背景的良好淡化效果之間來作一折 衷妥協。 19 針對灰階水印，乃推薦該背景應在RGB值(239,239,239) ， 、下，而相當於HTML^|#EFEFEF。就彩色水印而言，則推 薦忒3景要在此臨界值以下（即比此顏色更淡），例如任何顏 5色的含量皆少於42%。該複製字樣背景之值應與該複製字 樣同色，但應要儘可能的輕淡。 水印影印的檢查會被使用。昔已公知對一頻寬受限的 影像而言，若取樣頻率fs>2fn，則將可能完全重組該原始影 像，其中fs為該影像中出現的最高頻率，而fn為Nyquist(尼 _ 1至斯特）頻率。真實的字樣影像並無頻寬限制，因此可能會 產生誤差。該誤差能在取樣之前以低通濾波該影像而來消 減但亦會有衰減南空間頻率的代價。此即在列印產業中 所明的網紋現象’’(Moire phenomena)。網紋效應會以拍差 圖木的形式出現，其係在若影像含有接近於取樣頻率一半 的週期性時將會產生。此效應會在當該顯示光點尺寸較小 15時(相較於該取樣距離)來發生，因此該重組渡波器所切掉的 範圍會遠超過理想的低通濾波器所切掉者。嗣一在頻率 # 叹的信號會干擾-在fn〜f的伴隨信號，而造成一拍差圖 當該水印影像的頻率低於主體影像時，該水印將能藉 影印低通滤波來擷取。大部份的影印機係以⑼〜45〇dpi來 掃描，而優良品質的雷射印表機之解析度約為6〇〇邮，因 此該水印將能容易地藉影印來擷取。 本發明亦包含-電腦可用的媒體，其具有一電腦程式 瑪。該電腦程式碼係可令至少一處理器來執行一或多項功 20 200406717 能，以使一電腦能夠進行上述方法。 對熟習本發明相關技術的專業人員而言，本發明尚有 許多結構上的變化，大不相同的實施例和應用等將可被得 知，而不超出後附申請專利範圍所界定之本發明的範圍。 5 以上揭露和描述純為舉例說明，而毫無任何作為限制之意。 t圖式簡單說明3 第1圖為一灰階水印的原始圖案； 第2圖為一彩色水印的原始公司名稱影像； _ 第3圖為一彩色水印的複製字樣影像； 10 第4圖為供成型一水印的圖案影像； 第5圖示出一複製標記之點的初始圖案； 第6圖示出初始的水印影像； 第7圖示出伸展之後的初始水印影像； 第8圖示出控制強度之後的水印； 15 第9圖示出雜訊被施加於整體背景之後的水印； 第10圖為本發明產生水印之較佳過程的方塊圖； ® 第11圖示出本發明的像元伸展範圍； 弟12圖為弟10圖之水印產生程序的流程圖， 第13圖為第10圖之水印埋設程序的流程圖； 20 第14圖示出一具有水印的文件； 第15圖示第14圖的文件在影印機的濃度太淡時的影印 結果，及 第16圖示出第14圖的文件在影印機的濃度太暗時的影 印結果。 21 200406717 圖式之主要元件代表符號表】 101···公司字樣 108…複製字樣背景 102···複印字樣 109…水印背景 103···原始背景 120〜125···產生水印的步驟 107···小點 1301〜1304…埋設水印的步驟 2216 200406717 Step 1303 is buried by the pixel modulation of the grayscale watermark. If the pixels of the original file fall outside the range of the watermark pixels, the resulting pixels will be the same as the pixel values of the original file. If the pixels of the original file are white, the resulting pixels will be the same as the watermark pixel values. However, if the cells of the original file are black, the resulting cells will be the same as the cell values of the original file. Finally, if the original file cell is between white and black cells, the resulting cell will be the same as the original file cell value. If the watermark must be shaped using a shape, as shown in Figure 4, the 10 modulation will be different. If the pixels of the original file fall outside the range of the watermark pixels, the resulting pixels will be the same as the pixel values of the original file. If the original document cell is white and falls outside the shape, the resulting cell will be a white cell. However, if the original document cell is white and falls inside the shape, the resulting cell will be the same as the watermark cell. If the original file cell is black and falls outside the shape, the resulting cell will be the same as the original file cell. However, if the original file cell is black and falls inside the shape, the resulting cell will be the same as the original file cell. If the original document cell is neither white nor black, then the resulting pixel will be the same as the original document cell value. The '〇 $ procedure will occur every pixel of the watermark image. After the watermark image has been embedded, the file will be converted into an appropriate image format. For example, a jpeg or gif file. For a color watermark, in step 1303, the watermark will be embedded in the overall background of a document. To embed—watermark, you need to mark the water. 17 200406717 The original document printed will be converted into the appropriate image format, such as if, sandal. The location where the water e 卩 will be buried will be found. In the case of a color document, the entire background will be selected as a buried area and good visual effects will be obtained. 5 _color watermark will be buried with pixels_. If the original document cell falls outside the watermark cell range, the resulting cell will be the same as the original document cell value. If the pixels of the original file are white pixels, the resulting image 7C will be the same as the watermark pixel value. If the original file cell is a black _ cell, the resulting cell will be the same as the original file cell value. If the pixel value of the original document is greater than the watermark pixel, the resulting pixel will be the same as the watermark pixel value. If the pixels of the original file are neither white nor black, the pixels formed by Be J will be the same as the pixel values of the original file. If a color watermark must be formed, the color modulation to embed the watermark must take that shape into account. If the original file cell falls outside the 15-pixel range of the watermark, the resulting cell will be the same as the original cell. If the original file cell value is white and falls outside the shape, the resulting _ image will be the same as the original file cell. If the original file cell value is white and falls inside the shape, the resulting cell will be the same as the watermark cell value. If the original file cell value is black and falls outside of the shape, the resulting cell will be the same as the original file cell value. If the original file cell value is black and falls inside the shape, the resulting cell will be the same as the edge original file cell value. If the original file cell value is neither black nor white and falls outside the shape, the resulting cell will be the same as the original file cell value. If the original file cell value is neither black nor 18 200406717 white and falls inside the shape, the resulting cell will be the same as the original file cell value. After the watermark image is embedded, the image will be converted into an appropriate image format. For example, a jpeg or gif crotch. 