[go: up one dir, main page]

CN112288757A - An optimization method for image segmentation in encrypted domain based on data packaging technology - Google Patents

An optimization method for image segmentation in encrypted domain based on data packaging technology Download PDF

Info

Publication number
CN112288757A
CN112288757A CN202011197779.2A CN202011197779A CN112288757A CN 112288757 A CN112288757 A CN 112288757A CN 202011197779 A CN202011197779 A CN 202011197779A CN 112288757 A CN112288757 A CN 112288757A
Authority
CN
China
Prior art keywords
image
image segmentation
execution server
service provider
encrypted
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
Application number
CN202011197779.2A
Other languages
Chinese (zh)
Other versions
CN112288757B (en
Inventor
陈德霖
张冬傲
郑培嘉
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Sun Yat Sen University
Original Assignee
Sun Yat Sen University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Sun Yat Sen University filed Critical Sun Yat Sen University
Priority to CN202011197779.2A priority Critical patent/CN112288757B/en
Publication of CN112288757A publication Critical patent/CN112288757A/en
Application granted granted Critical
Publication of CN112288757B publication Critical patent/CN112288757B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/12—Edge-based segmentation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F21/00—Security arrangements for protecting computers, components thereof, programs or data against unauthorised activity
    • G06F21/60—Protecting data
    • G06F21/602—Providing cryptographic facilities or services
    • 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
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/136—Segmentation; Edge detection involving thresholding

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioethics (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Computer Hardware Design (AREA)
  • Computer Security & Cryptography (AREA)
  • Software Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Storage Device Security (AREA)
  • Computer And Data Communications (AREA)

Abstract

本发明提出一种基于数据打包技术的加密域图像分割优化方法，涉及加密域图像分割优化的技术领域，解决了当前图像分割方法在分割过程中存在空间资源占用大，计算复杂度高的问题，首先隐私服务提供商生成公钥和私钥发送至客户端，并将私钥发送给图像分割执行服务器，客户端对图像进行加密并进行数据打包，基于数据打包技术实现了计算复杂度及空间资源占用度的降低，提高图像分割的速度，图像分割执行服务器与隐私服务提供商之间通过多方安全计算进行交互，获取加密分割图像，图像分割执行服务器将加密分割图像发送给客户端解密，加密的图像仅能由图像拥有者进行解密，保障了安全隐私。

Figure 202011197779

The invention proposes an encryption domain image segmentation optimization method based on data packaging technology, relates to the technical field of encryption domain image segmentation optimization, and solves the problems of large space resource occupation and high computational complexity in the current image segmentation method in the segmentation process, First, the privacy service provider generates a public key and a private key and sends it to the client, and sends the private key to the image segmentation execution server. The client encrypts the image and packs the data. Based on the data packing technology, the computational complexity and space resources are realized. The occupancy is reduced and the speed of image segmentation is improved. The image segmentation execution server and the privacy service provider interact through multi-party secure computing to obtain encrypted segmented images. The image segmentation execution server sends the encrypted segmented images to the client for decryption. Images can only be decrypted by the image owner, ensuring security and privacy.

Figure 202011197779

Description

一种基于数据打包技术的加密域图像分割优化方法An optimization method for image segmentation in encrypted domain based on data packaging technology

技术领域technical field

本发明涉及加密域图像分割优化的技术领域，更具体地，涉及一种基于数据打包技术的加密域图像分割优化方法。The present invention relates to the technical field of encryption domain image segmentation optimization, and more particularly, to an encryption domain image segmentation optimization method based on data packaging technology.

背景技术Background technique

随着社会经济的飞速发展，图像分割技术在我们生活中的应用场景越来越多，包括公安、交通、医院等领域，提高了公共安全、信息管理的效率。With the rapid development of society and economy, image segmentation technology has more and more application scenarios in our lives, including public security, transportation, hospitals and other fields, improving the efficiency of public security and information management.

