CN107682585A - A kind of Watermarking for Halftone Images method based on two-sided embedded error diffusion - Google Patents

Abstract

A kind of Watermarking for Halftone Images method based on two-sided embedded error diffusion proposed in the present invention, its main contents include：Image watermarking conjugation error diffusion, dual conjugation error diffusion and two-sided embedded error diffusion, its process is, first in image watermarking conjugation error diffusion, pass through the stealthy half tone image of error diffusion generation of the executing rule in original-gray image, define quantization error, then stealthy half tone image will appear from the certain type of artifact of referred to as boundary artefacts, then according to two policy calculations, two minimum distortions, the formula of in general halftoning visual watermark problem is finally proposed in two-sided embedded error diffusion, change output halftone level by performing necessary switching.The present invention is using the two-sided embedded error diffusion of content recognition, it is contemplated that expected watermark decoding performance and cover image and the certain content of watermark, as a result produces the half tone image with good visual quality, can preferably put into practical application.

Description

Half-tone image watermarking method based on double-sided embedded error diffusion

Technical Field

The invention relates to the field of halftone image watermarking, in particular to a halftone image watermarking method based on double-sided embedded error diffusion.

Background

With the rapid development of the internet, digital media (digital audio, digital image, digital video, etc.) on the network can be downloaded and spread at will, which brings great economic and mental damage to copyright owners, and people pay more and more attention to copyright protection and authentication of digital media. Digital watermarking is an important digital media content protection technique, which, unlike conventional encryption systems, does not primarily aim to restrict access to the media, but rather to ensure that the watermark in the media is not altered or removed, providing the media with the necessary authentication information. Digital halftoning refers to the process of reproducing continuous tone images (e.g., grayscale images and color images) on binary devices or limited grayscale devices. Nowadays, digital halftone technology has been widely applied in the fields of printing, faxing, newspaper and magazine printing, etc., and information such as company names, logos or all authors can be embedded in pictures or videos to solve the problem of copyright protection of halftone images. However, most conventional watermarking methods for grayscale images, such as least significant bit embedding, cannot be performed efficiently on halftone images, resulting in poor visual quality of the halftone images.

The invention provides a halftone image watermarking method based on double-sided embedded error diffusion, which comprises the steps of firstly generating an invisible halftone image by performing regular error diffusion on an original gray level image in data hiding conjugate error diffusion, defining quantization errors, then generating certain type of artifacts called boundary artifacts on the invisible halftone image, then calculating two minimum distortions according to two strategies, finally providing a general halftone visible watermarking problem formula in double-sided embedded error diffusion, and changing an output halftone value by performing necessary switching. The invention adopts content identification double-sided embedded error diffusion, considers the expected watermark decoding performance and the specific content of the cover image and the watermark, and consequently generates a halftone image with good visual quality, so that the halftone image can be better put into practical application.

Disclosure of Invention

Aiming at the problem that least significant bit embedding cannot be effectively executed on a halftone image, the invention aims to provide a halftone image watermarking method based on double-sided embedding error diffusion.

In order to solve the above problems, the present invention provides a halftone image watermarking method based on double-sided embedded error diffusion, which mainly comprises:

data hiding conjugate error diffusion;

(II) double conjugate error diffusion;

and (III) double-sided embedded error diffusion.

Wherein, the watermark is characterized in that a message called watermark is embedded into another message called cover message to generate another message called hidden message, and the message is similar to the cover message; the watermark can typically be decoded by processing the hidden message; the message may be an image, video, audio, voice, or other media content.

Wherein, the halftone is a special image processing technology for printing a gray image on a printing material; printed materials are generally available in only two shades: the color of the paper is white, and the ink is black; halftoning approximates an original 8-bit (grayscale) image using only 1 bit (black and white), which will resemble the original grayscale image when viewed from a distance; the existing halftone method includes: ordered dithering, error diffusion, point diffusion, and direct binary search.

