CN110796585B - Image hiding method based on deep learning - Google Patents
Image hiding method based on deep learning Download PDFInfo
- Publication number
- CN110796585B CN110796585B CN201911033271.6A CN201911033271A CN110796585B CN 110796585 B CN110796585 B CN 110796585B CN 201911033271 A CN201911033271 A CN 201911033271A CN 110796585 B CN110796585 B CN 110796585B
- Authority
- CN
- China
- Prior art keywords
- image
- block
- carrier
- target hidden
- deep learning
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000013135 deep learning Methods 0.000 title claims abstract description 17
- 238000000034 method Methods 0.000 title claims abstract description 14
- 230000011218 segmentation Effects 0.000 claims description 8
- 230000001131 transforming effect Effects 0.000 claims description 6
- 238000013507 mapping Methods 0.000 claims description 5
- 238000003709 image segmentation Methods 0.000 description 3
- 239000000969 carrier Substances 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000005336 cracking Methods 0.000 description 1
- 238000013501 data transformation Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T1/00—General purpose image data processing
- G06T1/0021—Image watermarking
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/50—Image enhancement or restoration using two or more images, e.g. averaging or subtraction
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/10—Segmentation; Edge detection
- G06T7/11—Region-based segmentation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T9/00—Image coding
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20021—Dividing image into blocks, subimages or windows
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20212—Image combination
- G06T2207/20221—Image fusion; Image merging
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Multimedia (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses an image hiding method based on deep learning, which specifically comprises the following operation steps: s1: importing a target hidden image a and a carrier image A as a carrier thereof; s2: respectively segmenting the target hidden image a and the carrier image A according to the same pixel size; s3: acquisition block a m And block A m The characteristic parameters of (1); s4: deep learning, namely respectively performing cross comparison on data in the image block a database and the image block A database; s5: using block a m Replacing the closest parameter tile A m And all the recombined image blocks A m And recombining to form an image for output. The invention replaces the image blocks of the carrier image with the image blocks of the target hidden image, the image blocks of the target hidden image have higher dispersibility, and finally the image blocks of the carrier image containing the image blocks of the target hidden image are recombined, so that the target hidden image is dispersed in the carrier image by micro image blocks, is difficult to detect and has stronger confidentiality.
Description
Technical Field
The invention belongs to the technical field of image hiding, and particularly relates to an image hiding method based on deep learning.
Background
In the image hiding technology, images, videos, voices and text digital texts are used as carriers, and information to be hidden is distributed in the carriers and cannot be easily found. The image hiding technology has important significance for military, intelligence and national security. The existing image hiding method is that after data transformation, a target hidden image is hidden in a certain area of a carrier image as a whole in a centralized way, so that the boundary of the target hidden image is obviously different from the carrier image, and the defect that the target hidden image is easy to recognize exists.
Disclosure of Invention
The present invention is directed to an image hiding method based on deep learning to solve the above problems in the background art.
In order to achieve the purpose, the invention provides the following technical scheme: an image hiding method based on deep learning specifically comprises the following operation steps:
s1: importing a target hidden image a and a carrier image A serving as a carrier of the target hidden image a, comparing the overall parameters of the target hidden image a and the carrier image A, mapping and transforming the parameters of the target hidden image a to approximate the parameters of the carrier image A, and encrypting the mapping and transforming parameters;
s2: respectively segmenting the target hidden image a and the carrier image A according to the same pixel size to obtain a plurality of image blocks a of the target hidden image a m And a plurality of blocks A of the carrier image A m ；
S3: acquisition block a m And block A m Respectively and correspondingly storing the characteristic parameters in an image block a database and an image block A database;
s4: deep learning, cross-comparing the data in the image block a database and the image block A database respectively, and determining the data to be compared with the image block a m Segment A with the closest parameters m ；
S5: using block a m Segment A with closest replacement parameters m And all the recombined image blocks A m And recombining to form an image for output.
Preferably, the overall parameters in step S1 include Tamura texture features, HOG features, and LBP features.
Preferably, the segmentation size of the image in step S2 is 5-20px × 5-20px.
Preferably, in the step S2, when the image is divided, the divided image block a is divided by a key m And block A m The location parameter of (2) is encrypted.
Preferably, step S3 block a m And block A m The characteristic parameters include gray values and color moments.
The invention has the technical effects and advantages that:
the invention can obtain the processing image which is close to the parameters of the carrier image by processing the target hidden image in advance, finely dividing the processed target hidden image and the carrier image, obtaining the image blocks of the carrier image which are similar to the image blocks of the target hidden image by comparison, replacing the image blocks of the carrier image by the image blocks of the target hidden image, and finally recombining the image blocks of the carrier image which contain the image blocks of the target hidden image to ensure that the target hidden image is dispersed in the carrier image by the tiny image blocks, thereby realizing the hiding of the target hidden image in the carrier image, being difficult to find, and the image blocks are encrypted to have stronger confidentiality.