现有的加密信号处理技术(SPED)为云服务器提供了在保证隐私安全的同时完成大量对图像数据的处理能力。例如，云服务器可以实现加密图像水印、去噪和特征提取等工作。边缘检测和图像分割是图像处理的重要课题之一，然而云计算的场景下进行图像分割可能会引起隐私泄露问题。从数据收集环节来看，用户将需要进行图像分割的高清卫星图像、医疗影像和交通路况等隐私资源存放在端上。在云环境下进行图像分割的过程中，客户端将隐私资源传送至云服务器上进行边缘识别，而在这个过程中，由于目前传输的数据均为加密数据，因此用户的隐私数据往往由云服务器或主动或被动地泄露，而这种隐私数据泄露所带来的潜在的安全风险十分严重，可能会被人用于获利，更甚者被犯罪分子捕获，并用于非法活动。The existing encrypted signal processing technology (SPED) provides the cloud server with the ability to process a large amount of image data while ensuring privacy and security. For example, cloud servers can implement tasks such as encrypted image watermarking, denoising, and feature extraction. Edge detection and image segmentation are one of the important topics in image processing. However, image segmentation in cloud computing scenarios may cause privacy leakage. From the perspective of data collection, users store private resources such as high-definition satellite images, medical images, and traffic conditions that require image segmentation on the terminal. In the process of image segmentation in the cloud environment, the client transmits private resources to the cloud server for edge recognition. In this process, since the currently transmitted data is encrypted data, the user's private data is often stored by the cloud server. Either actively or passively, the potential security risks brought by such private data leakage are very serious, and may be used for profit, or even captured by criminals and used for illegal activities.

当前，Paillier密码系统和DGK密码系统是最流行的加性同态加密系统，基于同态加密的图像分割技术也在被广泛引用，如2018年3月，张敏情，李天雪，狄富强等人在郑州大学学报上发表了“基于Paillier同态公钥加密系统的可逆信息隐藏算法”的文章(50(001)：8-14)，图像拥有者通过公钥加密分割后的图像平面，然后进行组合，中间操作者在密文域下再分割图像，基于同态乘法将分割后的图像嵌入后组合，传输至图像接收者解密，在此方法下，中间操作者不需要使用私钥即可通过同态乘法嵌入数据后组合，而图像接收者通过安全通道接收到的私钥解密图像，非法敌手无法获取私钥，使图像在传输分割过程中的隐私安全性得到了保障，但一方面同态加密本身占用的计算资源大，加密域下矩阵运算耗时过长，容易导致尺寸灾难，无法满足实时性要求；另一方面，此方法中虽然借助了安全通道传递私钥，但加密的图像在解密时非同一人，也会存在隐私泄露的风险。Currently, Paillier cryptosystem and DGK cryptosystem are the most popular additive homomorphic encryption systems, and image segmentation technology based on homomorphic encryption is also widely cited, such as in March 2018, Zhang Minqing, Li Tianxue, Di Fuqiang and others in Zhengzhou The article "Reversible Information Hiding Algorithm Based on Paillier Homomorphic Public Key Encryption System" was published in the University Journal (50(001): 8-14), the image owner encrypts the segmented image planes through the public key, and then combines them. The intermediate operator divides the image in the ciphertext domain, embeds the segmented images based on homomorphic multiplication and combines them, and transmits it to the image recipient for decryption. Under this method, the intermediate operator does not need to use the private key to pass the homomorphic After multiplying the embedded data and combining, the image receiver decrypts the image through the private key received through the secure channel, and the illegal adversary cannot obtain the private key, which ensures the privacy and security of the image during the transmission and segmentation process. However, on the one hand, homomorphic encryption itself It takes up a lot of computing resources, and the matrix operation in the encryption domain takes too long, which can easily lead to size disaster and cannot meet the real-time requirements. Not the same person, there is also the risk of privacy leakage.

发明内容SUMMARY OF THE INVENTION

为解决现有基于同态加密的图像分割方法在分割过程中存在空间资源占用大，计算复杂度高的问题，本发明提出一种基于数据打包技术的加密域图像分割方法，在保证隐私安全性的同时，降低图像分割过程中的计算复杂度和空间资源占用度，从而优化分割速度，满足实时性要求。In order to solve the problems of large space resource occupation and high computational complexity in the existing image segmentation method based on homomorphic encryption in the segmentation process, the present invention proposes an encryption domain image segmentation method based on data packaging technology, which can ensure privacy and security. At the same time, it reduces the computational complexity and space resource occupancy in the image segmentation process, thereby optimizing the segmentation speed and meeting real-time requirements.

为了达到上述技术效果，本发明的技术方案如下：In order to achieve above-mentioned technical effect, technical scheme of the present invention is as follows:

一种基于数据打包技术的加密域图像分割优化方法，至少包括：An encryption-domain image segmentation optimization method based on data packaging technology, comprising at least:

S1.隐私服务提供商生成公钥pk和私钥sk，将公钥pk和私钥sk均发送至客户端，并将私钥sk发送给图像分割执行服务器；S1. The privacy service provider generates the public key pk and the private key sk, sends both the public key pk and the private key sk to the client, and sends the private key sk to the image segmentation execution server;

S2.客户端利用公钥pk对图像进行加密并进行数据打包，压缩加密图像尺寸，并将加密打包后的图像发送至图像分割执行服务器；S2. The client uses the public key pk to encrypt the image and package the data, compress the size of the encrypted image, and send the encrypted and packaged image to the image segmentation execution server;

S3.图像分割执行服务器与隐私服务提供商之间通过多方安全计算及乱码电路技术进行交互，获取加密分割图像；S3. The image segmentation execution server and the privacy service provider interact through multi-party secure computing and garbled circuit technology to obtain encrypted segmentation images;

S4.图像分割执行服务器将加密分割图像发送给客户端，客户端解密后得到最终边缘图像。S4. The image segmentation execution server sends the encrypted segmented image to the client, and the client decrypts to obtain the final edge image.

在本技术方案中，客户端利用公钥pk对图像进行加密并进行数据打包，压缩加密图像尺寸，利用数据打包技术实现计算复杂度及空间资源占用度的降低，提高图像分割的速度，而且通过同态加密，加密的图像仅能由图像拥有者进行解密，保证了图像分割执行服务器和客户端双方对隐私性的要求。In this technical solution, the client uses the public key pk to encrypt the image and perform data packaging, compresses and encrypts the size of the encrypted image, uses the data packaging technology to reduce the computational complexity and space resource occupancy, improves the speed of image segmentation, and Homomorphic encryption, the encrypted image can only be decrypted by the image owner, which ensures the privacy requirements of both the image segmentation server and the client.

优选地，步骤S2所述的数据打包，压缩加密图像尺寸的过程为：Preferably, in the data packaging described in step S2, the process of compressing and encrypting the image size is as follows:

客户端利用公钥pk对图像加密后形成加密图像

Figure BDA0002754491850000021
尺寸为sm×sn，将加密图像
Figure BDA0002754491850000022
分割成相同尺寸大小的L块图片，表示为：I0，I1，…，IL-1，执行公式：The client uses the public key pk to encrypt the image to form an encrypted image
Figure BDA0002754491850000021
of size s m ×s n , will encrypt the image
Figure BDA0002754491850000022
Divide the picture into L blocks of the same size, expressed as: I 0 , I 1 , ..., I L-1 , and execute the formula:

Figure BDA0002754491850000023
Figure BDA0002754491850000023

其中，

Figure BDA0002754491850000024
利用公钥pk对IP打包成尺寸为
Figure BDA0002754491850000025
的压缩图像
Figure BDA0002754491850000026
Figure BDA0002754491850000027
表示压缩图像
Figure BDA0002754491850000028
的尺寸参数，t，Q均表示打包参与参数，
Figure BDA0002754491850000029
Ik(i，j)表示第k块图片中索引为(i，j)的图像的像素值，通过数据打包技术，压缩图像的尺寸，降低空间资源占用度，提高图像分割速度。in,
Figure BDA0002754491850000024
Use the public key pk to pack the IP into a size of
Figure BDA0002754491850000025
compressed image of
Figure BDA0002754491850000026
Figure BDA0002754491850000027
Represents a compressed image
Figure BDA0002754491850000028
The size parameters of , t, Q all represent the packaging participation parameters,
Figure BDA0002754491850000029
I k (i, j) represents the pixel value of the image with index (i, j) in the k-th block picture. Through data packing technology, the size of the image is compressed, the space resource occupancy is reduced, and the image segmentation speed is improved.

优选地，步骤S2之后，步骤S3之前还包括：图像分割执行服务器将压缩图像

Figure BDA0002754491850000031
进行高斯滤波，高斯滤波后的结果表示为
Figure BDA0002754491850000032
Figure BDA0002754491850000033
满足：Preferably, after step S2 and before step S3, the method further includes: the image segmentation execution server will compress the image
Figure BDA0002754491850000031
Gaussian filtering is performed, and the result after Gaussian filtering is expressed as
Figure BDA0002754491850000032
Figure BDA0002754491850000033
Satisfy:

Figure BDA0002754491850000034
保留乘法运算结果，表示为：
Figure BDA0002754491850000034
The result of the multiplication operation is preserved, expressed as:

Figure BDA0002754491850000035
Figure BDA0002754491850000035

其中，m＝0，1，…，sm；n＝0，1，…，sn，均表示图像尺寸参数，G为高斯滤波模板，Q为高斯滤波模板量化参数，sg为高斯滤波模板尺寸。 Among them, m = 0 , 1, . size.

