CN111709867A - Novel full convolution network-based equal modulus vector decomposition image encryption analysis method - Google Patents
Novel full convolution network-based equal modulus vector decomposition image encryption analysis method Download PDFInfo
- Publication number
- CN111709867A CN111709867A CN202010524057.7A CN202010524057A CN111709867A CN 111709867 A CN111709867 A CN 111709867A CN 202010524057 A CN202010524057 A CN 202010524057A CN 111709867 A CN111709867 A CN 111709867A
- Authority
- CN
- China
- Prior art keywords
- training
- encryption
- analysis
- block
- image
- 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.)
- Granted
Links
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
- G06T1/005—Robust watermarking, e.g. average attack or collusion attack resistant
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Abstract
The invention provides an equal-modulus vector decomposition image encryption analysis method based on a novel full convolution network. The method comprises the following steps: the method comprises four parts of an encryption system based on equimode vector decomposition, a network model of encryption analysis, network training and encryption system analysis. The designed encryption analysis network model is trained by inputting a plaintext-ciphertext pair, and then a ciphertext image is input through the trained encryption analysis network model, so that an attack analysis result, namely a recovered high-quality plaintext image, can be obtained. Compared with the traditional attack method, the provided encryption analysis method can realize effective encryption analysis without knowing an encryption key or a private key and other encryption system parameters and the like, and can recover and recover a high-quality plaintext image; the deep learning method has short training time, and the training speed is improved by 7 times compared with the traditional method; the method has better generalization capability, and can adopt one image library for training and another image library for successful test; finally, the method has better robustness to noise and clipping in transmission.
Description
Technical Field
The invention relates to the technical field of information security and information optics, in particular to an image encryption analysis method.
Background
With the advent of the information age, information security has received more and more attention. Images tend to provide rich information, so image encryption becomes a crucial issue. In 2015, tsai proposes an image encryption method based on equal modulus separation (EMD), in which a two-dimensional vector is decomposed into two-dimensional vectors, and a safe one-way trapdoor function is provided for image encryption, so that the method is widely applied to gray image encryption. Deep learning is also used for image encryption analysis as an effective method for feature extraction, but at present, only an encryption system based on double random phase coding and the like are broken through by a deep learning method, and for an asymmetric encryption method with higher encryption security, such as an image encryption method based on equimode vector decomposition, an encryption analysis method of the deep learning method is not reported. And the deep learning network needs to be redesigned to adapt to the attack analysis method of a new encryption system. Therefore, the deep learning-based isomode vector decomposition image encryption analysis still has great research potential.
Disclosure of Invention
The invention provides an equimode vector decomposition image encryption analysis method based on a novel full convolution network, aiming at the problem that the traditional deep learning network can not attack and analyze the image encryption based on equimode vector decomposition. The method comprises the following steps: the method comprises four parts of an encryption system based on equimode vector decomposition, a network model of encryption analysis, network training and encryption system analysis.
The encryption system based on the equimodular vector decomposition comprises an encryption process and a decryption process, wherein the encryption process is specifically described as follows: the image to be encrypted is first subjected to the complex number and domain transformation, and then an equal modulus vector decomposition method is adopted to obtain a ciphertext and a private key, wherein the encryption process is expressed as [ C, P ]]=EMD{DT[F(I, M)]That is, a decryption process is an inverse process of the encryption process, and may be expressed as I = | IDT (C + P) | gaming2Wherein, I is the image to be encrypted, and M is in the range of [0, 2 pi ]]The random phase plate of (1), F (I, M) is a complex number, and the specific process can be expressed as F (I, M) = sqrt (I) × exp (I × M), DT [.]For domain transformation, EDM is equal modulus vector decomposition, IDT [.]For inverse domain transformation, | - | is modulo operation, C and P represent the ciphertext and the private key, respectively, with the modulo of each pixel being equal.
The network model of the encryption analysis is a novel full convolution network, belongs to a deep learning neural network, and comprises an input layer, a hidden layer and an output layer, wherein the input layer is a ciphertext image of 32 multiplied by 1, and the output layer is an encryption analysis result image of 32 multiplied by 1; the hidden layer comprises 9 layers of networks in total, the first 4 layers are down-sampling layers, the sizes of the down-sampling layers are 16 multiplied by 28, 8 multiplied by 36, 4 multiplied by 38 and 2 multiplied by 42 in sequence, the first 4 layers and the input layer are connected by down-sampling blocks, the second 5 layers are up-sampling layers, the sizes of the up-sampling layers are 2 multiplied by 48, 4 multiplied by 48, 8 multiplied by 46, 16 multiplied by 48 and 32 multiplied by 48, the last 5 layers and the first 4 layers are connected by convolution blocks, the layers of the last 5 layers are connected by up-sampling blocks, and the last 5 layers and the output layer are connected by convolution blocks; the downsampling block comprises 2 continuous convolution blocks, the convolution kernel size of the convolution blocks is 3 multiplied by 3, and the step sizes are 1 and 2 respectively; the convolution block comprises convolution operation, regularization and activation units, the convolution kernel size of the first convolution block is 3 x 3, and the convolution kernel size of the second convolution block is 1 x 1.
The network training is a training process of a deep learning neural network model, firstly, data preprocessing is carried out, the sizes of images to be encrypted are unified to be 32 multiplied by 32, pixel values are normalized to be between-1 and 1, the images of a selected training database are subjected to the constant modulus vector decomposition encryption process to obtain ciphertext and plaintext true value pairs, the ciphertext and plaintext true value pairs are divided into a training set and a testing set, the training set is input into the network model to be trained, and the trained network model is obtained through training.
The encryption system analysis is to input any ciphertext of the test set into the trained network model, so as to obtain a corresponding plaintext, and simultaneously, the ciphertext can be compared with a true value of the corresponding plaintext to test the encryption analysis effect of the novel full convolution network.
The domain transformation includes but is not limited to: the transformation between the spatial domain and the frequency domain can be realized optically, such as Fourier transformation, or Fresnel diffraction transformation corresponding to the optical diffraction process.
The method has the beneficial effects that: compared with the traditional attack method, the provided encryption analysis method can realize effective encryption analysis without knowing an encryption key or a private key and other encryption system parameters and the like, and can recover and recover a high-quality plaintext image; the deep learning method has short training time, and the training speed is improved by 7 times compared with the traditional method; the method has better generalization capability, and can adopt one image library for training and another image library for successful test; finally, the method has better robustness to noise and clipping in transmission.
Drawings
Fig. 1 is a schematic diagram of the structure of the novel full convolution network of the invention.
FIG. 2 is a schematic diagram of an encryption process of the encryption system based on the equimod vector decomposition according to the present invention.
FIG. 3 is a graph of loss function convergence curves and attack analysis test accuracy.
FIG. 4 is a diagram showing the result of the encryption analysis performed by the encryption method of 2 domain transformations according to the present invention.
Detailed Description
An exemplary embodiment of the method for analyzing the image encryption by using the uniform modulus vector decomposition based on the novel full convolution network according to the present invention is described in detail below, and the method is further described in detail. It is to be noted that the following examples are given for the purpose of illustration only and are not to be construed as limiting the scope of the present invention, and that the skilled person will be able to make insubstantial modifications and adaptations of the method based on the teachings of the method described above and still fall within the scope of the invention.
The invention provides an equal modulus vector decomposition image encryption analysis method based on a novel full convolution network, which comprises the following steps: the method comprises four parts of an encryption system based on equimode vector decomposition, a network model of encryption analysis, network training and encryption system analysis.
The specific network structure is shown in fig. 1, and the schematic diagram of the encryption process is shown in fig. 2.
The encryption system based on the equimodular vector decomposition comprises an encryption process and a decryption process, wherein the encryption process is specifically described as follows: the image to be encrypted is first subjected to the complex number and domain transformation, and then an equal modulus vector decomposition method is adopted to obtain a ciphertext and a private key, wherein the encryption process is expressed as [ C, P ]]=EMD{DT[F(I, M)]The decryption process is an encryption processCan be expressed as I = | IDT (C + P) converter2Wherein, I is the image to be encrypted, and M is in the range of [0, 2 pi ]]The random phase plate of (1), F (I, M) is a complex number, and the specific process can be expressed as F (I, M) = sqrt (I) × exp (I × M), DT [.]For domain transformation, EDM is equal modulus vector decomposition, IDT [.]For inverse domain transformation, | - | is modulo operation, C and P represent the ciphertext and the private key, respectively, with the modulo of each pixel being equal.
The network model of the encryption analysis is a novel full convolution network, belongs to a deep learning neural network, and comprises an input layer, a hidden layer and an output layer, wherein the input layer is a ciphertext image of 32 multiplied by 1, and the output layer is an encryption analysis result image of 32 multiplied by 1; the hidden layer comprises 9 layers of networks in total, the first 4 layers are down-sampling layers, the sizes of the down-sampling layers are 16 multiplied by 28, 8 multiplied by 36, 4 multiplied by 38 and 2 multiplied by 42 in sequence, the first 4 layers and the input layer are connected by down-sampling blocks, the second 5 layers are up-sampling layers, the sizes of the up-sampling layers are 2 multiplied by 48, 4 multiplied by 48, 8 multiplied by 46, 16 multiplied by 48 and 32 multiplied by 48, the last 5 layers and the first 4 layers are connected by convolution blocks, the layers of the last 5 layers are connected by up-sampling blocks, and the last 5 layers and the output layer are connected by convolution blocks; the downsampling block comprises 2 continuous convolution blocks, the convolution kernel size of the convolution blocks is 3 multiplied by 3, and the step sizes are 1 and 2 respectively; the convolution block comprises convolution operation, regularization and activation units, the convolution kernel size of the first convolution block is 3 x 3, and the convolution kernel size of the second convolution block is 1 x 1.
The network training is a training process of a deep learning neural network model, firstly, data preprocessing is carried out, the sizes of images to be encrypted are unified to be 32 multiplied by 32, pixel values are normalized to be between-1 and 1, the images of a selected training database are subjected to the constant modulus vector decomposition encryption process to obtain ciphertext and plaintext true value pairs, the ciphertext and plaintext true value pairs are divided into a training set and a testing set, the training set is input into the network model to be trained, and the trained network model is obtained through training.
The encryption system analysis is to input any ciphertext of the test set into the trained network model, so as to obtain a corresponding plaintext, and simultaneously, the ciphertext can be compared with a true value of the corresponding plaintext to test the encryption analysis effect of the novel full convolution network.
The domain transformation includes but is not limited to: the transformation between the spatial domain and the frequency domain can be realized optically, such as Fourier transformation, or Fresnel diffraction transformation corresponding to the optical diffraction process.
In the present example, the image databases used are the MNIST and Fashinon-MNIST databases. In the proposed full convolution network, the activation function of the activation unit is ReLU, and the learning rate is dynamically set: the first 5 rounds were 0.01, the next 5 rounds were 0.005 and the last 5 rounds were 0.001. The implementation environment of the deep learning network is a Baidu fly-plasma framework, Inviad Tesla V100 GPU. The loss function adopts mean square error, and the accuracy adopts the correlation of adjacent pixels. Fig. 3 is a graph showing the loss function convergence curve and the attack analysis test accuracy, fig. 3 (I) is a result of the cryptanalysis of the fourier transform-based domain transform, and fig. 3 (II) is a result of the cryptanalysis of the fresnel diffraction transform-based domain transform. It can be seen from the figure that the encryption analysis of the 2 domain transformation encryption methods has fast convergence rate, i.e. short training time and high accuracy, i.e. high quality of recovered plaintext. Fig. 4 is a diagram showing the result of the cryptanalysis performed by the proposed cryptanalysis network on 2 kinds of domain-transformed cryptanalysis methods, where fig. 4(I) is a cryptanalysis result image of a domain transform based on fourier transform, and fig. 4(II) is a cryptanalysis result image of a domain transform based on fresnel diffraction transform. Fig. 4(i), 4(ii), and 4(iii) are images of plaintext true value, ciphertext, and corresponding attack analysis result, respectively, fig. 4(a) -4(c) are results of MNIST training and MNIST testing, and fig. 4(d) -4(f) are results of MNIST training and fast-MNIST testing. The result shows that the proposed method has better generalization capability to the database.
Claims (2)
1. An equal modulus vector decomposition image encryption analysis method based on a novel full convolution network is characterized by comprising the following steps: four parts of an encryption system based on equimode vector decomposition, a network model of encryption analysis and network training and encryption system analysis; the encryption system based on the equimodular vector decomposition comprises an encryption process and a decryption process, wherein the encryption process is specifically describedThe method comprises the following steps: the image to be encrypted is first subjected to the complex number and domain transformation, and then an equal modulus vector decomposition method is adopted to obtain a ciphertext and a private key, wherein the encryption process is expressed as [ C, P ]]=EMD{DT[F(I, M)]That is, a decryption process is an inverse process of the encryption process, and may be expressed as I = | IDT (C + P) | gaming2Wherein, I is the image to be encrypted, and M is in the range of [0, 2 pi ]]The random phase plate of (1), F (I, M) is a complex number, and the specific process can be expressed as F (I, M) = sqrt (I) × exp (I × M), DT [.]For domain transformation, EDM is equal modulus vector decomposition, IDT [.]The method comprises the steps of carrying out inverse domain transformation, | i.e. modulus operation, | i.e. C and P respectively represent a ciphertext and a private key, the modulus of each pixel of the ciphertext and the private key are respectively equal, the network model for encryption analysis is a novel full convolution network and belongs to a deep learning neural network, the network model comprises an input layer, a hidden layer and an output layer, the input layer is a ciphertext image of 32 × 32 × 1, the output layer is an encryption analysis result image of 32 × 032 ×, the hidden layer totally comprises 9 layers, the front 4 layer is a downsampling layer, the size of the downsampling layer is 16 × ×, 8 × ×, 4 5864 ×, 2 × ×, the layers of the front 4 layer and the input layer are connected by a downsampling block, the last 5 layer is an upsampling layer, the size of the downsampling layer is 2 × 2 049 6868, 4 × ×, 8 × ×, 16 × 6867, 32 36516, 32 and 72, the size of the training block is a training block 72, the training data is obtained by connecting the training block 72, the training process of the training block for training the training block, the ciphertext and the training data are input to be a training data to be a training result image, the convolution data, the training data is obtained by a training process of the training block comprising the training process of the training block 363, the training block 72, the training process of the training block, the training block for training process of the training a training block, the training process of the training block, the training block of the training block, the ciphertext and the training data of the training block, the training data of the training process of the training block 363, the training process of the training block is unified convolution analysis process of the training block, the training process of the training block is a training block, the training process of the training processCorresponding plaintext can be obtained, and meanwhile, the encryption analysis effect of the novel full convolution network can be tested by comparing with the true value of the corresponding plaintext.
2. The novel full convolutional network based encryption analysis method for equimodular vector decomposition images as claimed in claim 1, wherein the domain transformation includes but is not limited to: the transformation between the spatial domain and the frequency domain can be realized optically, such as Fourier transformation, or Fresnel diffraction transformation corresponding to the optical diffraction process.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010524057.7A CN111709867B (en) | 2020-06-10 | 2020-06-10 | Novel full convolution network-based equal-modulus vector decomposition image encryption analysis method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010524057.7A CN111709867B (en) | 2020-06-10 | 2020-06-10 | Novel full convolution network-based equal-modulus vector decomposition image encryption analysis method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111709867A true CN111709867A (en) | 2020-09-25 |
| CN111709867B CN111709867B (en) | 2022-11-25 |
Family
ID=72539471
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010524057.7A Active CN111709867B (en) | 2020-06-10 | 2020-06-10 | Novel full convolution network-based equal-modulus vector decomposition image encryption analysis method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111709867B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112241931A (en) * | 2020-10-22 | 2021-01-19 | 海南大学 | Frequency domain digital veil design method for image easiness |
| CN112351151A (en) * | 2020-11-10 | 2021-02-09 | 广东工业大学 | FCN-based image information encryption method and device |
| CN116310719B (en) * | 2023-02-10 | 2024-04-19 | 中国人民解放军军事科学院国防科技创新研究院 | Time-frequency domain-based optical diffraction complex model training method and image processing method |
Citations (24)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2002057909A (en) * | 2000-06-20 | 2002-02-22 | Ricoh Co Ltd | Method and system for processing color medium tone, and recording medium |
| WO2004023715A1 (en) * | 2002-09-03 | 2004-03-18 | The Regents Of The University Of California | Block cipher mode of operation for constructing a wide-blocksize block cipher from a conventional block cipher |
| EP1833237A2 (en) * | 2006-03-07 | 2007-09-12 | Kabushiki Kaisha Toshiba | Retrieval of information embedded in an image |
| CN101185122A (en) * | 2005-06-03 | 2008-05-21 | 皇家飞利浦电子股份有限公司 | Homomorphic encryption for secure watermarking |
| WO2010064333A1 (en) * | 2008-12-05 | 2010-06-10 | Kabushiki Kaisha Toshiba | Image processing method and image processing apparatus |
| US20130063550A1 (en) * | 2006-02-15 | 2013-03-14 | Kenneth Ira Ritchey | Human environment life logging assistant virtual esemplastic network system and method |
| CN103442156A (en) * | 2013-08-20 | 2013-12-11 | 宁波大学 | Image steganography and secret information extraction method based on intelligent mobile terminal |
| WO2015052351A1 (en) * | 2013-10-11 | 2015-04-16 | Mauna Kea Technologies | Method for characterizing images acquired through a video medical device |
| WO2015054666A1 (en) * | 2013-10-10 | 2015-04-16 | Board Of Regents, The University Of Texas System | Systems and methods for quantitative analysis of histopathology images using multi-classifier ensemble schemes |
| CN106600517A (en) * | 2016-11-11 | 2017-04-26 | 陕西师范大学 | Multi-carrier secret image separate storage and reconstruction method based on EMD-3 |
| CN106845339A (en) * | 2016-12-13 | 2017-06-13 | 电子科技大学 | A kind of mobile phone individual discrimination method based on bispectrum and EMD fusion features |
| EP3242239A1 (en) * | 2016-05-02 | 2017-11-08 | Scantrust SA | Optimized authentication of graphic authentication code |
| CN108959478A (en) * | 2018-06-21 | 2018-12-07 | 中南林业科技大学 | Ciphertext image search method and system under a kind of cloud environment |
| US20190028608A1 (en) * | 2017-07-24 | 2019-01-24 | Samsung Electronics Co., Ltd. | Electronic device and method for controlling the electronic device |
| WO2019050994A1 (en) * | 2017-09-05 | 2019-03-14 | Gritstone Oncology, Inc. | Neoantigen identification for t-cell therapy |
| US20190113973A1 (en) * | 2012-09-14 | 2019-04-18 | Interaxon Inc | Systems and methods for collecting, analyzing, and sharing bio-signal and non-bio-signal data |
| CN109784023A (en) * | 2018-11-28 | 2019-05-21 | 西安电子科技大学 | Stable state vision inducting brain electricity personal identification method and system based on deep learning |
| CN109858264A (en) * | 2019-01-22 | 2019-06-07 | 四川大学 | A kind of color image encipher-decipher method of the moulds resolutions of vectors such as score stochastic transformation domain |
| CN110071798A (en) * | 2019-03-21 | 2019-07-30 | 深圳大学 | A kind of equivalent key acquisition methods, device and computer readable storage medium |
| CN110269625A (en) * | 2019-05-31 | 2019-09-24 | 杭州电子科技大学 | A kind of electrocardio authentication method and system of novel multiple features fusion |
| CN110334881A (en) * | 2019-07-17 | 2019-10-15 | 深圳大学 | A kind of Financial Time Series Forecasting method based on length memory network and depth data cleaning, device and server |
| US20190372754A1 (en) * | 2017-02-15 | 2019-12-05 | Lg Electronics Inc. | Apparatus and method for generating ciphertext data with maintained structure for analytics capability |
| CN110706144A (en) * | 2019-09-17 | 2020-01-17 | 四川大学 | Optical color image encryption method based on spiral phase transformation and equal-mode decomposition |
| WO2020103411A1 (en) * | 2018-11-22 | 2020-05-28 | 九牧厨卫股份有限公司 | Fall detection method, device, and system |
-
2020
- 2020-06-10 CN CN202010524057.7A patent/CN111709867B/en active Active
Patent Citations (24)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2002057909A (en) * | 2000-06-20 | 2002-02-22 | Ricoh Co Ltd | Method and system for processing color medium tone, and recording medium |
| WO2004023715A1 (en) * | 2002-09-03 | 2004-03-18 | The Regents Of The University Of California | Block cipher mode of operation for constructing a wide-blocksize block cipher from a conventional block cipher |
| CN101185122A (en) * | 2005-06-03 | 2008-05-21 | 皇家飞利浦电子股份有限公司 | Homomorphic encryption for secure watermarking |
| US20130063550A1 (en) * | 2006-02-15 | 2013-03-14 | Kenneth Ira Ritchey | Human environment life logging assistant virtual esemplastic network system and method |
| EP1833237A2 (en) * | 2006-03-07 | 2007-09-12 | Kabushiki Kaisha Toshiba | Retrieval of information embedded in an image |
| WO2010064333A1 (en) * | 2008-12-05 | 2010-06-10 | Kabushiki Kaisha Toshiba | Image processing method and image processing apparatus |
| US20190113973A1 (en) * | 2012-09-14 | 2019-04-18 | Interaxon Inc | Systems and methods for collecting, analyzing, and sharing bio-signal and non-bio-signal data |
| CN103442156A (en) * | 2013-08-20 | 2013-12-11 | 宁波大学 | Image steganography and secret information extraction method based on intelligent mobile terminal |
| WO2015054666A1 (en) * | 2013-10-10 | 2015-04-16 | Board Of Regents, The University Of Texas System | Systems and methods for quantitative analysis of histopathology images using multi-classifier ensemble schemes |
| WO2015052351A1 (en) * | 2013-10-11 | 2015-04-16 | Mauna Kea Technologies | Method for characterizing images acquired through a video medical device |
| EP3242239A1 (en) * | 2016-05-02 | 2017-11-08 | Scantrust SA | Optimized authentication of graphic authentication code |
| CN106600517A (en) * | 2016-11-11 | 2017-04-26 | 陕西师范大学 | Multi-carrier secret image separate storage and reconstruction method based on EMD-3 |
| CN106845339A (en) * | 2016-12-13 | 2017-06-13 | 电子科技大学 | A kind of mobile phone individual discrimination method based on bispectrum and EMD fusion features |
| US20190372754A1 (en) * | 2017-02-15 | 2019-12-05 | Lg Electronics Inc. | Apparatus and method for generating ciphertext data with maintained structure for analytics capability |
| US20190028608A1 (en) * | 2017-07-24 | 2019-01-24 | Samsung Electronics Co., Ltd. | Electronic device and method for controlling the electronic device |
| WO2019050994A1 (en) * | 2017-09-05 | 2019-03-14 | Gritstone Oncology, Inc. | Neoantigen identification for t-cell therapy |
| CN108959478A (en) * | 2018-06-21 | 2018-12-07 | 中南林业科技大学 | Ciphertext image search method and system under a kind of cloud environment |
| WO2020103411A1 (en) * | 2018-11-22 | 2020-05-28 | 九牧厨卫股份有限公司 | Fall detection method, device, and system |
| CN109784023A (en) * | 2018-11-28 | 2019-05-21 | 西安电子科技大学 | Stable state vision inducting brain electricity personal identification method and system based on deep learning |
| CN109858264A (en) * | 2019-01-22 | 2019-06-07 | 四川大学 | A kind of color image encipher-decipher method of the moulds resolutions of vectors such as score stochastic transformation domain |
