CN108600574A - Secret sharing method based on Faster R-CNN ' s - Google Patents
Secret sharing method based on Faster R-CNN ' s Download PDFInfo
- Publication number
- CN108600574A CN108600574A CN201810375942.6A CN201810375942A CN108600574A CN 108600574 A CN108600574 A CN 108600574A CN 201810375942 A CN201810375942 A CN 201810375942A CN 108600574 A CN108600574 A CN 108600574A
- Authority
- CN
- China
- Prior art keywords
- image
- secret
- shadow
- carrier
- faster
- 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.)
- Pending
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N1/00—Scanning, transmission or reproduction of documents or the like, e.g. facsimile transmission; Details thereof
- H04N1/32—Circuits or arrangements for control or supervision between transmitter and receiver or between image input and image output device, e.g. between a still-image camera and its memory or between a still-image camera and a printer device
- H04N1/32101—Display, printing, storage or transmission of additional information, e.g. ID code, date and time or title
- H04N1/32144—Display, printing, storage or transmission of additional information, e.g. ID code, date and time or title embedded in the image data, i.e. enclosed or integrated in the image, e.g. watermark, super-imposed logo or stamp
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N1/00—Scanning, transmission or reproduction of documents or the like, e.g. facsimile transmission; Details thereof
- H04N1/44—Secrecy systems
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention belongs to computer secrets Image Sharing technical fields, provide a kind of secret sharing method based on Faster R CNN ' s.The model considered on choosing carrier image and shadow image Faster R CNN ' s image retrieval robustness and real-time.Public image data are introduced as image querying library, with tradition (t, n) Threshold Secret Image Sharing Scheme is compared, expression ability of the deep learning to feature is utilized in scheme used herein, and sender only needs to transmit an identical query image and the sequence retrieved by Faster RCNN ' s to corresponding all participants.The present invention provides a kind of new method of computer secrets Image Sharing, it can effectively improve the safety of Secret Image and be reduced the acquisition Secret Image time, a kind of new solution is provided for Secret Image Sharing field.
Description
Technical field
The invention belongs to computer secrets images share fields, are related to a kind of based on the (regions of acceleration Faster RCNN ' s
Convolutional neural networks) model Secret Image sharing method.
Background technology
Secret information shares (Secret Information Sharing) technology and initially originates from confidential information management.Silver
Interests loss and authority's deception caused by row system key is administered due to individual are not within minority, and military decision-making power rests in individual
Caused incorrect decision similarly exists in hand.It is secret by professional equipment preservation part by more people that past secret information, which is shared,
Confidential information so that the secret information after recovery is capable of providing accurate information to solve the unreasonable of unidirectional decision.
Nowadays, image information content is abundant, statement image, with the increase of network bandwidth, image information can it is fast
Transmission over networks.But image transmits and is easy to be stolen or be attacked by attacker in a network simultaneously, important Secret Image
Information risk higher during transmission, therefore carry out fast and accurately obtaining Secret Image using image retrieval technologies and share
Technology becomes ideal selection, the quick-searching that can not only preserve secret information distribution, but also be concentrated in image data
Attack of the image information in network transmission can be reduced.For some important Secret Images and cinefilm, directly transmit
Information criminal can be caused to obtain the Secret Image of unauthorized during to reciever, Secret Image shares technical finesse load
Body image (cover image) effect is smooth naturally, processing trace is not noticeable.Theory on computer vision thinks in image
A small amount of information change, human eye are the true and falses of image after None- identified initial carrier image and modification.Faster R-CNN ' s can
Image similar with query image target is retrieved in the database.It is first when progress Secret Image is shared by this theory
First each pixel of secret information will be embedded into carrier image by we, to ensure the quality of carrier image, it would be desirable to
The shadow image (shadow image) linearly combined with carrier information to secret information utilizes carrier image and shadow image
The indistinguishable property of vision proposes that the Secret Image based on Faster R-CNN ' s shares scheme.