在此，为了避免不必要的计算，减少除法运算带来的计算开销，降低计算复杂度，提高分割速度，在高斯滤波时没有进行除法运算，而是直接保留乘法运算结果。Here, in order to avoid unnecessary calculations, reduce the computational overhead caused by the division operation, reduce the computational complexity, and improve the segmentation speed, the Gaussian filtering does not perform the division operation, but directly retains the multiplication result.

优选地，步骤S3所述图像分割执行服务器与隐私服务提供商之间通过多方安全计算及乱码电路技术进行交互的过程包括：对高斯滤波后的加密打包数据

Figure BDA0002754491850000036
进行比较，步骤为：Preferably, the process of interacting between the image segmentation execution server and the privacy service provider in step S3 through multi-party secure computing and garbled circuit technology includes: encrypting the encrypted package data after Gaussian filtering.
Figure BDA0002754491850000036
For comparison, the steps are:

S31.以水平方向的Sobel滤波模板Gx为例，图像分割执行服务器构建两个临时模板：模板系数为正的部分表示为

Figure BDA0002754491850000037
和模板系数为负的部分表示为
Figure BDA0002754491850000038
构建公式为：S31. Taking the Sobel filter template G x in the horizontal direction as an example, the image segmentation execution server constructs two temporary templates: the part with positive template coefficients is expressed as
Figure BDA0002754491850000037
and the part where the template coefficient is negative is expressed as
Figure BDA0002754491850000038
The build formula is:

Figure BDA0002754491850000039
Figure BDA0002754491850000039

Figure BDA00027544918500000310
Figure BDA00027544918500000310

其中，

Figure BDA00027544918500000311
和
Figure BDA00027544918500000312
分别是水平方向Sobel卷积核中系数为正和负的部分，in,
Figure BDA00027544918500000311
and
Figure BDA00027544918500000312
are the positive and negative coefficients of the Sobel convolution kernel in the horizontal direction, respectively.

Figure BDA00027544918500000313
Figure BDA00027544918500000313

S32.图像分割执行服务器生成两个随机数α+、α-，并执行公式对两个随机数打包：S32. The image segmentation execution server generates two random numbers α + , α - , and executes the formula to pack the two random numbers:

Figure BDA00027544918500000314
Figure BDA00027544918500000314

Figure BDA00027544918500000315
Figure BDA00027544918500000315

其中，

Figure BDA0002754491850000041
表示随机数α+进行数据打包后的值；
Figure BDA0002754491850000042
表示随机数α-进行数据打包后的值，基于打包后的随机数，图像分割执行服务器生成中间密文
Figure BDA0002754491850000043
及垂直
Figure BDA0002754491850000044
分别满足：in,
Figure BDA0002754491850000041
Represents the random number α + the value after data packing;
Figure BDA0002754491850000042
Indicates the random number α - the value after data packing, and based on the packed random number, the image segmentation execution server generates the intermediate ciphertext
Figure BDA0002754491850000043
and vertical
Figure BDA0002754491850000044
respectively satisfy:

Figure BDA0002754491850000045
Figure BDA0002754491850000045

Figure BDA0002754491850000046
Figure BDA0002754491850000046

图像分割执行服务器将中间密文

Figure BDA0002754491850000047
及
Figure BDA0002754491850000048
发送至隐私服务提供商；The image segmentation execution server converts the intermediate ciphertext
Figure BDA0002754491850000047
and
Figure BDA0002754491850000048
to a privacy service provider;

S33.隐私服务提供商利用私钥sk对中间密文

Figure BDA0002754491850000049
及
Figure BDA00027544918500000410
进行解密，分段得到
Figure BDA00027544918500000411
和
Figure BDA00027544918500000412
并利用随机数α1生成集合{δ}，集合{δ}中任意一个值δi满足：S33. The privacy service provider uses the private key sk to pair the intermediate ciphertext
Figure BDA0002754491850000049
and
Figure BDA00027544918500000410
Decrypt, segment to get
Figure BDA00027544918500000411
and
Figure BDA00027544918500000412
And use the random number α 1 to generate the set {δ}, any value δ i in the set {δ} satisfies:

Figure BDA00027544918500000413
Figure BDA00027544918500000413

其中，i＝0，1，…，L-1，c为额外变量，c∈{0，1}，为隐私服务提供商随机挑选；Among them, i=0, 1, ..., L-1, c is an additional variable, c∈{0, 1}, which is randomly selected by the privacy service provider;

