201010408 九、發明說明： 卜 【發明所屬之技術領域】 • 本發明係關於一種資訊隱藏方法’特別是關於以邊 緣吻合及最佳替代為基礎之資訊隱藏方法。 【先前技術】 目前，網際網路已成為生活不可或缺的工具，各種 多媒體資訊，例如：影像（Image)、語音（Audio)或視 ❺ 訊（Video)等’皆能輕易透過網路廣泛的傳遞與散播， ，而，透過網際網路進行機密資訊傳輸的過程中，仍具有 谷易受到監聽、*·改或擷取機密資訊等風險。因此，目前 於機密資訊之傳輪過程中，普遍仍必須輔以資訊隱藏技術 ’以提供機密資訊較佳的安全性。 如第1圖所示，習用資訊隱藏方法，如「簡單最不 重要位元（Simple Least Significant Bit，Simple LSB)」替 代技術’其概念係為將一原始機密資訊11隱藏於一原始 ❹ 掩護影像12 (Host image)之「最不重要位元」中，該「 最不重要位元」的改變對於該原始掩護影像12之像素值 (Pixel)的影響不大，藉以達到資訊隱藏之目的。更詳言 之’其係將該原始機密資訊11分解為數個像素值，各像 ， 素值係為k-bit，k係指該原始機密資訊11欲藏入一原始 . 掩護影像12各像素值中最右邊最不重要位元的位元個數 ’而由數個k-bit為單位的像素值所組成之資訊稱為一第 一機密資訊111 ;另外，將該原始掩護影像12則分割為 一掩護影像121及一殘留影像122 ’其中該掩護影像121 201010408 係為扣除該原始掩護影像12每一個像素值之「 位元（例如位於每一個像素值最右邊之位元）最不重要 影像，該殘留影像m係為萃取該原始掩護影」像所^組成， 最不重要位元」所組成的影像；最後，再將該第一2 =「 訊111之每-個k-bit為單位的像素值取代該機密資 122’再依附於該掩護影像121原糾萃取^最留影像 位元的相對位置，以獲得一偽裝影像13。 不重要201010408 IX. Description of the invention: 卜 [Technical field to which the invention pertains] • The present invention relates to an information hiding method, particularly to an information hiding method based on edge matching and optimal replacement. [Prior Art] At present, the Internet has become an indispensable tool for life. Various multimedia information, such as Image, Audio, or Video, can be easily accessed through the Internet. Passing and disseminating, while in the process of transmitting confidential information over the Internet, there is still the risk of being vulnerable to monitoring, *·changing or capturing confidential information. Therefore, in the current process of transmitting confidential information, it is still necessary to supplement the information hiding technology to provide better security for confidential information. As shown in Figure 1, the conventional information hiding method, such as "Simple Least Significant Bit (Simple LSB)" alternative technology, is to hide a piece of original confidential information 11 from a raw 掩 cover image. In the "least significant bit" of the (Host image), the change of the "least significant bit" has little effect on the pixel value (Pixel) of the original cover image 12, thereby achieving the purpose of information hiding. More specifically, it decomposes the original confidential information 11 into a number of pixel values, each image, the prime value is k-bit, and k means that the original confidential information 11 is intended to be hidden into an original. The information of the number of bits of the rightmost least significant bit in the middle and the pixel value of several k-bit units is called a first confidential information 111; in addition, the original cover image 12 is divided into a cover image 121 and a residual image 122', wherein the cover image 121 201010408 is the least significant image of the "bit" (eg, the rightmost bit at each pixel value) for each pixel value of the original cover image 12, The residual image m is an image formed by extracting the image of the original cover image, and the least significant bit. Finally, the first 2 = "each of the k-bits of the signal 111" The pixel value replaces the secret asset 122' and then is attached to the relative position of the mask image 121 original correction extraction ^most image bit to obtain a camouflage image 13.
當欲取出隱藏於該偽裝影像13之機密資訊 ^ 依序取出該第-機密資訊⑴先前置人於°、’僅須 的每-個識為單㈣像素值，崎麵組^該影像一^ 密資訊111，再進-步將該第—機㈣訊U1還^ 始機密資訊：U。_，前述f知資訊隱藏方法 驟雖相當簡單，惟當欲進行隱藏之原始機密資訊i擔 案愈大時，該偽裝影像13所呈現出之品質亦相對容易產 生失真情形’進^容g被他人查覺該偽裝影像n藏有機 密資訊。再者，由於該方法之實作步驟相#簡易故當該 偽裝影像U不慎遭受他人竊取後，亦相對料取出該^ 密資訊，其安全性不佳。 ^ A請參閱第2騎示’習用資訊隱藏方法，如邊緣吻 : side-match)藏匿法，其概念係利用相鄰之像素來估 计欲藏入機密資訊之像素的藏人容量。#欲藏人機密資訊 之像素位於平滑區域，將藏入較少的機密資訊量；當欲藏 入機密資訊之像素在邊緣區域，將藏入較多的機密資訊量 習用一邊緣吻合（Two-Sided side-match，TSM)藏匿法 201010408 係以兩個相鄰的像素作為基準點進行判斷，如第2圖所示 ’ IV為欲藏入機密資訊之像素，在它上面及左邊的像素分 別為Pu和Pi，d為？11與P!的平均值減去像素？3£後的差異 值，如公式2所示： d= (Pu + Pi) /2-Px …公式 2 若差異值d為-1、0或1，則利用上述習用最不重要 位元替代技術將一個機密資訊藏入像素Px中；否則，利 用公式3求出須藏入之機密資訊個數N。公式3如下所示 N=[l〇g2 | d 丨]...公式 3 而須藏入之機密資訊個數N需轉換成十進位8，並 利用公式4計算出新的差異值d，，最後新的像素Ρχ值為 Pu與Pi的平均值減d’的差異值。公式4如下所示： d’=j* 2N+S，若 d>l -(2N+S)，若 d<l …公式 4 舉例而言，假設像素卩，為115，像素匕為14〇，像 素Pi為150,而欲藏入之原始機密資訊以二進位代表為 10101111。