The correlative study present situation that Secret Image shares technology is as follows:
Naor proposes first Secret Image with Shamir and shares scheme, still, due to shadow image be it is meaningless,
So when shadow image transmits in unsafe network, it is easy to attacker be attracted to repair meaningless shadow image
The malicious operations such as change, delete.
In order to solve the problems, such as that shadow picture is meaningless, Zhang Zhencheng et al. is proposed chooses significant carrier picture manually
Embedded Secret Image information generates shadow picture, and the shadow image generated is made with carrier image to be visually inseparable
It distinguishes so that the transmission of shadow picture in a network is not easy to cause attacker's note that but for different size of Secret Image
It is reluctant in the same scheme.
Chen, which is proposed, to be solved, using the insertion of two schemes processing different size Secret Image, to increase the robust of system
Property.But timely countermeasure can not be made when transmitting shadow image and encountering attack.
Present invention utilizes Faster R-CNN ' s can be in Image-Database Retrieval to the energy containing same target image
Power can automatically choose carrier image and retrieve corresponding shadow image for participant, meanwhile, use scheme to transmit herein
What it is to sender is no longer shadow image, but without the query image of any modification, therefore, attacker cannot have found query graph
Any exception of picture.The present invention proposes a kind of Secret Image effectively, safe and shares scheme.
Invention content
It is an object of the present invention to provide a kind of Secret Images based on FasterR-CNN ' s models to share model, can solve
The shortcomings that certainly secure context is not enough the considerations of existing algorithm no longer transmits risky shadow image in transmission process, passes
Defeated is unmodified query image, avoid attacker by query image compared with the true picture that public data is concentrated and
It was found that in transmission process image exception.The data of 5,063 pictures of Oxford Buildings of a safety are set herein
Collection, in addition to this, in order to ensure the safety of shadow image, once find that attack, sender update image data base at once.
Technical scheme of the present invention:
Secret sharing method based on Faster R-CNN ' s, steps are as follows:
Step 1):Sender (Sender) selects a query image Q arbitrarily from image data base first, uses
The Secret Image of Faster RCNN ' s, which is shared algorithm and retrieved from image data base, obtains n carrier image C;If current secret
Image is s, is shared Secret Image s to n carrier images using linear integer secret sharing (LISS), obtains n shadow figures
Picture;
Step 1.1):First by each pixel c of carrier image CkIt is handled, b=ckModm, wherein k=
{ 1 ..., Dl*Dh, Dl*DhIt is the length and width of carrier image, m is the prime number that sender chooses;
Step 1.2):Distribution M is built according to access strategy using LISS so that each variable M in access strategy
Statement, secret sharing matrix ρ=(s, ρ2,...,ρe), ρiIt is random integer;Pass through M ρ=(si1,si2,...,sid) obtain
Pre-process secret information sij, the value range of i and j are i={ 1 ..., logm255 }, j={ 1 ..., d }, d are participant's
Number enables dk=sij-b;
Step 1.3):D is changed according to formula (1)kObtain d'k
Step 1.4):According to obtained d'kChange the pixel value of carrier image, p'k=ck+d'k;
Step 1.5):Pixel value overflows in order to prevent, and the pixel value of carrier image is adjusted according to formula (2)
Step 1.6):Repeat above step until all Secret Image s pixel values all share into carrier image, and
Generate shadow image;
Step 2):Original data library is adjusted, obtained n shadow images are replaced into the n in original image data base
Open carrier image;
Step 3):It reuses Faster RCNN ' s and retrieves n shadow images, record n at this time from image data base
Shadow image position sequence is denoted as ti=(1 ..., n);Sender is that each participant is packaged portion<Query image, position
Set sequence>It is right, then use Advanced Encryption Standard (AES) to its encrypting and transmitting to recipient;
Step 4):Each participant is according to obtaining<Query image, > pairs of position sequence, by query image and position-order
In row input value Faster R-CNN ' s, the unique shadow image for belonging to each participant is obtained;Only when there is t participants
Or more than t participants cooperate to restore final Secret Image information s;
When participant gets shadow image, the step of Restore Secret image, is as follows:
Step 4.1):sijIt is calculated by formula (3), m is the prime number that sender chooses;
Step 4.2):T or the shadow image cooperation more than t participants obtain final Secret Image information, select
One suitable subset sA∈sijSecret is calculated, Secret Image information obtains by formula (4):
Beneficial effects of the present invention:Introduce image data set, participant only needs to use identical inquiry picture and respective
Position sequence utilize FasterR-CNN ' s query results.Compared to other image repair schemes based on (t, n) thresholding, when
For shadow image when transmission process is under attack, scheme herein can adjust image data collection immediately, to make attacker attack
Hit failure.Scheme herein can handle different size of Secret Image, but becoming larger with Secret Image, mean square error MSE
It can become smaller.The present invention provides a kind of new methods of computer secrets images share, can effectively improve the safety of Secret Image
Property, provide a kind of new solution for the trouble free service of Secret Image.