Drawings
FIG. 1 is a schematic flow diagram of the present invention;
FIG. 2 is a schematic diagram of a deep learning structure according to the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be obtained by a person skilled in the art without making any creative effort based on the embodiments in the present invention, belong to the protection scope of the present invention.
As shown in fig. 1-2, an image hiding method based on deep learning specifically includes the following operation steps:
s1: introducing a target hidden image a and a carrier image A serving as a carrier of the target hidden image a, comparing the target hidden image a with the overall parameters of the carrier image A, and mapping and transforming the parameters of the target hidden image a to approximate the parameters of the carrier image A, wherein in order to improve the hiding effect, the area of the carrier image A occupied by the target hidden image a is not more than 20%, and after transforming Tamura texture characteristics, HOG characteristics and LBP characteristic parameters of the target hidden image a, the target hidden image a is more similar to the carrier image A, which is beneficial to hiding the target hidden image a in the carrier image A, wherein the transformation rule of the target hidden image a is encrypted, so that a lawless person is prevented from cracking the target hidden image a in the carrier image A;
s2: respectively dividing the target hidden image a and the carrier image A according to the same pixel sizeCutting to obtain a plurality of image blocks a of the target hidden image a m And a plurality of blocks A of the carrier image A m In the image segmentation, theoretically, the smaller the image segmentation is, the easier the image segmentation is to hide in the carrier image, and considering the actual technical conditions, the segmentation size of the image is selected from the segmentation size of 5-20px × 5-20px, and the segmented image block a is subjected to key pair segmentation m And block A m The position parameter of the image block a is encrypted, so that the image block a is convenient to decrypt m And block A m Carrying out reduction;
s3: acquisition block a m And block A m Characteristic parameter of (1), block a m And block A m The characteristic parameters comprise gray value and color moment, and are respectively and correspondingly stored in the image block a database and the image block A database, and the image block a in a small range m And block A m Due to the small span, the image block a m And block A m The color image can be processed similar to a monochromatic image in a small range, and the smaller the segmented image is, the more single the chromaticity is, so that the gray value and color moment parameters can be collected more conveniently;
s4: deep learning, cross-comparing the data in the image block a database and the image block A database respectively, and determining the data to be compared with the image block a m Is the closest segment A m ；
S5: using block a m Replacing the closest parameter tile A m After determination, use block a m Replacement of Picture Block A m Up to all the picture blocks a m Entering the image block A database through replacement, recording the corresponding relationship of the replacement as a secret key for storage, and recording all the reconstructed image blocks A in the image block A database m And recombining to form an image for output.
The invention can obtain the processing image which is close to the parameters of the carrier image by processing the target hidden image in advance, obtains the image blocks of the carrier image which are similar to the image blocks of the target hidden image by performing fine segmentation on the processed target hidden image and the carrier image, replaces the image blocks of the carrier image by the image blocks of the target hidden image, has higher dispersity when replacing the image blocks of the carrier image, and finally recombines the image blocks of the carrier image which contain the image blocks of the target hidden image to ensure that the target hidden image is dispersed in the carrier image by the small image blocks, thereby realizing the purpose that the target hidden image is hidden in the carrier image, difficult to find and has stronger confidentiality.
When decryption is needed, segmentation is carried out according to the same specification, carrier image blocks with a replacement relation are made to correspond to target hidden image blocks, then decryption is carried out on position parameters of the target hidden image blocks according to a secret key during segmentation, the target hidden images are combined into a complete image, and the Tamura texture features, the HOG features and the LBP feature parameters of the image are restored to obtain the target hidden image.
Finally, it should be noted that: although the present invention has been described in detail with reference to the foregoing embodiments, it will be apparent to those skilled in the art that modifications may be made to the embodiments or portions thereof without departing from the spirit and scope of the invention.
Claims (5)
1. An image hiding method based on deep learning is characterized in that: the method specifically comprises the following operation steps:
s1: importing a target hidden image a and a carrier image A serving as a carrier of the target hidden image a, comparing the overall parameters of the target hidden image a and the carrier image A, mapping and transforming the parameters of the target hidden image a to approximate the parameters of the carrier image A, and encrypting the mapping and transforming parameters;
s2: respectively segmenting the target hidden image a and the carrier image A according to the same pixel size to obtain a plurality of image blocks a of the target hidden image a m And a plurality of blocks A of the carrier image A m ；
S3: acquisition block a m And block A m The characteristic parameters of (a) are set,respectively and correspondingly storing the characteristic parameters in an image block a database and an image block A database;
s4: deep learning, cross-comparing the data in the image block a database and the image block A database, and determining the cross-comparison with the image block a m Is the closest segment A m ；
S5: using block a m Segment A with closest replacement parameters m And all the recombined image blocks A m And recombining to form an image for output.