首先，計算差異值 d= (140+150) /2-115=30 , 接著利用差異值計算像素Ρχ所須藏入之機密資訊個數 | 30 | ]=4，從該原始機密資訊取得前4個位元 1010 ’並轉換成十進位（1010) 2=10後，求得新的差異 值d’為24+10=26，最後新的Ρχ像素值即為（140+150) /2一26=119 〇 201010408 求出新的差異值d， 在萃取的階段，首先利用公式 公式5如下所示: d，= (Pu+Pi) /2-Px …公式 5 倘若差異值為心〇或1的話，直接從像素值中取得 最後一個位元，否則，利用公式6 ..i 计算出此像素被藏入 多^資訊量Ν’_用公式7’求得機密資訊s，最後 Ο ❹When the confidential information hidden in the camouflage image 13 is to be taken out, the first confidential information (1) is previously selected, and each of the only required ones is a single (four) pixel value, and the image is a ^ Confidential Information 111, and then step-by-step the first machine (four) message U1 also to start confidential information: U. _, the above-mentioned information hiding method is quite simple, but when the original confidential information to be hidden is larger, the quality of the camouflage image 13 is relatively easy to produce a distortion situation. Others found that the camouflage image n possesses confidential information. Moreover, since the implementation step of the method is simple, when the camouflage image U is inadvertently stolen by others, the security information is relatively poor. ^ A Please refer to the 2nd riding 'Utility Information Hiding Method, such as the edge-match: side-match.' The hiding method is based on the use of adjacent pixels to estimate the Tibetan capacity of pixels that want to be hidden in confidential information. The pixel of the confidential information of the Tibetan is located in the smooth area, which will hide less confidential information. When the pixel that wants to hide the confidential information is in the edge area, it will hide more confidential information and use an edge to match (Two- Sided side-match, TSM) The hiding method 201010408 uses two adjacent pixels as the reference point for judgment. As shown in Fig. 2, 'IV is the pixel to be hidden in the confidential information, and the pixels on the top and the left side are respectively Pu and Pi, d? 11 minus the average of P! minus the pixels? The difference value after 3 £, as shown in Equation 2: d = (Pu + Pi) / 2-Px ... Equation 2 If the difference value d is -1, 0 or 1, then the least used bit substitution technique is used. Hide a confidential information into the pixel Px; otherwise, use Equation 3 to find the number N of confidential information to be hidden. Equation 3 is as follows: N=[l〇g2 | d 丨]...Form 3 The number of confidential information N to be hidden needs to be converted into decimal digit 8, and the new difference value d is calculated using Equation 4, Finally, the new pixel Ρχ value is the difference between the average value of Pu and Pi minus d'. Equation 4 is as follows: d'=j* 2N+S, if d>l -(2N+S), if d<l...Formula 4 For example, suppose the pixel 卩 is 115 and the pixel 匕 is 14〇. The pixel Pi is 150, and the original confidential information to be hidden is represented by the binary 10101111. First, calculate the difference value d = (140 + 150) / 2-115 = 30, and then use the difference value to calculate the number of confidential information to be hidden in the pixel | | 30 | ] = 4, the first 4 from the original confidential information After the bit 1010 ' is converted into decimal (1010) 2=10, the new difference value d' is obtained as 24+10=26, and the last new pixel value is (140+150) /2-26. =119 〇201010408 Find the new difference value d. In the extraction stage, first use the formula 5 as follows: d,= (Pu+Pi) /2-Px ... Equation 5 If the difference value is 〇 or 1 Directly obtain the last bit from the pixel value. Otherwise, use the formula 6. .i to calculate that the pixel is hidden in the information amount Ν '_ use the formula 7 to obtain the confidential information s, and finally Ο ❹
將其轉換成二進制便可得到原始的機密資訊。公式6及公 式7如下所示： A 如㈦治I 4, | ] …公式6 S=厂 d，-2N ，若 d，>l -< L-d，-2N，若 d’<l ...公式 7 别述邊緣吻合藏匿法係衍生出另一個缺點，其原始 掩護影像至少有兩個側緣之像素並未藏入機密資訊，例如 ’於上述例子中，上方及左方側緣之像素並未藏入機密資 訊，亦即第一行及第一列之像素並未藏入機密資訊，因此 藏入之機密資訊量較少；再者，該邊緣吻合藏匿法於藏入 機密資訊後’影像品質將受到影響，而使其容易被偵測出 藏有機密資訊。 【發明内容】 本發明之主要目的係提供一種資訊隱藏方法，以於 維持影像品質之前提下，有效提升一掩護影像之資訊藏入 量。 為了達到上述之發明目的，本發明的技術手段係於 該一掩護影像之至少二相鄰之最外緣藏入一替代矩陣及部 201010408 分機密資訊’並以邊緣吻合藏匿法結合該替代矩陣轉換機 密資訊進行資訊藏匿，以提升該掩護影像之資訊入量。Convert it to binary to get the original confidential information. Equation 6 and Equation 7 are as follows: A. For example, (7) rule I 4, | ] ... Formula 6 S = plant d, -2N, if d, > l -< Ld, -2N, if d'<l. .. Equation 7: The edge-matching hiding system has another disadvantage. The original screening image has at least two side-edge pixels that do not contain confidential information, such as 'in the above example, the upper and left side edges Pixels are not hidden in confidential information, that is, the pixels in the first row and the first column are not hidden in confidential information, so the amount of confidential information hidden is less; in addition, the edge is consistent with the hiding method after hiding confidential information. 'Image quality will be affected, making it easy to detect hidden information. SUMMARY OF THE INVENTION The main object of the present invention is to provide a method for hiding information, which is provided before the image quality is maintained, thereby effectively increasing the amount of information hidden in a cover image. In order to achieve the above object, the technical means of the present invention is to hide at least two adjacent outermost edges of the cover image into a replacement matrix and the portion 201010408 to divide the confidential information ' and combine the substitution matrix with the edge matching method. Confidential information is used to hide information to enhance the information input of the cover image.