Description of the drawings
Fig. 1 is that the Secret Image for the technique algorithm that the present invention uses shares process schematic.
Fig. 2 is Faster R-CNN ' the s algorithm schematic diagrames that the present invention uses.
During Fig. 3 is of the invention, with becoming larger for embedded Secret Image, the change of divergence between carrier image and shadow image
Show schematic diagram.
Fig. 4 is in the present invention, with becoming larger for embedded Secret Image, the displaying schematic diagram of the secret time change of insertion.
Specific implementation mode
To make the objectives, technical solutions, and advantages of the present invention clearer, below to the specific implementation mode of the present invention
It is described in further detail.
Herein in pycharm, in the PC machine of GTX750 video memorys, " FasterR- is based on using proposed in this paper by 3.40GHz
The Secret Image of CNN ' s is shared " algorithm, share algorithm then in conjunction with the Secret Image of linear integer model, completes entire secret
The flow of images share.LISS is utilized to build distribution M according to access strategy herein so that each variable in access strategy can
To be stated with M, secret sharing matrix ρ=(s, ρ2,...,ρe),ρiIt is random integer.It can pass through M ρ=(s hereini1,
si2,...,sid) pretreatment secret information is obtained, d is the number of participant, and it is as follows that specific Secret Image shares details:
Step 1):First by each pixel c of carrier image CkIt is handled, b=ckMod m, wherein k=
{ 1 ..., Dl*Dh, Dl*DhIt is the length and width of carrier image, m is the prime number that sender chooses.
Step 2):Enable dk=sij- b, sijIt is by M ρ=(si1,si2,...,sid) obtain, the value range of i and j are i=
{1,...,logm255 }, j={ 1 ..., d }.
Step 3):D is changed according to following equation 1kObtain d'k。
Step 4):According to obtained d'kChange the pixel value of carrier image, p'k=ck+d'k。
Step 5):Pixel value overflows in order to prevent, and the pixel value of carrier image is adjusted according to formula 2.
Step 6):Repeat all of above step until all Secret Image pixel values all share into carrier image, and
And generate shadow image.
Assuming that the system of (2, a 4) thresholding, access strategy is Γ={ P1,P2},{P3,P4, according to LISS, can obtain
To distribution matrixM=11, it is assumed that the pixel value of first secret information is 189, converts it to 11
System, s1={ 1,6,2 }11, to s1First value choose ρ=(1,4,7)T, according to M ρ=(si1,si2,...,sid), it obtains
s1Four parts can be distributed, be expressed as s1={ 5,4,8,7 }T, by s11=5 are embedded into carrier image, it is assumed that carrier image first
Pixel value is ck=168, use b=ckB=3, d is calculated in mod mk=2.According to step 3, d is updatedk'=dk=2, most
Eventually, according to step 4 and 5, p'k=ck+d'k=170.In this way, by first pixel value c of original carrier imagek=168 is embedding
Enter secret information and obtains the pixel value p' of shadow imagek=170.