5 If the density of the photocopier is too light or too dark, or its resolution is similar to that of a printer, it will be more difficult to distinguish the copied text. The resolution of a printer is typically 600 to 1200 dpi, while the resolution of a photocopier is typically 150 to 600 dpi (upper eyes higher than this range can also be found in more advanced photocopiers). To overcome this problem, a color detector will be used. As shown in Figures 14 to 16, the color detection includes a high grayscale background and an indication of the nature of the reproduction, such as the words "invalid". The word "invalid" will be clearly visible when photocopied at very dark densities (Figure 16); at very light densities (Figure 15) it will be very vague. The color detector can be used to identify the density of the copier. Two color detectors can be used: light color detector and dark color detector. When the density of the photocopier is too light, the light color detector will become invisible, which means that the choice of the density is incorrect. When the density of the photocopier is too dark, the dark detector becomes too dark to be seen, which also means that its density has not been properly selected. In the process of marking the watermark, the background of the watermark is an important factor. If the background is too dark, it will not fade after photocopying, so the watermark effect will not be as clearly visible as it should be. If the background is too light, the original markers will not be clearly visible in the watermark. Therefore, a compromise must be made between the visibility of the original logo and the good fade effect of the background. 19 For grayscale watermarking, it is recommended that the background should be at the RGB value (239,239,239), which is equivalent to HTML ^ | #EFEFEF. As far as color watermarking is concerned, it is recommended that the 3 scenes be below this threshold (that is, lighter than this color), for example, the content of any 5 colors is less than 42%. The value of the background of the copy should be the same color as the copy, but it should be as light as possible. Watermark copy checks will be used. It has been known in the past that for a bandwidth-limited image, if the sampling frequency fs> 2fn, the original image may be completely reconstructed, where fs is the highest frequency appearing in the image, and fn is Nyquist (Ni_1 to Est.) Frequency. True type images have no bandwidth limitation, so errors may occur. This error can be reduced by low-pass filtering the image before sampling, but also at the cost of attenuating the south space frequency. This is the moire phenomenon that is known in the printing industry. The moire effect will appear in the form of a beat map, which will occur if the image contains periodicity close to half the sampling frequency. This effect occurs when the size of the display light spot is 15 (compared to the sampling distance), so the range cut by the recombined wavelet will far exceed that cut by the ideal low-pass filter. The signal at the frequency # sigh will interfere with the accompanying signal at fn ~ f, resulting in a beat difference. When the frequency of the watermark image is lower than the main image, the watermark will be captured by photocopy low-pass filtering. Most photocopiers scan at ⑼ ~ 450 dpi, and the resolution of a good-quality laser printer is about 600 mail, so the watermark can be easily captured by photocopying. The present invention also includes a computer-usable medium having a computer program. The computer code can enable at least one processor to perform one or more functions, so that a computer can perform the above method. For those who are familiar with the related technology of the present invention, there are still many structural changes in the present invention. The widely different embodiments and applications will be known without exceeding the scope of the present invention as defined by the appended patents Range. 5 The above disclosures and descriptions are purely for illustration, without any limitation. t Schematic description 3 The first picture is the original pattern of a gray-scale watermark; the second picture is the image of the original company name with a color watermark; the third picture is a copy type image of a color watermark; 10 the fourth picture is for A pattern image of a watermark is formed; FIG. 5 shows an initial pattern of a point where a mark is copied; FIG. 6 shows an initial watermark image; FIG. 7 shows an initial watermark image after stretching; and FIG. 8 shows control intensity Watermark afterwards; 15 Figure 9 shows the watermark after noise is applied to the overall background; Figure 10 is a block diagram of the preferred process of generating a watermark in the present invention; ® Figure 11 shows the pixel extension range of the present invention ; Figure 12 is a flowchart of the watermark generation procedure of Figure 10, Figure 13 is a flowchart of the watermark embedding procedure of Figure 10; 20 Figure 14 shows a document with a watermark; Figure 15 shows Figure 14 The photocopy result of the document when the density of the photocopier is too light, and FIG. 16 shows the photocopy result of the document of FIG. 14 when the density of the photocopier is too dark. 21 200406717 Schematic representation of the main components of the diagram] 101 ... The company type 108 ... Copy type background 102 ... Copy type 109 ... Watermark background 103 ... Original background 120 ~ 125 ... Steps to generate watermark 107 ... ·· Dots 1301 ~ 1304 ... Step 22 of embedding watermark