2. The image hiding method based on deep learning of claim 1, wherein: the overall parameters in step S1 include Tamura texture features, HOG features, and LBP features.
3. The image hiding method based on deep learning of claim 1, wherein: the segmentation size of the image in step S2 is 5-20px × 5-20px.
4. The image hiding method based on deep learning of claim 1, wherein: in step S2, when the image is divided, the divided image block a is divided by the secret key m And block A m The location parameter of (2) is encrypted.
5. The image hiding method based on deep learning of claim 1, wherein: step S3 Block a m And block A m The characteristic parameters include gray values and color moments.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911033271.6A CN110796585B (en) | 2019-10-28 | 2019-10-28 | Image hiding method based on deep learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911033271.6A CN110796585B (en) | 2019-10-28 | 2019-10-28 | Image hiding method based on deep learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110796585A CN110796585A (en) | 2020-02-14 |
| CN110796585B true CN110796585B (en) | 2023-02-28 |
Family
ID=69441678
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911033271.6A Active CN110796585B (en) | 2019-10-28 | 2019-10-28 | Image hiding method based on deep learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110796585B (en) |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| TW201128574A (en) * | 2010-02-11 | 2011-08-16 | Univ Nat Taiwan Science Tech | Image data processing systems for hiding secret information and data hiding methods thereof |
| CN102208022A (en) * | 2010-03-31 | 2011-10-05 | 富士通株式会社 | Shaded character recovery device and method thereof, shaded character recognition device and method thereof |
| US11080809B2 (en) * | 2017-05-19 | 2021-08-03 | Google Llc | Hiding information and images via deep learning |
| CN109547663B (en) * | 2018-09-13 | 2020-06-19 | 华南理工大学 | Improved LSB image steganography method combined with cryptography |
| CN110276753B (en) * | 2019-06-20 | 2021-07-23 | 兰州理工大学 | Target self-adaptive hiding method based on feature space statistical information mapping |
-
2019
- 2019-10-28 CN CN201911033271.6A patent/CN110796585B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN110796585A (en) | 2020-02-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Liu et al. | Reversible data-hiding in encrypted images by redundant space transfer | |
| Ferreira et al. | A review of digital image forensics | |
| Chen et al. | Novel coverless steganography method based on image selection and stargan | |
| Wei et al. | Generative steganography network | |
| Weng et al. | High capacity reversible data hiding in encrypted images using SIBRW and GCC | |
| CN109635576A (en) | A kind of hiding data method and system in the picture | |
| CN112070647A (en) | Reversible high dynamic range image information hiding method | |
| CN106454367A (en) | Method and system for identifying multimedia file | |
| Xiao et al. | Securing the socio-cyber world: Multiorder attribute node association classification for manipulated media | |
| Zeng et al. | Quantum watermark algorithm based on maximum pixel difference and tent map | |
| Sirichotedumrong et al. | Visual security evaluation of learnable image encryption methods against ciphertext-only attacks | |
| Liu et al. | JPEG robust invertible grayscale | |
| Sherine et al. | CMY color spaced-based visual cryptography scheme for secret sharing of data | |
| Li et al. | High-capacity coverless image steganographic scheme based on image synthesis | |
| Wagdarikar et al. | Design and development of a multiobjective cost function for robust video watermarking using wavelet transform | |
| Roselinkiruba et al. | Secure video steganography using key frame and region selection technique | |
| CN110069907A (en) | Big data source tracing method and system based on digital watermarking | |
| CN110796585B (en) | Image hiding method based on deep learning | |
| CN110766597B (en) | Picture camouflage method and device, electronic equipment and storage medium | |
| Davidson et al. | Locating secret messages in images | |
| RU2761417C1 (en) | Method for steganographying a digital image using a graphic digital label and method for decrypting a steganographied image | |
| Umbarkar et al. | COMPARATIVE STUDY OF EDGE BASED LSB MATCHING STEGANOGRAPHY FOR COLOR IMAGES. | |
| Cevik et al. | Improved exploiting modification direction steganography for hexagonal image processing | |
| Qin et al. | Robust steganography via patch-based texture synthesis | |
| Guo et al. | Constructing New Backbone Networks via Space-Frequency Interactive Convolution for Deepfake Detection |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |