A robust image steganography method based on adaptive STC-ECC strategy
By adopting an adaptive STC-ECC strategy, dynamically selecting the carrier sequence and optimizing the generation of the checksum, the problems of redundant checksums and error propagation in image steganography are solved, improving the robustness and security of image steganography and making it suitable for covert communication in social networks.
Patent Information
- Application Number
- CN202310964979.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-08-02
- Publication Date
- 2026-11-10
- Estimated Expiration
- 2043-08-02
AI Technical Summary
Existing image steganography techniques in social networks suffer from problems such as redundant checksums, error propagation, and reduced robustness and security due to fixed checksums, especially when message payloads change, making it difficult to maintain stealth.
An adaptive STC-ECC strategy is adopted, which dynamically selects the carrier sequence and optimizes the generation of the check code. By combining Reed-Solomon error correction code and STC encoding, secret information is embedded in segments to reduce error propagation and flexibly adjust the check code overhead, thereby generating an adaptive steganographic sequence.
It improves the robustness and security of image steganography, enabling it to remain anonymous in social networks, adapt to different message payload variations, reduce error propagation, and enhance the image's resistance to detection.
Smart Images

Figure CN117041442B_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of steganography technology, specifically relating to a robust image steganography method based on an adaptive STC-ECC strategy. Background Technology
[0002] The purpose of image steganography is to achieve covert communication through digital images. However, in images transmitted on social networks, operators often perform secondary compression due to bandwidth and storage limitations. To improve communication security, adaptive steganography uses Separate Check Cell (STC) coding to minimize image distortion. Robust steganography, on the other hand, can still correctly extract hidden messages even after secondary compression.
[0003] Over the past decade, academia has proposed a "compression-resistant domain & ECC-STC" framework to achieve robust adaptive embedding. This framework uses highly compression-resistant coefficients as carrier coefficients, processes the encoded secret message using error-correcting codes (ECC), and embeds the encoded data into the image using STC encoding. However, to ensure robustness, a large number of additional check codes need to be embedded, which leads to a significant decrease in security as the message payload increases. Furthermore, since STC is a matrix encoding, error propagation is prone to occur during the decoding process, and even with the addition of a large number of check codes, good robustness cannot be achieved.
[0004] To address this issue, the academic community proposed the "Construction of Compression-Resistant Domain & STC-ECC" framework. During the embedding process, this framework divides the carrier sequence into two parts: the first part is used to embed the secret message, and then ECC is used to encode a portion of the hidden sequence, embedding the checksum into the second part. This approach effectively reduces error propagation and decreases the overhead of the checksum, thereby improving robustness and security. However, this framework also introduces a problem: regardless of the amount of embedded payload, there is a fixed checksum overhead, making it easily detectable by steganalysts even when transmitting small payloads.
[0005] The aforementioned technical deficiencies have become a problem that urgently needs to be solved in this field. Summary of the Invention
[0006] To address the aforementioned problems, this invention proposes a robust image steganography method based on an adaptive STC-ECC strategy. Through a novel robust steganography framework, it solves the problems of redundant checksums, error propagation, and fixed checksums in existing frameworks, improving the robustness and security of robust adaptive steganography in social networks, while eliminating the need to share other information. The specific steps of the robust image steganography method based on the adaptive STC-ECC strategy provided by this invention are as follows:
[0007] Step S1: Decode the JPEG file. The carrier image used to embed the secret message is decoded in the JPEG domain, and the Discrete Cosine Transform (DCT) coefficient matrix X is obtained for use in subsequent steps.
[0008] Step S2: Calculate the modification loss ρ for each coefficient in X, quantifying the modification distortion of each coefficient in X. This can be done using existing distortion functions such as UERD and J-UNIWARD.
[0009] Step S3, extract candidate vector element sequence C c For different image transmission channels or social platforms, the carrier selection frequency band can be adjusted, and candidate carrier sequences C can be extracted using a dithering modulation algorithm. c .
[0010] Step S4: Embed secret information using the adaptive "STC-ECC" strategy. Based on the candidate vector sequence C... c By filtering the payload and embedding density, the secret message is embedded into the carrier image to generate a steganographic sequence S.
[0011] Step S41, extract the candidate vector sequence C. c Random scrambling is performed to ensure that the coefficients of various distortion magnitudes are evenly distributed.
[0012] Step S42: Because the selected Reed-Solomon (RS) error correction code is a block code, it is necessary to divide the candidate carrier sequence C into blocks. c Cut into n b Each of equal length l b The carrier element block is used to generate a checksum with fixed parameters.