| CN110071798A (en) * | 2019-03-21 | 2019-07-30 | 深圳大学 | A kind of equivalent key acquisition methods, device and computer readable storage medium |
| CN110269625A (en) * | 2019-05-31 | 2019-09-24 | 杭州电子科技大学 | A kind of electrocardio authentication method and system of novel multiple features fusion |
| CN110334881A (en) * | 2019-07-17 | 2019-10-15 | 深圳大学 | A kind of Financial Time Series Forecasting method based on length memory network and depth data cleaning, device and server |
| CN110706144A (en) * | 2019-09-17 | 2020-01-17 | 四川大学 | Optical color image encryption method based on spiral phase transformation and equal-mode decomposition |
Non-Patent Citations (1)
| Title |
|---|
| 胡克亚: "基于分块压缩感知和改进幻方变换的图像加密", 《激光技术》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112241931A (en) * | 2020-10-22 | 2021-01-19 | 海南大学 | Frequency domain digital veil design method for image easiness |
| CN112351151A (en) * | 2020-11-10 | 2021-02-09 | 广东工业大学 | FCN-based image information encryption method and device |
| CN112351151B (en) * | 2020-11-10 | 2022-09-16 | 广东工业大学 | FCN-based image information encryption method and device |
| CN116310719B (en) * | 2023-02-10 | 2024-04-19 | 中国人民解放军军事科学院国防科技创新研究院 | Time-frequency domain-based optical diffraction complex model training method and image processing method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111709867B (en) | 2022-11-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Wang et al. | A novel triple-image encryption and hiding algorithm based on chaos, compressive sensing and 3D DCT | |
| CN110136063B (en) | Single image super-resolution reconstruction method based on condition generation countermeasure network | |
| CN111709867B (en) | Novel full convolution network-based equal-modulus vector decomposition image encryption analysis method | |
| CN112217825B (en) | Asymmetric encryption system attack method based on resolution self-adaptive network | |
| Wei et al. | Generative steganography network | |
| Zhang et al. | Generative reversible data hiding by image-to-image translation via GANs | |
| Sun et al. | Multiple-image encryption based on cascaded gyrator transforms and high-dimensional chaotic system | |
| Bao et al. | Image encryption in the wavelet domain | |
| CN112016112A (en) | Method for encrypting image by compounding Fourier transform and differential transform | |
| CN113285797B (en) | Multi-image encryption method for optical rotation domain based on compressed sensing and deep learning | |
| CN104143174A (en) | Image disguising and reconstructing method based on Tangram algorithm and two-dimensional double-scale rectangular mapping | |
| Gopan et al. | Video super resolution with generative adversarial network | |
| Singh et al. | GIMD-Net: An effective General-purpose Image Manipulation Detection Network, even under anti-forensic attacks | |
| CN116743934A (en) | Equal resolution image hiding and encrypting method based on deep learning and ghost imaging | |
| CN116309164A (en) | Image processing method and device | |
| Wang et al. | A customized deep network based encryption-then-lossy-compression scheme of color images achieving arbitrary compression ratios | |
| Wang et al. | Applying deep learning to known-plaintext attack on chaotic image encryption schemes | |
| CN112188030B (en) | Gray level image encryption coding and decoding method based on pixel combination and complex number transformation | |
| Zhou et al. | Deep learning based attack on phase-truncated optical encoding | |
| Huang et al. | Development of a novel hyperchaos-based image encryption algorithm consisting of two scrambling-diffusion operations | |
| Bai et al. | Reconstruction of chaotic grayscale image encryption based on deep learning | |
| Ping et al. | Image hiding using invertible neural network and similarity of bits pairs | |
| Kaushik et al. | A two stage hybrid model for image encryption and compression to enhance security and efficiency | |
| CN113313775B (en) | Nonlinear optical encryption system attack method based on deep learning | |
| Narula et al. | Implementation of hybrid technique from spatial and frequency domain steganography: Along with cryptography to withstand statistical attacks |
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 |