根據本發明之資訊隱藏方法’其包含一前處理步驟 ，係將一原始機密>訊切割為一第一機密資訊及一第二機 密資訊’完成後進行一喪入步称；該嵌入步驟係獲取一掩 護影像包含數個第一嵌入像素及數個第二嵌入像素，且該 數個第一嵌入像素係位於該掩護影像至少二相鄰之最外緣 的像素，並將該第一機密資訊及一替代矩陣嵌入該掩護影 像之第-I人像素中，使該H人像素形祕裝像素， 完成後進行一推算步驟；該推算步驟係透過邊緣吻合藏匿 法，以推算出同時與二個偽裝像素相鄰之第二嵌入像素的 藏入容量，並由該第二機密資訊中取出對應該藏入容量之 位元 一第一判斷步驟；該第一判斷步驟係判 斷該藏入容量是否等於該第二喪人像素之最不4要位元之 預叹個數，若為「是」，則先進行一轉換步驟後，再進行 :藏入步驟，若為「否」，則直接進行城人步驟；該轉 ^驟係透過該替代矩陣對欲藏人該m像素之資訊 換並替代欲藏人n人像素之資訊；該藏入步 =該第二叙像素藏人該推算步驟中對應該藏入容量 判斷:否3後=第二判斷步驟’·該第二判斷步雜係 掩料^第—機密資誠未技藏人__像中且該 若1L之所有第二嵌人像素尚未完全藏有機密資訊， 總藏入谷量與影像品質作為一目前解，並進行 201010408 一第三判斷步驟；該第三判斷步驟係判斷該目前解是否優 * 於目則最佳解，若為「是」，則更新目前最隹解後，再進 行一第四判斷步驟，若為「否」，則直接進行進行該第四 判斷步驟，其中，若為初次進行，並無目前最佳解則以 該目别解作為目前最佳解後，進行該第四判斷步碌；及兮 第四判斷步驟係判斷是否已搜尋過所有的替代矩陣，若^ 「是」，則終止，若為「否」，則搜尋另一替代矩陣，並以 Q 該另一替代矩陣進行該嵌入步驟。 .【實施方式】 為讓本發明之上述及其他目的、特徵及優點能更明 . 顯易懂，下文特舉本發明之較佳實施例，並配合所附圖式 ’作詳細說明如下： 請參照第3圖所示，本發明之資訊隱藏方法係藉由 電腦系統（未繪示）連接至少一資料庫（未繪示）作為 執行架構，用以執行一前處理步驟S1、一嵌入步驟S2、 © 推算步驟S3、一第一判斷步驟S4、一轉換步驟S5、一 資訊藏入步驟S6、一第二判斷步驟S7、一第三判斷步驟 S8及一第四判斷步驟沾。藉此，本發明以邊緣吻合為基 礎並結合最佳替代之資訊隱藏方法，可用於將一機密資訊 . (例如：影像或文字等資訊）隱藏於如影像（Image)等 多媒體資訊中，進而提升藏入容量。 請再參照第3及4圖所示，本發明較佳實施例之前 置處理步驟S1係：將一原始機密資訊2轉換為以數個位 π表示，並將該原始機密資訊2切割為一第一機密資訊 201010408 21及-第二機㈣訊…更詳言之，透過該·系統將 * 該原始機密資訊2 轉換為以數個位元表示.，以形成—長辛 《位元组’再將該-長串之位元組進行切t卜以形成兩段 位元組’分別為該第一機密資訊21及第二機密資訊22。 另外，較佳係將該第一機密資訊21及第二機密資訊22透 過加密技術（例如常見之金鑰加密技術等）以分別獲得一 第-加密資訊及-第二加密資訊，並以該第一加密資訊及 ❹ 帛二加密資訊分別取代該第-機密資訊21及-第二機密 資訊22後再進行該嵌入步驟S2，以確保其安全性。 接著，如第5及6圖所示，本發明接著進行該嵌入 步驟S2係：獲取一掩護影像3包含數個第一嵌入像素31 及數個第二嵌入像素32，且該數個第一嵌入像素31係位 於該掩護影像3至少二相鄰之最外緣之像素，並將該第一 機密資訊21及一替代矩陣4嵌入該掩護影像3之第一嵌 入像素31中，使該第一嵌入像素31形成偽裝像素33， ❹ 其中，該替代矩陣4依該第二嵌入像素32之最不重要位 元之預故個數而隨機決定。更詳言之，該掩護影像3係由 該第一嵌入像素31及第二嵌入像素32所組成，且該數個 第一嵌入像素31係位於該掩護影像3之至少二相鄰之最 外緣，舉例而言，如第5圖所示，本實施例之第一嵌入像 素31係為該掩護影像3之第一行及第一列之像素，該掩 護影像3之其餘像素係為第二後入像素％ ;該替代矩陣4 係透過該電腦系統所隨機產生。其中，該替代矩陣4較佳 係定義為An={a[i]|〇^i$2N}，其中N係指該第二嵌入像 201010408 素32最右邊最不重要位元的位元個數，亦即該第二欲入 : 像素32可藏入資訊之藏入容量，n係為讓替代矩俥4之 大小，該替代矩陣4之大小n=2N，i係為陣列元素的位 置值’舉例而言，當該第二嵌入像素32最右邊最不重要 位元的位元個數N預設為2時，該替代矩陣4之大小 n=22=4 ’因此該替代矩陣4内之陣列元素係為〇、卜2及 3之組合，該替代矩陣4共有（2n) != (22)卜24種組合 ❹，例如[2 3 〇 U或[2 0 3 1]…等等。由該24種 可能中隨機挑選一種作為該替代矩陣4之初始解，本實施 例之替代矩陣4係以八4=[2 3 0 1]作為例子，以利後 續說明。 而該第一機密資訊21及替代矩陣4係先轉換為二進 位表示後，透過最不重要位元匹配（least 3丨职丨丘 matching’ LSB matching)藏匿法藏入該數個第一嵌入像素 31中，使該第一嵌入像素31形成偽裝像素33 ;最不重要 ❹ 位元匹配藏匿法係詳述如下··此方法每一次係將兩個機密 資訊藏入二個第一嵌入像素31中，而在藏入之前，利用最 不重要位元函數（LSB函數）及二元函數（Binary Functi〇n) 來判別如何更改像素值；請參照第5及6圖所示，以一前第 嵌入像素A及一後第一欲入像素B為一組，第一位元叫 • 及第二位元則為由該替代矩陣4或第一機密資訊21依序 取出之二個位元，例如若欲先藏入該第一機密資訊21，則每 次藏入機密資訊時，由該第一機密資訊21最左方依序取出二 個位元作為第一位元mi及第二位元叫，以藏入該前第一嵌 —12 — 201010408 入像素A及後第一嵌入像素B中。前偽裝像素C為該前第一 喪入像素A藏入第一位元mi後所形成之儀裝像素33，後撝 裝像素D為後第一礙入像素B藏入第二位元m2後所形成之 偽裝像素33 ’ LSB函數主要是計算像素的最後一個最不重要 位元的值為0或1，F係為二元函數，用來藏入及取出資訊， 其定義如公式8所示。 F (A3) =LSB ([A/2]+B)…公式 8 請參照第7圖所示，首先，判斷該前第一嵌入像素a是 否等於0 ’再判斷該前第一嵌入像素A的LSB函數值是否等 於 m!;若 A=〇 且 LSB (A) =111!，則 C=A 且 D=B-LSB (B )+m2 ;若 A=〇 且 LSB (A) —mi，則 C=A+1 且 D=B-LSB (B) +m2。若 A关〇 且 LSB (a) ，則利用 F (A，B)的 二元函數來判別是否與第二位元m2相同，若F (A»B) =m2， 則 C=A，且 D=B，若 F (A，B)妾m2，則 C=A 且 EHB+i 或 D=B-1 ;若 A关0 且 LSB (A)妾m!，則利用 F (Α·1，Β)的 二元函數來判別是否與第二位元m2相同，若F (Α-1，Β) ==!η2 ，則C=A-1且D=B ;若F (Α，Β)关m2，則再判斷該前第一 嵌入像素A是否等於255，若F (A，B)关1¾且A=255，則 C=A-1 且 D=B+1 或 B-1，若 F (A，B)爹m2 且 A尹255，則 C=A+1 且 D=B。 舉例而言，假設欲先藏入該第一機密資訊31，前第一叙 入像素A之值為140，後第一嵌入像素B之值為150，而該 第一機密資訊31中第一個位元m!及第二個位元1¾分別為〇 和1，首先，A^O，且LSB (A) =0，與機密資訊m〗的值相 -13 - 201010408 同，因此接著判斷F (A，B)是否等於第二位元吨尹（ ， 鐵測4，其值不等於第讀元吨’进此計算出修改後 的值為C=A=140及D=B_W49，如此一來便完成藏入的動作 。重複上述藏人動作，直至該替代矩陣4及第—機密資訊Μ 被完全藏入該第一嵌入像素31中，如第8圖所示，且使該第 -喪入像素31形成偽裝像素33後，再進行該推算步驟心。 其中，在資料萃取的過程時，只要取出C值的LSB，便可得 到機密資訊mi，再利用F (C，D)取得機密資訊m2，便可完 成資料取出的動作。延續上述例子，mi=LSB (c) =lsb ( 140) =0 ’ 而 m尸F (C，D) =F (140，149)=卜 請參照第8及9圖所示，本發明接著進行該推算步 称S3係：透過邊緣吻合藏匿法，以推算出同時與二個偽 裝像素33相鄰之第二嵌入像素32的藏入容量，並由該^ 二機密資訊22中取出對應該藏人容量之位元，以作^欲 藏入該一個第二嵌入像素32之資訊。更詳言之，本實施 φ 例係以二邊緣吻合藏匿法（TSM)進行該藏入容量之推算 ，舉例而言，假設Pu=140，P丨=150，該第二嵌入像素32 係為139，透過公式2可獲得差異值扣（14〇+15〇/2_ 139=6，在透過公式3可得知藏入容量Ν==[1〇&丨6丨]=2 / 並選擇由該第一機雄、資訊22中最左邊梅取2個位元，以 作為欲藏入該一個第二嵌入像素32之資訊。 請參照第3圖所示，本發明接著進行該第一判斷步 驟S4係：判斷該藏入容量是否等於該一個第二嵌入像素 32之最不重要位元之預設個數，若判斷結果為是，則進 201010408 #該轉換㈣，若觸結果為否，則進行該資訊藏入步驟 Γ 。更詳言之，廷饋土述例子，由於該一嗰第二嵌入像素 ， 32之最不重要位元之預設個數係預設為2，又所推算出之 該藏入容量等於2，判斷與該第二嵌入像素32之最不重 要位元之預設個數相同，因此先進行該轉換步驟S5，透 過該替代矩陣4對該二個位元進行轉換後，再進行該資訊 藏入步驟S6，以提升藏入資訊之安全性；若為「否」，則 Q 直接進行該資訊藏入步驟S6 ’以將該二個位元藏入該一 個第二嵌入像素32中。 請參照第3及9圖所示，本發明接著進行一轉換步 驟S5係：透過該替代矩陣4對欲藏入該一個第二嵌入像 素32之資訊進行轉換，並替代欲藏入該一個第二嵌入像 素32之資訊。更詳言之，延續上述例子，假設欲藏入該 一個第二嵌入像素32之資訊係為最左方二個位元〇1) 2 ，並將其轉換為十進位後為（4) 1Q，並利用該替代矩陣4 ❹ ，將該第二機密資訊22之該二個位元進行轉換，由於該 二個位元為（4) 10,因此選擇該替代矩陣4第四個位置 之陣列元素（1) ίο取代（4) ίο，再轉換為二進位後為（ 01) 2，以作為欲藏入該一個第二嵌入像素32之資訊，完 . 成後進行該藏入步驟S6。 $參照第9 _示，本發明接著進行該藏人步驟S6 係：將欲藏入該一個第二嵌入像素32之資訊藏入該第二 嵌入像素32中，以使該一個第二嵌入像素32完成資訊藏 入而形成偽裝像素。更詳言之，延續上述例子，欲藏入該 —15 一 201010408 一個第二嵌入像素32之資訊於轉換後為（οι) 2，再透過 ; 公式{分求出著的差異值 代入公式（2) ’便可推出該一個第二嵌入像素32於藏入 資訊後所形成之偽裝像素33的像素值係為（140+150) /2-5=140 »如此’便可將該第二機密資訊22藏入該第二 嵌入像素32中’而形成偽裝像素33，完成後進行該第二 判斷步驟S7。 ❹請參照第3圖所示，本發明接著進行該第二判斷步 驟S7係：判斷是否該第二機密資訊22尚未完全藏入該掩 護影像3中且該掩護影像3中之所有第二嵌入像素尚 未完全藏有機密資訊，若為「是」，則進行該推算步驟S3 ，若為「否」，則以目前該掩護影像3之總藏入容量與影 像品質作為一目前解，並進行該第三判斷步驟S8。更詳 呂之，若該第二機密資訊尚未完全藏入該掩護影像3之第 二嵌入像素32中，且該掩護影像3中之所有第二嵌入像 〇 素32尚未完全藏有機密資訊，則重新進行該推算步驟S3 ;否則，則以目前該掩護影像3之總藏入容量與影像品質 作為一目前解，再進行該第三判斷步驟S8。 請參照第3圖所示，本發明接著進行該第三判斷步 . 驟邡係：判斷該目前解是否優於目前最佳解，若為「是 - ，則更新目前最佳解後，再進行該第四判斷步驟，若1 為Γ否」，則直接進行進行該第四判斷步驟S9，其中，若 為初次進行，並無目前最佳解，則以該目前解作為目前 佳解後，進行該第四判斷步驟S9。 ^ 201010408 請參照第3 _示’本㈣接著進行該第四判斷步 • 称S9 I·•鑛是否域尋過所有的替代矩陣4 ， 是」，則終止，若為「否」，則搜尋另一替代矩陣4，並以 其進行該嵌入步驟S2。更詳言之，延續上述例子，該替 代矩陣共有24種可能，若已分別利用24種替代矩陣4進 行本方法，則完成本發明之資訊隱藏方法，且可獲得該 24種結果中之最佳解；否貝，卜選擇另一種替代矩陣4 ,進 Α 行該嵌入步驟S2。 請參照表1及表2所示，其分別為本發明與習用資 訊藏匿方法（TSM)之藏入容量及影像品質之比較表，表 1係將圖片一至圖片六作為機密資訊，圖片三作為掩護影 像，並將圖片一至圖片六分別藏入圖片三，而圖片一至圖 片六之内容較大，並無法完全藏入掩護影像圖片三中，因 此可以得知掩護影像圖片三之最大藏入容量為多少。表i 之AD係為本發明之藏入容量減去習用TSM之藏入容量 ❹ 的差異值，由結果可得知’本發明之資訊隱藏方法之藏入 容量皆比習用TSM大了約一千個位元左右，因此可驗證 本發明之資訊隱藏方法確實可有效提升掩護影像之藏入容 量0 ^ 1.本發明與習用TSM之蘊入交署fch軔夹 法 機密資 習用TSM 本發明 △D 圖片一 424,184 425,596 1,412 —17 — 201010408 圖片二 423,913 425,476 1,563 圖1Γ昱 424,320 425,663 1,343 圖片四 424,392 425,691 1,299 圖片五 424,387 425,615 1,228 圖片六 424,438 425,680 1,242 平均 424,272 425,620 1,348 表2係為針對表1之結果進行偽裝影像品質之分析 ，表2之AD係為本發明之影像品質減去習用TSM之影 像品質的差異值，由結果可得知，本發明之影像品質皆較 習用TSM高，因此可有效驗證本發明確實可在維持影像 品質之前提下，有效提升掩護影像之藏入容量。 表2.本發明與習用TSM之偽裝影像品質比較表According to the information hiding method of the present invention, the method includes a pre-processing step of cutting an original secret message into a first confidential information and a second confidential information, and performing a stepping step; the embedding step is Obtaining a cover image includes a plurality of first embedded pixels and a plurality of second embedded pixels, and the plurality of first embedded pixels are located at at least two adjacent outermost pixels of the cover image, and the first confidential information is obtained And an alternative matrix is embedded in the first-person pixel of the cover image, and the H-pixel pixel-shaped secret pixel is completed, and an estimation step is performed; the estimation step is performed by using the edge-matching hiding method to calculate the simultaneous Determining the hidden capacity of the second embedded pixel adjacent to the pixel, and extracting, from the second confidential information, a first determining step of the bit corresponding to the hidden capacity; the first determining step determines whether the hidden capacity is equal to The number of pre-sighs of the least number of pixels of the second funeral pixel is "yes", then a conversion step is performed first: the hiding step, if "no", the city is directly executed. a step of translating the information of the m pixel to the Tibetan by the substitution matrix and replacing the information of the pixel of the n person to be hidden; the hidden step = the second pixel of the Tibetan Judging the capacity judgment: No 3 after = Second judgment step '· The second judgment step Miscellaneous mask ^ The first - secret Zicheng unskilled __ image and if the 1L of all the second embedded pixels have not yet Completely concealing confidential information, total storage and image quality as a current solution, and proceeding to 201010408, a third judgment step; the third judgment step is to determine whether the current solution is superior* to the best solution, if Yes, after updating the current best solution, a fourth determination step is performed. If "No", the fourth determination step is directly performed. If the initial determination is made, there is no current best solution. After the target solution is the current best solution, the fourth determination step is performed; and the fourth determination step determines whether all the substitution matrices have been searched, and if "Yes", the termination is performed, and if "No" , searching for another substitution matrix and replacing it with Q Array for the embedding step. The above and other objects, features and advantages of the present invention will become more apparent. As shown in FIG. 3, the information hiding method of the present invention is connected to at least one database (not shown) as an execution architecture by a computer system (not shown) for performing a pre-processing step S1 and an embedding step S2. , the estimating step S3, the first determining step S4, the converting step S5, the information hiding step S6, the