[0013] Step S43, based on the payload, i.e., the length l of the embedded secret message. m Determine the required number of carrier element blocks n e And select n e Block candidate carrier element blocks are combined to form carrier sequence C * ;, n e As shown in the formula below:
[0014]
[0015] Where w is the embedding density parameter.
[0016] The vector sequence C * Cut into front-end vector sequences at a 3:1 ratio. and the subsequent vector sequence
[0017] Step S44: Based on the modification loss ρ, embed the secret message m into the front-end carrier sequence using STC encoding. Obtain the front-end steganography sequence
[0018] Step S45: Use RS(n,k,t) error correction code to correct the preceding steganographic sequence. Encode the code and obtain its checksum portion.
[0019] Step S46: Embed the check code into the subsequent carrier sequence using STC encoding. In the process, the back-end steganographic sequence is obtained.
[0020] Step S47, steg sequence and Combine them and fill in any unselected carrier element blocks (if any), to obtain the initial steganographic sequence S. * .
[0021] Step S48, using the same random number seed, the steganographic sequence S * Reverse the scramble to obtain the steganographic sequence S.
[0022] Step S5: Modify carrier elements. Based on the steganographic sequence S generated in the previous step, the coefficients in the carrier image are modified using a dithering modulation algorithm to hide the secret information.
[0023] Step S6: Output the steganalytic image. The coefficients modified in step S5 are JPEG encoded to generate a JPEG image file, which is the steganalytic image.
[0024] The beneficial effects of this invention are as follows:
[0025] 1. The improved embedding algorithm effectively reduces error propagation by ECC encoding the steganalytic sequence after STC embedding, ensuring reliable extraction of secret messages.
[0026] 2. A dynamic carrier sequence selection and optimization check code generation mechanism is introduced to adjust the check code overhead according to changes in message payload, increasing the flexibility and adaptability of the method.
[0027] 3. While maintaining the visual quality of the stegana, it enhances robustness, making the stegana difficult to detect under normal observation and analysis.
[0028] The advantage of this invention lies in its highly robust performance, making it widely applicable to transmission environments such as social networks. This robust steganography method is suitable for both individual and enterprise users, providing a reliable and secure solution for covert communication. Through this invention, users can conduct covert communication on social networks without sharing other information, while simultaneously achieving excellent robustness and confidentiality. Attached Figure Description
[0029] Figure 1 This is a flowchart of the method provided in an embodiment of the present invention;
[0030] Figure 2 This is a flowchart of the adaptive STC-ECC embedding process provided in an embodiment of the present invention;
[0031] Figure 3 A schematic diagram of the frequency band embedded in an 8*8 DCT coefficient block in the JPEG domain;
[0032] Figure 4 This is a schematic diagram of a dithering modulation algorithm;
[0033] Figure 5 This is a diagram illustrating the detection error rate in an example. Detailed Implementation
[0034] This invention provides a robust image steganography method based on an adaptive STC-ECC strategy for embedding secret information. Its core is to achieve both security and robustness in steganography. This method is applicable to various covert communication services, providing users with highly reliable covert communication services.
[0035] To better understand the technical solution of the present invention, the embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that all other embodiments obtained by those skilled in the art without creative effort, as described herein, fall within the scope of protection of the present invention.
[0036] Please refer to Figure 1 , Figure 1 The flowchart illustrates a robust steganography method based on adaptive STC-ECC, as provided in an embodiment of the present invention.
[0037] The specific implementation method is as follows:
[0038] Step S1: Decode the JPEG file, decode the carrier image into the JPEG domain, and obtain the Discrete Cosine Transform (DCT) coefficient matrix X for use in subsequent steps.
[0039] Step S2: Calculate the modification loss ρ of the coefficients, quantizing the modification distortion of each coefficient in X. This can be calculated using existing distortion functions UERD and J-UNIWARD.
[0040] Step S3, extract candidate vector element sequence C c For different image transmission channels or social platforms, the carrier selection frequency band can be adjusted, and candidate carrier sequences C can be extracted using a dithering modulation algorithm. c .
[0041] For example, when the channel compression quality is known, lower frequency coefficients can be selected for embedding to ensure steganography security. When the channel is unknown, mid-to-low frequency coefficients can be selected for embedding to ensure robustness. Please refer to [reference needed]. Figure 3 , Figure 3 For an 8x8 DCT block, where the embedding field E k The frequency range is from k=1 to k=15, with each frequency representing a different level of security and robustness.
[0042] For example, E k =4,5 represents a lower frequency band, suitable for a known channel, while E k =7,8,9 represent the mid-to-low frequency band, suitable for unknown channels.
[0043] The jitter modulation algorithm mentioned above, such as Figure 4 As shown, the DCT coefficients are divided into set A and set B according to the quantization value Δ, representing 0 or 1 respectively. The selected coefficient values are then mapped to a binary sequence S.
[0044] Step S4: Embed secret information using the adaptive "STC-ECC" strategy. Select candidate vector sequence C. c Then, by filtering the payload and embedding density, the secret message is embedded in the carrier image to generate a steganographic sequence.
[0045] Step S41, extract the candidate vector sequence C. c Randomize the coefficients to ensure a uniform distribution of coefficients with varying degrees of distortion.
[0046] Step S42: Since the selected Reed-Solomon (RS) error correction code is a block code, the candidate carrier sequence needs to be divided into n parts. b Each of equal length l b The carrier element block is used to generate a checksum with fixed parameters.
[0047] Step S43, based on the payload, i.e., the length l of the embedded secret message. m Determine the required number of carrier element blocks n e And select n e Block candidate carrier element blocks are combined to form carrier sequence C * ;n e As shown in the formula below:
[0048]
[0049] Where w is the embedding density parameter.
[0050] Simultaneously, the vector sequence C * Cut into front-end vector sequences at a 3:1 ratio. and the subsequent vector sequence
[0051] Step S44: Based on the modification loss ρ, embed the secret message m into the front-end carrier sequence using STC encoding. Obtain the front-end steganography sequence
[0052] The STC encoding includes the original STC matrix, the STC encoded matrix, the carrier matrix, the steganographic matrix, and the secret information. A specific example is as follows:
[0053]
[0054]
[0055]
[0056]
[0057] m = [0 1 1 1] T
[0058] Among them, H 2×2 The matrix is the original matrix, H 4×8 The STC encoding matrix is derived from the original matrix basis. For the carrier matrix, This is a steganographic matrix.
[0059] Step S45: Use RS(n,k,t) error correction code to correct the preceding steganographic sequence. Encode the code and obtain its checksum portion.
[0060] The RS(n,k,t) error correction code, by treating t bits as an integer, outputs an encoding of n integers for k integers, which includes (nk) check codes, and has the ability to correct (nk) / 2 integer errors.
[0061] Step S46: Embed the check code into the subsequent carrier sequence using STC encoding. In the process, the back-end steganographic sequence is obtained. like Figure 2 As shown.
[0062] Step S47, steg sequence and Combine them and fill in any unselected carrier element blocks (if any), to obtain the initial steganographic sequence S. * .
[0063] Step S48, using the same random number seed, the steganographic sequence S * Reverse the scramble to obtain the steganographic sequence S.
[0064] Step S5: Modify carrier elements. Based on the steganographic sequence S generated in the previous step, the coefficients in the carrier image are modified using a dithering modulation algorithm to hide the secret information.
[0065] The jitter modulation algorithm mentioned above, such as Figure 4 As shown, the DCT coefficients are divided into set A and set B according to the quantization value Δ, representing 0 or 1 respectively. If the element in the steganographic sequence does not correspond to the value mapped by the actual coefficient, the coefficient value is modified to match the steganographic element.
[0066] Step S6: Output the steganographic image. The coefficients modified in step S5 are JPEG encoded to generate a JPEG image file.
[0067] Using the BOSSbase1.01 image library, 10,000 images in the library were first compressed using JPEG with a quality factor of 75 to serve as a candidate carrier image set in the frequency domain. Then, representative steganography algorithms such as TCM, DMMR, ESS, SSR, and Adaptive-GMAS, as well as the robust steganography algorithm in this scheme, were used with RS(200,192,8) error correction coding. Randomly generated secret messages were embedded with embedding payloads ranging from 0.01 to 0.1 bpnzAC to generate the corresponding steganography images. Specific experimental settings are shown in Table 1.
[0068] Table 1
[0069]
[0070] The synchronization of the channel matching network at the embedding and receiving ends was tested. First, for the candidate carrier image set, attacks were launched on the steganographic images generated by different algorithms at different embedding ratios. Then, for each steganography algorithm, the secret message was extracted from the corresponding steganographic image under each embedding payload, and the number of perfectly extracted images was calculated, as shown in Table 2.
[0071] Table 2
[0072]
[0073] The experimental results in Table 2 demonstrate that the robust steganography algorithm embedded using the adaptive STC-ECC strategy employed in this invention exhibits strong robustness against JPEG recompression. This is primarily due to the method of encoding the STC-embedded sequence with ECC, which avoids the spread of STC decoding errors present in previous algorithms. This ensures the complete and correct extraction of secret information at the receiving end.