When participant gets shadow image, the details of Restore Secret image is as follows:
Step 1) sijIt can be calculated by formula 3, m is one prime number of sender.
Step 2) t or shadow image cooperation more than t participants can obtain final secret, select a conjunction
Suitable subset sA∈sijSecret is calculated, Secret Image information can obtain by following formula.
According in ciphering process it is assumed that p'k=170, according to formula 3, obtain s11=5, secret process is shared in verification
Correctness.
Carrier picture is obtained by Faster R-CNN ' s, in order to ensure that image remains able to obtain in retrieval, herein
The gray level image for obtaining carrier image is operated, and not only be ensure that visual unidentifiable property, is also ensured Faster R-
CNN ' s can retrieve the image again.
For the algorithm of this paper, for the efficiency of objective relatively different size Secret Image insertion, i.e., Secret Image is at this
The insertion time of text and embedding capacity.Herein, under the premise of the safety for ensureing image data set, they are objectively surveyed
Amount.For this purpose, introducing the common technical parameter of objective measurement:Mean square deviation (MSE).
Wherein f (x, y) is shadow image, f0(x, y) is carrier image, and the size of image is M × N.Present invention utilizes
Faster R-CNN ' s obtain the carrier of shadow image using image data base as the tool for obtaining shadow image, can
When person under attack attacks, data set is adjusted in time, to avoid secret caused by channel is dangerous in transmission process from losing
It loses.
It is specific embodiments of the present invention and the technical principle used described in above, if conception under this invention institute
The change of work when the spirit that generated function is still covered without departing from specification and attached drawing, should belong to the present invention's
Protection domain.
Claims (1)
1. a kind of secret sharing method based on Faster R-CNN ' s, which is characterized in that steps are as follows:
Step 1):Sender selects a query image Q arbitrarily from image data base first, uses Faster RCNN ' s'
Secret Image, which is shared algorithm and retrieved from image data base, obtains n carrier image C;If current secret image is s, using linear
Integer secret sharing shares Secret Image s to n carrier images, obtains n shadow images;
Step 1.1):First by each pixel c of carrier image CkIt is handled, b=ckMod m, wherein k=1 ...,
Dl*Dh, Dl*DhIt is the length and width of carrier image, m is the prime number that sender chooses;
Step 1.2):Distribution M is built according to access strategy using LISS so that each variable in access strategy is stated with M,
Secret sharing matrix ρ=(s, ρ2,...,ρe), ρiIt is random integer;Pass through M ρ=(si1,si2,...,sid) obtain pre- place
Manage secret information sij, the value range of i and j are i={ 1 ..., logm255 }, j={ 1 ..., d }, d are the number of participant,
Enable dk=sij-b;
Step 1.3):D is changed according to formula (1)kObtain d'k
Step 1.4):According to obtained d'kChange the pixel value of carrier image, p'k=ck+d'k;
Step 1.5):Pixel value overflows in order to prevent, and the pixel value of carrier image is adjusted according to formula (2)
Step 1.6):Above step is repeated until all Secret Image s pixel values are all shared into carrier image, and generate
Shadow image;
Step 2):Original data library is adjusted, obtained n shadow images, which are replaced the n in original image data base, to be carried
Body image;
Step 3):It reuses Faster RCNN ' s and retrieves n shadow images from image data base, n opens shadow to record at this time
Image position sequence is denoted as ti=(1 ..., n);Sender is that each participant is packaged portion < query images, position-order
Then > pairs of row use Advanced Encryption Standard to its encrypting and transmitting to recipient;
Step 4):For each participant according to obtained < query images, > pairs of position sequence is defeated by query image and position sequence
Enter in value Faster R-CNN ' s, obtains the unique shadow image for belonging to each participant;Only when there is t participants or more
Cooperate to restore final Secret Image information s in t participants;
When participant gets shadow image, the step of Restore Secret image, is as follows:
Step 4.1):sijIt is calculated by formula (3), m is the prime number that sender chooses;
Step 4.2):T or the shadow image cooperation more than t participants obtain final Secret Image information, select one
Suitable subset sA∈sijSecret is calculated, Secret Image information obtains by formula (4):
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810375942.6A CN108600574A (en) | 2018-04-20 | 2018-04-20 | Secret sharing method based on Faster R-CNN ' s |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810375942.6A CN108600574A (en) | 2018-04-20 | 2018-04-20 | Secret sharing method based on Faster R-CNN ' s |
Publications (1)
Publication Number | Publication Date |
---|---|
CN108600574A true CN108600574A (en) | 2018-09-28 |
Family
ID=63609269
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810375942.6A Pending CN108600574A (en) | 2018-04-20 | 2018-04-20 | Secret sharing method based on Faster R-CNN ' s |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108600574A (en) |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111416915A (en) * | 2020-02-21 | 2020-07-14 | 成都信息工程大学 | Image secret sharing method based on threshold reduction and application |
CN114612317A (en) * | 2022-02-07 | 2022-06-10 | 中国人民解放军国防科技大学 | Secret image sharing method and system for resisting mean filtering |
CN114782563A (en) * | 2022-05-26 | 2022-07-22 | 中国人民解放军国防科技大学 | Secret image sharing method and system for JPEG image |
CN115131253A (en) * | 2022-05-26 | 2022-09-30 | 中国人民解放军国防科技大学 | Secret image sharing method and system for resisting JPEG recompression |
CN115330890A (en) * | 2022-05-26 | 2022-11-11 | 中国人民解放军国防科技大学 | Secret image sharing method and system based on global adjustment and stable block conditions |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101882992A (en) * | 2010-06-21 | 2010-11-10 | 中国农业大学 | Threshold secret information distribution and recovery method and device |
JP2011186129A (en) * | 2010-03-08 | 2011-09-22 | Dainippon Printing Co Ltd | Encryption pattern generating method and encryption recording medium |
CN103761702A (en) * | 2014-01-09 | 2014-04-30 | 太原科技大学 | Image hiding and authentication method based on secret sharing |
CN105830086A (en) * | 2013-11-06 | 2016-08-03 | 净睿存储股份有限公司 | Data protection in a storage system using external secrets |
CN106339975A (en) * | 2016-10-10 | 2017-01-18 | 中科院成都信息技术股份有限公司 | Digital image secret sharing method, restoration method and devices |
CN106980791A (en) * | 2017-03-30 | 2017-07-25 | 上海海洋大学 | Weighted Threshold ocean remote sensing image secret sharing scheme and method are extended under cloud environment |
-
2018
- 2018-04-20 CN CN201810375942.6A patent/CN108600574A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2011186129A (en) * | 2010-03-08 | 2011-09-22 | Dainippon Printing Co Ltd | Encryption pattern generating method and encryption recording medium |
CN101882992A (en) * | 2010-06-21 | 2010-11-10 | 中国农业大学 | Threshold secret information distribution and recovery method and device |
CN105830086A (en) * | 2013-11-06 | 2016-08-03 | 净睿存储股份有限公司 | Data protection in a storage system using external secrets |
CN103761702A (en) * | 2014-01-09 | 2014-04-30 | 太原科技大学 | Image hiding and authentication method based on secret sharing |
CN106339975A (en) * | 2016-10-10 | 2017-01-18 | 中科院成都信息技术股份有限公司 | Digital image secret sharing method, restoration method and devices |
CN106980791A (en) * | 2017-03-30 | 2017-07-25 | 上海海洋大学 | Weighted Threshold ocean remote sensing image secret sharing scheme and method are extended under cloud environment |
Non-Patent Citations (1)
Title |
---|
J. LI, N. WANG, Z. WANG, H. LI, C. CHANG AND H. WANG: "《New Secret Sharing Scheme Based》", 《IEEE ACCESS》 * |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111416915A (en) * | 2020-02-21 | 2020-07-14 | 成都信息工程大学 | Image secret sharing method based on threshold reduction and application |
CN114612317A (en) * | 2022-02-07 | 2022-06-10 | 中国人民解放军国防科技大学 | Secret image sharing method and system for resisting mean filtering |
CN114612317B (en) * | 2022-02-07 | 2023-04-07 | 中国人民解放军国防科技大学 | Secret image sharing method and system for resisting mean filtering |
CN114782563A (en) * | 2022-05-26 | 2022-07-22 | 中国人民解放军国防科技大学 | Secret image sharing method and system for JPEG image |
CN115131253A (en) * | 2022-05-26 | 2022-09-30 | 中国人民解放军国防科技大学 | Secret image sharing method and system for resisting JPEG recompression |
CN115330890A (en) * | 2022-05-26 | 2022-11-11 | 中国人民解放军国防科技大学 | Secret image sharing method and system based on global adjustment and stable block conditions |
CN114782563B (en) * | 2022-05-26 | 2023-07-14 | 中国人民解放军国防科技大学 | Secret image sharing method and system for JPEG image |
CN115330890B (en) * | 2022-05-26 | 2023-12-12 | 中国人民解放军国防科技大学 | Secret image sharing method and system based on global adjustment and stable block conditions |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108600574A (en) | Secret sharing method based on Faster R-CNN ' s | |
CN105847629B (en) | A kind of reversible information hidden method of encrypted image | |
CN108566500A (en) | The reversible hidden method of adapting to image encrypted domain based on Hybrid Encryption mechanism | |
CN112417467B (en) | Image encryption method based on anti-neurocryptography and SHA control chaos | |
Blesswin et al. | Enhanced semantic visual secret sharing scheme for the secure image communication | |
Islam et al. | Application of homomorphism to secure image sharing | |
CN106530197A (en) | Image encryption method based on Kent mapping and generalized Gray codes | |
Li et al. | Fidelity preserved data hiding in encrypted highly autocorrelated data based on homomorphism and compressive sensing | |
CN105897709A (en) | User attribute encryption and decryption method of non-monotonic access structure in distributed network | |
CN111581648A (en) | Method of federal learning to preserve privacy in irregular users | |
Wang et al. | TPE-ISE: approximate thumbnail preserving encryption based on multilevel DWT information self-embedding | |
Kaushik et al. | A pixel-based digital medical images protection using genetic algorithm | |
Qin et al. | Application of CSK encryption algorithm in video synergic command systems | |
Zhou et al. | A survey of security aggregation | |
Kayani et al. | Efficient Region of Interest Encryption Based on a New Chaotic Map | |
CN114398671A (en) | Privacy calculation method, system and readable storage medium based on feature engineering IV value | |
Rani et al. | Region based data hiding in medical images | |
Jayahari Prabhu | Multimodal medical imaging security using hybridization of honey encryption algorithm with binary particle swarm optimization | |
Theramban et al. | Colour image encryption algorithm with quantum coding | |
Cheng et al. | Information hiding mechanism based on QR code and information sharing algorithm | |
Radke et al. | Review of Image Security Approaches: Concepts, issues, challenges and applications | |
Sajitha et al. | Analysis of Various Visual Cryptographic Techniques and Their Issues Based on Optimization Algorithms | |
Brabin et al. | Region based visual cryptography scheme for color images | |
Shyamala | Chaos based image encryption scheme | |
Rani et al. | Design and Implementation of Image Encryption Algorithm Using Chaos |
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 | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20180928 |
|
WD01 | Invention patent application deemed withdrawn after publication |