second determining step S7, the third determining step S8, and the fourth determining step. Therefore, the present invention is based on edge matching and combined with the best alternative information hiding method, which can be used to hide a confidential information (for example, information such as images or text) in multimedia information such as an image, thereby improving Hide capacity. Referring to FIG. 3 and FIG. 4 again, the pre-processing step S1 of the preferred embodiment of the present invention converts an original confidential information 2 into a plurality of bits π, and cuts the original confidential information 2 into one. First Confidential Information 201010408 21 and - Second Machine (4) News... In more detail, the original confidential information 2 is converted to a number of bits by the system to form - Chang Xin "Bytes" Then, the long string of bits is cut to form a two-segment byte, which is the first confidential information 21 and the second confidential information 22, respectively. In addition, the first confidential information 21 and the second confidential information 22 are preferably obtained by using an encryption technology (for example, a common key encryption technology) to obtain a first-encrypted information and a second encrypted information, respectively. The encryption information and the second encryption information replace the first-confidential information 21 and the second confidential information 22, respectively, and then the embedding step S2 is performed to ensure the security. Next, as shown in FIGS. 5 and 6, the present invention proceeds to the embedding step S2: acquiring a mask image 3 including a plurality of first embedded pixels 31 and a plurality of second embedded pixels 32, and the plurality of first embeddings The pixel 31 is located at at least two adjacent outermost edges of the cover image 3, and the first confidential information 21 and a substitute matrix 4 are embedded in the first embedded pixel 31 of the cover image 3, so that the first embedded The pixel 31 forms a dummy pixel 33, wherein the replacement matrix 4 is randomly determined according to the number of pre-emptive bits of the second embedded pixel 32. In more detail, the cover image 3 is composed of the first embedded pixel 31 and the second embedded pixel 32, and the plurality of first embedded pixels 31 are located at least two adjacent outermost edges of the cover image 3. For example, as shown in FIG. 5, the first embedded pixel 31 of the embodiment is the pixel of the first row and the first column of the mask image 3, and the remaining pixels of the mask image 3 are the second image. Into the pixel %; the substitution matrix 4 is randomly generated by the computer system. Wherein, the substitution matrix 4 is preferably defined as An={a[i]|〇^i$2N}, where N is the number of bits of the rightmost least significant bit of the second embedded image 201010408 prime 32 , that is, the second desire: the pixel 32 can hide the hidden capacity of the information, n is the size of the substitution matrix 4, the size of the substitution matrix 4 is n=2N, and the position value of the array element is ' For example, when the number N of bits of the rightmost least significant bit of the second embedded pixel 32 is preset to be 2, the size of the replacement matrix 4 is n=22=4', so the array in the replacement matrix 4 The element system is a combination of 〇, 卜 2 and 3. The substitution matrix 4 has (2n) != (22) 卜 combinations of 24, such as [2 3 〇U or [2 0 3 1]... and so on. One of the 24 possibilities is randomly selected as the initial solution of the substitution matrix 4, and the substitution matrix 4 of the present embodiment is exemplified by eight 4 = [2 3 0 1] for the sake of subsequent explanation. The first confidential information 21 and the substitution matrix 4 are first converted into binary representations, and the first embedded pixels 31 are hidden by the least significant bit matching (LSB matching) hiding method. The first embedded pixel 31 is formed into a camouflage pixel 33; the least important ❹ bit matching hiding system is detailed as follows: · This method hides two confidential information into two first embedded pixels 31 each time, Before hiding, use the least important bit function (LSB function) and binary function (Binary Functi〇n) to determine how to change the pixel value; please refer to the 5th and 6th, and insert the pixel first. A and a first desired pixel B are a group, and the first bit is called • and the second bit is two bits sequentially extracted by the substitution matrix 4 or the first confidential information 21, for example, if desired First, the first confidential information 21 is hidden, and each time the confidential information is hidden, two bits are sequentially taken from the leftmost side of the first confidential information 21 as the first bit mi and the second bit call, Hiding the first first inlay—12 — 201010408 into pixel A and the first embedded image B, respectively. The front camouflage pixel C is the instrument-mounted pixel 33 formed after the first first lost pixel A is hidden in the first bit mi, and the rear-mounted pixel D is the second hindered pixel B after the second bit m2 is hidden. The formed camouflage pixel 33' LSB function is mainly to calculate the value of the last least significant bit of the pixel is 0 or 1, and F is a binary function for hiding and fetching information, which is defined as shown in Equation 8. . F (A3) = LSB ([A/2] + B)... Equation 8 Referring to Figure 7, first, it is determined whether the first first embedded pixel a is equal to 0' and then the first embedded pixel A is determined. Whether the LSB function value is equal to m!; if A=〇 and LSB (A) =111!, then C=A and D=B-LSB (B )+m2 ; if A=〇 and LSB (A) —mi, then C=A+1 and D=B-LSB (B) +m2. If A is closed and LSB (a), then the binary function of F (A, B) is used to determine whether it is the same as the second bit m2. If F (A»B) = m2, then C = A, and D =B, if F (A, B) 妾 m2, then C = A and EHB + i or D = B-1; if A is 0 and LSB (A) 妾 m!, then F (Α·1, Β The binary function is used to determine whether it is the same as the second bit m2. If F (Α-1,Β) ==!η2, then C=A-1 and D=B; if F(Α,Β)off m2 , then determine whether the first first embedded pixel A is equal to 255, if F (A, B) is off 13⁄4 and A = 255, then C = A-1 and D = B + 1 or B - 1, if F (A , B) 爹m2 and A Yin 255, then C=A+1 and D=B. For example, suppose that the first confidential information 31 is to be hidden first, the value of the first first introduced pixel A is 140, and the value of the first embedded pixel B is 150, and the first of the first confidential information 31 is Bit m! and second bit 13⁄4 are 〇 and 1, respectively, first, A^O, and LSB (A) =0, which is the same as the value of confidential information m -13 - 201010408, so then judge F ( A, B) is equal to the second ton ton Yin (, iron test 4, its value is not equal to the first reading ton 'in this calculation of the modified value of C = A = 140 and D = B_W49, so that Completing the hidden action. Repeating the above Tibetan action until the replacement matrix 4 and the first confidential information Μ are completely hidden in the first embedded pixel 31, as shown in FIG. 8, and the first-incoming pixel is made After forming the camouflage pixel 33, the estimation step is performed. Among them, in the process of data extraction, the confidential information mi can be obtained by taking the LSB of the C value, and then the confidential information m2 is obtained by using F (C, D). The data removal action can be completed. Continue the above example, mi=LSB (c) =lsb (140) =0 ' and m corpse F (C,D) =F (140,149)=Bu please refer to the 8th and As shown in FIG. 9 , the present invention further performs the estimation step S3 system: through the edge-matching hiding method, to calculate the hiding capacity of the second embedded pixel 32 adjacent to the two camouflage pixels 33 simultaneously, and The secret information 22 is taken out of the bit corresponding to the Tibetan capacity, so as to be hidden in the information of the second embedded pixel 32. In more detail, the embodiment φ is performed by the two edge matching method (TSM). For the calculation of the hidden capacity, for example, suppose that Pu=140, P丨=150, and the second embedded pixel 32 is 139, and the difference value buckle can be obtained by formula 2 (14〇+15〇/2_ 139=6) In the formula 3, you can know the hidden capacity Ν==[1〇&丨6丨]=2 / and choose 2 bits from the leftmost plum in the first machine and the information 22 as the desire The information of the second embedded pixel 32 is hidden. Referring to FIG. 3, the present invention proceeds to the first determining step S4: determining whether the hidden capacity is equal to the least important of the second embedded pixel 32. The preset number of bits, if the judgment result is yes, then enter 201010408 # this conversion (four), if the result is no, then enter The information is hidden in the step 。. In more detail, the courtesy example, due to the second embedded pixel, the preset number of the least significant bits of 32 is preset to 2, and is calculated. The storage capacity is equal to 2, and the predetermined number of the least significant bits of the second embedded