Wherein, the data hide the conjugate error diffusion, let X₁And X₂The original gray scale images can be the same or different; let Y₁And Y₂For the generated latent halftone image, W is the binary pattern to be embedded, W_wIs the set of positions of white pixels in W, W_bIs the set of positions of the black pixels in W; order toRepresenting a binary sum (AND) operation, AND let ⊙ represent a binary exclusive nor (XNOR) operation;

in data hiding conjugate error diffusion, by applying a bias voltage at X₁Performing regular error diffusion to generate Y₁And the conventional error diffusion process is represented by the following equation:

u₁＝x₁(i,j)+∑h(k,l)×e₁(i-k,j-l) (1)

e₁(i,j)＝u₁(i,j)-y₁(i,j) (3)

wherein u is₁(i, j) is the current pixel value x carried by the current pixel₁(i, j) and previous error, h (k, l) is error diffusion kernel and e₁(i, j) is an error generated when the current pixel is processed; error e to be diffused₁(i, j), i.e. the quantization error is defined as u₁(i, j) and y₁(i, j) difference between (i, j).

Further, the data hiding conjugate error diffusion system obtains Y₁Thereafter, by applying reference X₂、Y₁And W produces a second halftone image Y₂(ii) a For (i, j) ∈ W_b，y₂(i, j) will be in contact with y₁(i, j) conjugation; for (i, j) ∈ W_wIf X is₁＝X₂Then y will be enforced₂(i,j)＝y₁(i, j); if X is₁≠X₂Then y is₂(i, j) will favor y₁(i, j) are the same;

the data hiding conjugate error diffusion system will be paired with u₂(i, j) performing experimental quantification; then ifWhereinIs y₂(i, j), calculating the minimum distortion Δ u (i, j) for switching the current pixel; if the minimum distortion is acceptable, i.e., Δ u (i, j) ≦ T, then a handover will be performed; the threshold T controls the contrast of the disclosed secret pattern and the latent halftone image Y₂A trade-off between visual quality of; as T increases, the contrast of the disclosed secret pattern increases, and the latent halftone image Y₂The visual quality of (2) is degraded.

Wherein the double conjugate error diffusion is at Y₂Will present some type of artifact called boundary artifact, mainly at X₁＝X₂When (i, j) is at W_bThe bottom and right border flat areas of the position in (1); if the filter is applied to Y₂Then the boundary image will be more apparent on the obtained edge map; to reduce boundary artifacts and improve the performance of watermarking techniques, Y may be generated₁And Y₂Performing modified data hiding conjugate error diffusion; in data hiding conjugate error diffusion, Y₁And Y₂Simultaneously generating; for (i, j) ∈ W_b/W_wData hiding conjugate error diffusion will first pair u₁(i, j) and u₂(i, j) experimental quantification was performed and then the two minimum distortions were calculated according to two strategies.

Further, the two strategies are that y is₂(i, j) favours the reaction with y₁(i, j) conjugation/identity to obtain Δ u₂(i, j); strategy II y₁(i, j) favours the reaction with y₂(i, j) conjugation/identity to obtain Δ u₁(i,j)；

Calculating Δ u₁(i, j) and Δ u₂(i, j) thereafter, selecting a strategy that causes less distortion; similar to data hiding conjugate error diffusion, in calculating Δ u₁(i, j) and Δ u₂T is also used during (i, j) to control distortion that is acceptable to the user; when X is present₁＝X₂When data hiding conjugate error diffusion will force y₁(i, j) and y₂(i, j) are the same.

Wherein, the double-sided embedded error diffusion proposes a formula of a general Halftone Visual Watermark (HVW) problem, as shown in equation (4):

minD_h+λ*D_w(4)

wherein D is_hRepresenting distortion caused during watermark embedding, D_wRepresents the difference between the decoded watermark and the original watermark, and lambda is less than or equal to 0; the output halftone values may be changed by performing the necessary switching.

Further, said changing the output halftone level, in the double-sided embedded error diffusion, in order to perform the necessary switching to change the output halftone level, will distort Δ u_1,iIs added to x_1,iAnd will distort Δ u_2,iIs added to x_2,iTo change the pixel value before quantization; by making Δ U₁And Δ U₂Is added to X during the watermark embedding process₁And X₂ED (·) is a conventional error diffusion process, and double-sided embedded error diffusion can be represented by equation (5):

wherein,representing AND operationsOr the decoding operation of the XNOR operation (⊙).

Further, the conventional error diffusion is assumed to be performed from the first pixel x in the conventional error diffusion processing sequence_1,1(x_2,1) To the last pixel x_1,N(x_2,N) Wherein N is X₁(X₂) Number of pixels in (1), Y₁And Y₂Concurrently, equation (5) may be changed to equations (6) and (7):

where N ∈ {2,3, …, N }, J₁And J_nIs the lowest cost achieved by the optimization process; by counting the current pixel x_1,iAnd x_2,iTo calculate the optimum Δ u_1,iAnd Δ u_2,iThereby solving the relaxation optimization problem and can be optimized by simply at X₁+ΔU₁And X₂+ΔU₂Performs conventional error diffusion to generate a final halftone image Y₁And Y₂。

Drawings

Fig. 1 is a system framework diagram of a halftone image watermarking method based on double-sided embedded error diffusion according to the present invention.

FIG. 2 is a data hiding conjugate error diffusion system of the halftone image watermarking method based on double-sided embedded error diffusion according to the present invention.

Fig. 3 is a dual conjugate error diffusion system of a halftone image watermarking method based on double-sided embedded error diffusion according to the present invention.

Detailed Description

It should be noted that the embodiments and features of the embodiments in the present application can be combined with each other without conflict, and the present invention is further described in detail with reference to the drawings and specific embodiments.

Fig. 1 is a system framework diagram of a halftone image watermarking method based on double-sided embedded error diffusion according to the present invention. The method mainly comprises data hiding conjugate error diffusion, dual conjugate error diffusion and double-sided embedded error diffusion.

In watermarking, a message called watermark is embedded into another message called cover message to generate another message called hidden message, which is similar to the cover message; the watermark can typically be decoded by processing the hidden message; the message may be an image, video, audio, voice, or other media content.

Halftoning is a special image processing technique for printing a grayscale image on a printed material; printed materials are generally available in only two shades: the color of the paper is white, and the ink is black; halftoning approximates an original 8-bit (grayscale) image using only 1 bit (black and white), which will resemble the original grayscale image when viewed from a distance; the existing halftone method includes: ordered dithering, error diffusion, point diffusion, and direct binary search.

Double-sided embedded error diffusion, a general formulation of the Halftone Visual Watermark (HVW) problem is proposed, as shown in equation (1):

minD_h+λ*D_w(1)

wherein D is_hRepresenting distortion caused during watermark embedding, D_wRepresents the difference between the decoded watermark and the original watermark, and lambda is less than or equal to 0; the output halftone values may be changed by performing the necessary switching.

Changing the output halftone level, in double-sided embedded error diffusion, to perform the necessary switching to change the output halftone level, the distortion Δ u_1,iIs added to x_1,iAnd will distort Δ u_2,iIs added to x_2,iTo change the pixel value before quantization; by making Δ U₁And Δ U₂Is added to X during the watermark embedding process₁And X₂ED (·) is a conventional error diffusion process, and double-sided embedded error diffusion can be represented by equation (5):

wherein,representing AND operationsOr the decoding operation of the XNOR operation (⊙).

Assume that the conventional error diffusion processing order is from the first pixel x_1,1(x_2,1) To the last pixel x_1,N(x_2,N) Wherein N is X₁(X₂) Number of pixels in (1), Y₁And Y₂Concurrently, equation (2) may be changed to equations (3) and (4):

where N ∈ {2,3, …, N }, J₁And J_nIs the lowest cost achieved by the optimization process; by counting the current pixel x_1,iAnd x_2,iTo calculate the optimum Δ u_1,iAnd Δ u_2,iThereby solving the relaxation optimization problem and can be optimized by simply at X₁+ΔU₁And X₂+ΔU₂Performs conventional error diffusion to generate a final halftone image Y₁And Y₂。