[0074] To test the anti-detection performance of the steganography algorithm for statistical features in generating encrypted images according to the algorithm of this invention, frequency domain steganography detection features such as CCPEV and DCTR were used, combined with an ensemble classifier to test robust steganography algorithms such as TCM, DMMR, ESS, SSR, and Adaptive-GMAS, as well as encrypted images generated by the algorithm proposed in this invention under different embedding payloads. For each set of encrypted images generated by different algorithms at different embedding ratios, half were randomly used for training and the other half for testing in the experiment, and the detection error rate P of the carrier images corresponding to different methods was obtained. E like Figure 5 As shown, the left figure is a schematic diagram of the detection error rate of CCPEV, and the right figure is a schematic diagram of the detection error rate of DCTR.
[0075] The experimental results above show that, compared with the robust steganography algorithms such as TCM, DMMR, ESS, SSR and Adaptive-GMAS proposed in the past, the algorithm proposed in this invention can generate steganography images with excellent anti-detection capabilities.
[0076] Through the above implementation methods, this invention can effectively improve the security and reliability of robust image steganography, and is applicable to various covert communication services. This method maintains a high degree of concealment during image transmission, providing a reliable and secure covert communication solution.
[0077] While specific embodiments of the invention have been described in detail by way of examples, those skilled in the art should understand that these examples are for illustrative purposes only and not intended to limit the scope of the invention. Those skilled in the art should understand that modifications can be made to the above embodiments without departing from the scope and spirit of the invention. The scope of the invention is defined by the appended claims.
[0078] In summary, the image robust steganography method provided in this application can improve the security and reliability of steganography, and is widely applicable to various covert communication services, providing users with highly reliable covert communication services.
Claims
1. A robust image steganography method based on an adaptive STC-ECC strategy, characterized in that, Includes the following steps: Step S1: Decode the carrier image used to embed the secret message in the JPEG domain and obtain the Discrete Cosine Transform (DCT) coefficient matrix X; Step S2: Calculate the modification distortion for each coefficient in X to obtain the modification loss. ; Step S3: Extract candidate carrier sequences using a dithering modulation algorithm. ; Step S4, based on the candidate vector sequence By filtering based on payload and embedding density, secret messages are embedded into the carrier image to generate a steganalytic sequence. The specific process is as follows: Step S41, extract the candidate vector sequence to obtain the candidate vector sequence. Randomize; Step S42, select the candidate vector sequence Cut into Equal length The carrier element block; Step S43, based on the payload, i.e., the length of the embedded secret message Determine the number of carrier element blocks required. , and select Block candidate carrier element blocks are combined into a carrier sequence ; vector sequence Cut into 3:1 ratio and the subsequent vector sequence ; Number of carrier element blocks The calculation is as follows: in, For embedding density parameters; Step S44, based on the modified loss Using the Check Cell Code (STC) to encode secret messages Embedded into the front-end vector sequence Obtain the front-end steganography sequence ; Step S45, using Error-correcting codes, for the preceding steganographic sequence Encode the code and obtain its verification code; Step S46: Embed the check code into the subsequent carrier sequence using STC encoding. In the process, the subsequent steganographic sequence is obtained. ; Step S47, steg sequence and Combine them to obtain a preliminary steganographic sequence. ; Step S48, using the same random number seed, the steganographic sequence Reverse scrambling to obtain the steganographic sequence ; Step S5, based on the steganography sequence By using a dithering modulation algorithm to modify the coefficients in the carrier image, the secret information can be hidden. Step S6: JPEG encode the modified coefficients from step S5 to generate a JPEG image file, i.e., a steganographic image.
2. The image robust steganography method based on the adaptive STC-ECC strategy according to claim 1, characterized in that, In step S2, the modified distortion of each coefficient in X is calculated using the distortion functions UERD and J-UNIWARD.
3. The image robust steganography method based on the adaptive STC-ECC strategy according to claim 1, characterized in that, In step S3, the candidate carrier sequence is extracted using the dithering modulation algorithm. During the process, the frequency band of the carrier is adjusted for different image transmission channels or social platforms.
4. The image robust steganography method based on the adaptive STC-ECC strategy according to claim 3, characterized in that, In step S44, the check lattice code STC includes the original STC matrix, the STC encoded matrix, the carrier matrix, the steganographic matrix, and the secret information, as follows: in, The matrix is the original matrix. The STC encoding matrix is derived from the original matrix basis. For the carrier matrix, This is a steganographic matrix.
5. The image robust steganography method based on the adaptive STC-ECC strategy according to claim 4, characterized in that, In step S45, the aforementioned Error correction codes, by Each bit is considered an integer. Output n integers. The encoding of an integer, which contains A verification code, for correction There are one integer error.
6. The image robust steganography method based on the adaptive STC-ECC strategy according to claim 5, characterized in that, In step S47, the steganographic sequence and During combination, if there are unselected carrier element blocks, then... and During combination, fill in the unselected carrier element block to obtain the steganographic sequence. .