pixel 32 is determined to be the same. Therefore, the converting step S5 is performed first, and the two bits are converted by the replacement matrix 4. Then, the information is further hidden in step S6 to improve the security of the hidden information; if "No", Q directly performs the information hiding in step S6' to hide the two bits in the second one. Embedded in pixel 32. Referring to Figures 3 and 9, the present invention further performs a conversion step S5 of converting information to be hidden in the second embedded pixel 32 through the replacement matrix 4, and instead of hiding the second one. Information embedded in pixel 32. More specifically, continuing the above example, it is assumed that the information to be hidden in the second embedded pixel 32 is the leftmost two bits 〇1) 2 and converted to decimal (4) 1Q, And using the substitution matrix 4 ❹ to convert the two bits of the second confidential information 22, since the two bits are (4) 10, the array element of the fourth position of the replacement matrix 4 is selected ( 1) ίο replaces (4) ίο, and then converts to binary (01) 2 as the information to be hidden in the second embedded pixel 32. After the completion, the hiding step S6 is performed. Referring to FIG. 9, the present invention proceeds to the Tibetan step S6: hiding information to be hidden in the second embedded pixel 32 into the second embedded pixel 32 such that the second embedded pixel 32 Complete the information hiding to form a camouflage pixel. More specifically, continuing the above example, the information of a second embedded pixel 32 to be hidden in the data is converted to (οι) 2, and then transmitted; the formula {the obtained difference value is substituted into the formula (2) The pixel value of the camouflage pixel 33 formed by the second embedded pixel 32 after the information is hidden is (140+150) /2-5=140 » so that the second confidential information can be 22 is hidden in the second embedded pixel 32 to form the camouflage pixel 33, and after completion, the second determining step S7 is performed. Referring to FIG. 3, the present invention proceeds to the second determining step S7: determining whether the second confidential information 22 has not been completely hidden in the cover image 3 and all the second embedded pixels in the cover image 3. The confidential information is not completely hidden. If it is "Yes", the estimation step S3 is performed. If "No", the total storage capacity and image quality of the current cover image 3 are used as a current solution, and the first Third, step S8 is judged. More specifically, if the second confidential information is not completely hidden in the second embedded pixel 32 of the cover image 3, and all the second embedded pixels 32 in the cover image 3 have not completely hidden confidential information, then The estimation step S3 is performed again; otherwise, the total storage capacity and image quality of the cover image 3 are currently used as a current solution, and the third determination step S8 is performed. Referring to FIG. 3, the present invention proceeds to the third determining step. Step: Determine whether the current solution is better than the current best solution. If yes, then update the current best solution and then proceed. In the fourth determining step, if 1 is no, the fourth determining step S9 is directly performed. If there is no current best solution for the first time, the current solution is used as the current best solution. The fourth determining step S9. ^ 201010408 Please refer to the 3rd _ show 'this (4) and then carry out the fourth judgment step. • S9 I·• mine is the domain to find all the substitution matrix 4, yes, then terminate, if it is no, search for another An alternative matrix 4 is used and the embedding step S2 is performed therewith. More specifically, continuing the above example, the substitution matrix has 24 possibilities. If the method is performed using 24 substitution matrices 4 respectively, the information hiding method of the present invention is completed, and the best of the 24 results can be obtained. Solution; no, the other alternative matrix 4 is selected, and the embedding step S2 is performed. Please refer to Table 1 and Table 2, which are respectively a comparison table of the hidden capacity and image quality of the invention and the conventional information hiding method (TSM). Table 1 uses Picture 1 to Picture 6 as confidential information, and Picture 3 as a cover. Image, and the picture one to the picture six are hidden in the picture three, and the content of the picture one to the picture six is large, and can not be completely hidden in the cover image picture three, so it can be known that the maximum hidden capacity of the cover image picture three is . The AD of Table i is the difference between the hidden capacity of the present invention and the hidden capacity of the conventional TSM. From the results, it can be known that the hidden capacity of the information hiding method of the present invention is about one thousand larger than the conventional TSM. It can be verified that the information hiding method of the present invention can effectively improve the hiding capacity of the cover image. 0 ^ 1. The present invention and the conventional TSM are included in the FX 轫 法 机 机 本 本 本 本 本 本 本 本 D Picture 424,184 425,596 1,412 —17 — 201010408 Picture 2 423,913 425,476 1,563 Figure 1Γ昱424,320 425,663 1,343 Picture 4 424,392 425,691 1,299 Picture 5 424,387 425,615 1,228 Picture 6 424,438 425,680 1,242 Average 424,272 425,620 1,348 Table 2 is a camouflage for the results of Table 1. For the analysis of image quality, the AD of Table 2 is the difference between the image quality of the present invention and the image quality of the conventional TSM. From the results, the image quality of the present invention is higher than that of the conventional TSM, so that the present invention can be effectively verified. It can be lifted before the image quality is maintained, effectively increasing the hiding capacity of the cover image. Table 2. Comparison of camouflage image quality between the present invention and conventional TSM
方法 機密資 習用之TSM 本發明 △D 圖片一 549,352 550,726 1,374 圖片二 546,959 548,985 2,026 圖片三 548,930 550,178 1,248 圖片四 549,546 550,309 763 圖片五 549,347 550,668 1,321 圖片六 548,869 549,850 981 平均 548,834 550,119 1,286 —18 — 201010408 如上所述，本發明透過於該掩護影像3之至少二最 y 外緣藏入資訊4益以邊緣吩合藏谨法結合該替代矩陣4轉 ^ 換機密資訊進行資訊藏匿，以使本發明於維持影像品質之 前提下，具有可有效提升資訊藏入量之功效。 雖然本發明已利用上述較佳實施例揭示，然其並非 用以限定本發明，任何熟習此技藝者在不脫離本發明之精 神和範圍之内，相對上述實施例進行各種更動與修改仍屬 ❹ 本發明所保護之技術_，因此本發明之保護範圍當視後 附之申請專利範圍所界定者為準。 201010408 L圖式簡單說明】 第1谓；習如簧訊薇藏方法之~執行流輕示意,, 第3圖 第4圖 第5圖 第6圖 第7圖 第 意圖。 第9圖 第2圓：另一習知資訊隱藏方法之執行流程示意圖。 第3圖：本發明之資訊隱藏方法之流程圖。 ° 本發明之前處理步驟之示意圖。 本發明之嵌入步驟之示意圖。 本發明之嵌入步驟之示意圖。Method Confidentiality TSM The invention ΔD picture one 549,352 550,726 1,374 picture two 546,959 548,985 2,026 picture three 548,930 550,178 1,248 picture four 549,546 550,309 763 picture five 549,347 550,668 1,321 picture six 548,869 549,850 981 average 548,834 550,119 1,286 —18 — 201010408 As described above, the present invention hides the information by using the at least two most y outer edges of the cover image 3 to perform information hiding with the replacement matrix 4 to change the confidential information, so that the present invention maintains the image. Before the quality is raised, it has the effect of effectively increasing the amount of information hidden. While the invention has been described in connection with the preferred embodiments described above, it is not intended to limit the scope of the invention. The invention is protected by the invention, and therefore the scope of the invention is defined by the scope of the appended claims. 201010408 L diagram simple explanation] The first sentence; Xiru spring exchange method of the implementation of ~ light flow, 3, 4, 5, 5, 6, 7 Figure 9 Circle 2: Schematic diagram of the execution flow of another conventional information hiding method. Figure 3: Flow chart of the information hiding method of the present invention. ° Schematic diagram of the processing steps prior to the present invention. A schematic diagram of the embedding step of the present invention. A schematic diagram of the embedding step of the present invention.
圖 本發明之嵌入步驟之操作過程示意圖。 :本發明藏入替代矩陣及第一機密資訊後之 :本發明之轉換步驟之示意圖 【主要元件符號說明】 11原始機密資訊 12原始掩護影像 122殘留影像 2 原始機密資訊 22第二機密資訊 31第一嵌入像素 33偽裝像素 A 前第一I入像素 C 前偽裝像素 111第一機密資訊 121掩護影像 13偽裝影像 21第一機密資訊 3 掩護影像 32第二嵌入像素 4 替代矩陣 B 後第一嵌入像素 D 後偽装像素 —20 —Figure Schematic diagram of the operation of the embedding step of the present invention. : The present invention is hidden in the replacement matrix and the first confidential information: a schematic diagram of the conversion step of the present invention [main component symbol description] 11 original confidential information 12 original cover image 122 residual image 2 original confidential information 22 second confidential information 31 An embedded pixel 33 camouflages the pixel A before the first I enters the pixel C, the front camouflage the pixel 111, the first secret information 121, the mask image 13, the camouflage image 21, the first confidential information 3, the mask image 32, the second embedded pixel 4, the first embedded pixel after the matrix B is replaced D after camouflage pixels - 20 -