FIG. 2 is a data hiding conjugate error diffusion system of the halftone image watermarking method based on double-sided embedded error diffusion according to the present invention. Let X₁And X₂The original gray scale images can be the same or different; let Y₁And Y₂For the generated latent halftone image, W is the binary pattern to be embedded, W_wIs the set of positions of white pixels in W, W_bIs the set of positions of the black pixels in W; order toRepresenting a binary sum (AND) operation, AND let ⊙ represent a binary exclusive nor (XNOR) operation;

in data hiding conjugate error diffusion, by applying a bias voltage at X₁Performing regular error diffusion to generate Y₁And the conventional error diffusion process is represented by the following equation:

u₁＝x₁(i,j)+∑h(k,l)×e₁(i-k,j-l) (5)

e₁(i,j)＝u₁(i,j)-y₁(i,j) (7)

wherein u is₁(i, j) is the current pixel value x carried by the current pixel₁(i, j) and previous error, h (k, l) is error diffusion kernel and e₁(i, j) is an error generated when the current pixel is processed; error e to be diffused₁(i, j), i.e. the quantization error is defined as u₁(i, j) and y₁(i, j) difference between (i, j).

In obtaining Y₁Thereafter, by applying reference X₂、Y₁And W produces a second halftone image Y₂(ii) a For (i, j) ∈ W_b，y₂(i, j) will be in contact with y₁(i, j) conjugation; for (i, j) ∈ W_wIf X is₁＝X₂Then y will be enforced₂(i,j)＝y₁(i, j); if X is₁≠X₂Then y is₂(i, j) will favor y₁(i, j) are the same;

the data hiding conjugate error diffusion system will be paired with u₂(i, j) performing experimental quantification; then ifWhereinIs y₂(i, j), calculating the minimum distortion Δ u (i, j) for switching the current pixel; if the minimum distortion is acceptable, i.e., Δ u (i, j) ≦ T, then a handover will be performed; the threshold T controls the contrast of the disclosed secret pattern and the latent halftone image Y₂A trade-off between visual quality of; as T increases, the contrast of the disclosed secret pattern increases, and the latent halftone image Y₂The visual quality of (2) is degraded.

Fig. 3 is a dual conjugate error diffusion system of a halftone image watermarking method based on double-sided embedded error diffusion according to the present invention. At Y₂Will present some type of artifact called boundary artifact, mainly at X₁＝X₂When (i, j) is at W_bThe bottom and right border flat areas of the position in (1); if the filter is applied to Y₂Then the boundary image will be more apparent on the obtained edge map; to reduce boundary artifacts and improve the performance of watermarking techniques, Y may be generated₁And Y₂Performing modified data hiding conjugate error diffusion; in data hiding conjugate error diffusion, Y₁And Y₂Simultaneously generating; for (i, j) ∈ W_b/W_wData hiding conjugate error diffusion will first pair u₁(i, j) and u₂(i, j) experimental quantification was performed and then the two minimum distortions were calculated according to two strategies.

The strategy is₂(i, j) favours the reaction with y₁(i, j) conjugation/identity to obtain Δ u₂(i, j); strategy II y₁(i, j) favours the reaction with y₂(i, j) conjugation/identity to obtain Δ u₁(i,j)；

Calculating Δ u₁(i, j) and Δ u₂(i, j) thereafter, selecting a strategy that causes less distortion; similar to data hiding conjugate error diffusion, in calculating Δ u₁(i, j) and Δ u₂T is also used during (i, j) to control distortion that is acceptable to the user; when X is present₁＝X₂When data hiding conjugate error diffusion will force y₁(i, j) and y₂(i, j) are the same.

It will be appreciated by persons skilled in the art that the invention is not limited to details of the foregoing embodiments and that the invention can be embodied in other specific forms without departing from the spirit or scope of the invention. In addition, various modifications and alterations of this invention may be made by those skilled in the art without departing from the spirit and scope of this invention, and such modifications and alterations should also be viewed as being within the scope of this invention. It is therefore intended that the following appended claims be interpreted as including preferred embodiments and all such alterations and modifications as fall within the scope of the invention.

Claims (10)

1. A half tone image watermark method based on double-sided embedded error diffusion is characterized in that the method mainly comprises a data hiding conjugate error diffusion (I); double conjugate error diffusion (two); double-sided embedded error diffusion (iii).

2. Watermark according to claim 1, characterized in that in the watermark a message called watermark is embedded in another message called cover message to generate another message called hidden message, which is similar to the cover message; the watermark can typically be decoded by processing the hidden message; the message may be an image, video, audio, voice, or other media content.

3. The halftone according to claim 1, wherein the halftone is a special image processing technique of printing a gray-scale image on a printing material; printed materials are generally available in only two shades: the color of the paper is white, and the ink is black; halftoning approximates an original 8-bit (grayscale) image using only 1 bit (black and white), which will resemble the original grayscale image when viewed from a distance; the existing halftone method includes: ordered dithering, error diffusion, point diffusion, and direct binary search.

4. The data concealment conjugated error diffusion (I) of claim 1, wherein X is₁And X₂The original gray scale images can be the same or different; let Y₁And Y₂For the generated latent halftone image, W is the binary pattern to be embedded, W_wIs the set of positions of white pixels in W, W_bIs the set of positions of the black pixels in W; order toRepresenting a binary sum (AND) operation, instructionRepresenting a binary exclusive nor (XNOR) operation;

in data hiding conjugate error diffusion, by applying a bias voltage at X₁Performing regular error diffusion to generate Y₁And the conventional error diffusion process is represented by the following equation:

u₁＝x₁(i,j)+∑h(k,l)×e₁(i-k,j-l) (1)

<mrow> <msub> <mi>y</mi> <mn>1</mn> </msub> <mrow> <mo>(</mo> <mi>i</mi> <mo>,</mo> <mi>j</mi> <mo>)</mo> </mrow> <mo>=</mo> <mfenced open = "{" close = ""> <mtable> <mtr> <mtd> <mrow> <mn>0</mn> <mo>,</mo> </mrow> </mtd> <mtd> <mrow> <msub> <mi>u</mi> <mn>1</mn> </msub> <mrow> <mo>(</mo> <mi>i</mi> <mo>,</mo> <mi>j</mi> <mo>)</mo> </mrow> <mo>&lt;</mo> <mn>128</mn> </mrow> </mtd> </mtr> <mtr> <mtd> <mrow> <mn>255</mn> <mo>,</mo> </mrow> </mtd> <mtd> <mrow> <msub> <mi>u</mi> <mn>1</mn> </msub> <mrow> <mo>(</mo> <mi>i</mi> <mo>,</mo> <mi>j</mi> <mo>)</mo> </mrow> <mo>&amp;GreaterEqual;</mo> <mn>128</mn> </mrow> </mtd> </mtr> </mtable> </mfenced> <mo>-</mo> <mo>-</mo> <mo>-</mo> <mrow> <mo>(</mo> <mn>2</mn> <mo>)</mo> </mrow> </mrow>

e₁(i,j)＝u₁(i,j)-y₁(i,j) (3)

wherein u is₁(i, j) is the current pixel value x carried by the current pixel₁(i, j) and previous error, h (k, l) is error diffusion kernel and e₁(i, j) is an error generated when the current pixel is processed; error e to be diffused₁(i, j), i.e. the quantization error is defined as u₁(i, j) and y₁(i, j) difference between (i, j).

5. Data hiding conjugate according to claim 4Error diffusion system, characterized in that in obtaining Y₁Thereafter, by applying reference X₂、Y₁And W produces a second halftone image Y₂(ii) a For (i, j) ∈ W_b，y₂(i, j) will be in contact with y₁(i, j) conjugation; for (i, j) ∈ W_wIf X is₁＝X₂Then y will be enforced₂(i,j)＝y₁(i, j); if X is₁≠X₂Then y is₂(i, j) will favor y₁(i, j) are the same;

the data hiding conjugate error diffusion system will be paired with u₂(i, j) performing experimental quantification; then ifWhereinIs y₂(i, j), calculating the minimum distortion Δ u (i, j) for switching the current pixel; if the minimum distortion is acceptable, i.e., Δ u (i, j) ≦ T, then a handover will be performed; the threshold T controls the contrast of the disclosed secret pattern and the latent halftone image Y₂A trade-off between visual quality of; as T increases, the contrast of the disclosed secret pattern increases, and the latent halftone image Y₂The visual quality of (2) is degraded.

6. The dual conjugate error diffusion (II) of claim 1, wherein Y is₂Will present some type of artifact called boundary artifact, mainly at X₁＝X₂When (i, j) is at W_bThe bottom and right border flat areas of the position in (1); if the filter is applied to Y₂Then the boundary image will be more apparent on the obtained edge map; to reduce boundary artifacts and improve the performance of watermarking techniques, Y may be generated₁And Y₂Performing modified data hiding conjugate error diffusion; in data hiding conjugate error diffusion, Y₁And Y₂Simultaneously generating; for (i, j) ∈ W_b/W_wData hiding conjugate error diffusion will first pair u₁(i, j) and u₂(i, j) experimental quantification was performed and then the two minimum distortions were calculated according to two strategies.

7. Two policies based on claim 6, wherein policy is y₂(i, j) favours the reaction with y₁(i, j) conjugation/identity to obtain Δ u₂(i, j); strategy II y₁(i, j) favours the reaction with y₂(i, j) conjugation/identity to obtain Δ u₁(i,j)；

Calculating Δ u₁(i, j) and Δ u₂(i, j) thereafter, selecting a strategy that causes less distortion; similar to data hiding conjugate error diffusion, in calculating Δ u₁(i, j) and Δ u₂T is also used during (i, j) to control distortion that is acceptable to the user; when X is present₁＝X₂When data hiding conjugate error diffusion will force y₁(i, j) and y₂(i, j) are the same.

8. The double-sided embedded error diffusion (c) of claim 1, wherein the general Halftone Visual Watermark (HVW) problem is formulated as shown in equation (4):

min D_h+λ*D_w(4)

wherein D is_hRepresenting distortion caused during watermark embedding, D_wRepresents the difference between the decoded watermark and the original watermark, and lambda is less than or equal to 0; the output halftone values may be changed by performing the necessary switching.

9. Changing the output halftone level according to claim 8, wherein in the double-sided embedded error diffusion, in order to perform the necessary switching to change the output halftone level, the distortion Δ u is distorted_1,iIs added to x_1,iAnd will distort Δ u_2,iIs added to x_2,iTo change the pixel value before quantization; by making Δ U₁And Δ U₂Is added during the watermark embedding processIs added to X₁And X₂ED (·) is a conventional error diffusion process, and double-sided embedded error diffusion can be represented by equation (5):

wherein,representing AND operationsOR XNOR operationThe decoding operation of (1).

10. Conventional error diffusion as claimed in claim 9, wherein the conventional error diffusion processing order is assumed to be from the first pixel x_1,1(x_2,1) To the last pixel x_1,N(x_2,N) Wherein N is X₁(X₂) Number of pixels in (1), Y₁And Y₂Concurrently, equation (5) may be changed to equations (6) and (7):

<mrow> <msub> <mi>J</mi> <mn>1</mn> </msub> <mo>=</mo> <munder> <mrow> <mi>m</mi> <mi>i</mi> <mi>n</mi> </mrow> <mrow> <msub> <mi>&amp;Delta;u</mi> <mrow> <mn>1</mn> <mo>,</mo> <mn>1</mn> </mrow> </msub> <mo>,</mo> <msub> <mi>&amp;Delta;u</mi> <mrow> <mn>2</mn> <mo>,</mo> <mn>1</mn> </mrow> </msub> </mrow> </munder> <mo>{</mo> <mo>|</mo> <msub> <mi>&amp;Delta;u</mi> <mrow> <mn>1</mn> <mo>,</mo> <mn>1</mn> </mrow> </msub> <msup> <mo>|</mo> <mi>p</mi> </msup> <mo>+</mo> <mo>|</mo> <msub> <mi>&amp;Delta;u</mi> <mrow> <mn>2</mn> <mo>,</mo> <mn>1</mn> </mrow> </msub> <msup> <mo>|</mo> <mi>p</mi> </msup> </mrow>

where N ∈ {2,3, …, N }, J₁And J_nIs the lowest cost achieved by the optimization process; by counting the current pixel x_1,iAnd x_2,iTo calculate the optimum Δ u_1,iAnd Δ u_2,iThereby solving the relaxation optimization problem and can be optimized by simply at X₁+ΔU₁And X₂+ΔU₂Performs conventional error diffusion to generate a final halftone image Y₁And Y₂。