S34.根据随机数α+、α-，图像分割执行服务器计算中间参数σ0及σ1，满足：σ0＝α+-α-，σ1＝α--α+，然后通过随机数α2修饰σi，公式为：S34. According to the random numbers α + and α - , the image segmentation execution server calculates the intermediate parameters σ 0 and σ 1 , satisfying: σ 0 =α + -α - , σ 1 =α - -α + , and then pass the random number α 2 Modified σ i , the formula is:

Figure BDA00027544918500000414
Figure BDA00027544918500000414

其中，

Figure BDA00027544918500000415
表示修饰后的σi，i＝0，1；图像分割执行服务器与隐私服务提供商之间执行茫然传输后，图像分割执行服务器持有
Figure BDA00027544918500000416
隐私服务提供商持有δi；in,
Figure BDA00027544918500000415
Represents the modified σ i , i=0, 1; after the ambiguity transmission is performed between the image segmentation execution server and the privacy service provider, the image segmentation execution server holds the
Figure BDA00027544918500000416
The privacy service provider holds δ i ;

S35.由图像分割执行服务器创建乱码电路，生成一个对应带额外输入的比较电路C的加密乱码电路表GCT，另一方对乱码电路表GCT解码，以持有值

Figure BDA00027544918500000417
及δi计算最终电路的输出
Figure BDA00027544918500000418
基于比较乱码电路完成加密打包数据
Figure BDA00027544918500000419
的比较，图像分割执行服务器获取比较结果r∈{0，1}。S35. Create a garbled circuit by the image segmentation execution server, generate an encrypted garbled circuit table GCT corresponding to the comparison circuit C with extra input, and the other party decodes the garbled circuit table GCT to hold the value
Figure BDA00027544918500000417
and δ i to calculate the output of the final circuit
Figure BDA00027544918500000418
Encrypted and packaged data based on comparison garbled circuit
Figure BDA00027544918500000419
For comparison, the image segmentation execution server obtains the comparison result r∈{0,1}.

在此，由于仅使用同态加密无法实现加密值的比较，引入乱码电路技术实现图像分割执行服务器、隐私服务提供商的交互，实现计算加性同态加密所不能计算的线性函数。Here, since the comparison of encrypted values cannot be achieved only by using homomorphic encryption, the garbled circuit technology is introduced to realize the interaction between the image segmentation execution server and the privacy service provider, and realize the linear function that cannot be calculated by the additive homomorphic encryption.

优选地，图像分割执行服务器与隐私服务提供商之间执行茫然传输的过程中，图像分割执行服务器持有

Figure BDA00027544918500000420
和
Figure BDA00027544918500000421
隐私服务提供商持有c∈{0，1}，图像分割执行服务器无法获得c的值，隐私服务提供商也无法获得
Figure BDA00027544918500000422
隐私服务提供商生成
Figure BDA00027544918500000423
并将
Figure BDA00027544918500000424
发送给图像分割执行服务器，图像分割执行服务器通过减去α2获取
Figure BDA00027544918500000425
保证了隐私安全性。Preferably, in the process of performing dazed transmission between the image segmentation execution server and the privacy service provider, the image segmentation execution server holds the
Figure BDA00027544918500000420
and
Figure BDA00027544918500000421
The privacy service provider holds c ∈ {0, 1}, the image segmentation execution server cannot obtain the value of c, and the privacy service provider cannot obtain
Figure BDA00027544918500000422
Privacy Service Provider Generated
Figure BDA00027544918500000423
and will
Figure BDA00027544918500000424
Send to the image segmentation execution server, and the image segmentation execution server obtains by subtracting α 2
Figure BDA00027544918500000425
Privacy security is guaranteed.

优选地，步骤S3所述图像分割执行服务器与隐私服务提供商之间通过多方安全计算及乱码电路技术进行交互的过程还包括：对高斯滤波后的加密打包数据

Figure BDA0002754491850000051
进行Sobel滤波边缘检测，得到加密图像
Figure BDA0002754491850000052
步骤为：Preferably, the process of interacting between the image segmentation execution server and the privacy service provider in step S3 through multi-party secure computing and garbled circuit technology further includes: encrypting the Gaussian filtered encrypted packet data
Figure BDA0002754491850000051
Perform Sobel filter edge detection to get encrypted image
Figure BDA0002754491850000052
The steps are:

隐私服务提供商对集合{δ}加密得到加密集合

Figure BDA0002754491850000053
并发送至图像分割执行服务器；The privacy service provider encrypts the set {δ} to get the encrypted set
Figure BDA0002754491850000053
And send it to the image segmentation execution server;

对于加密集合

Figure BDA0002754491850000054
中的每一个
Figure BDA0002754491850000055
持有比较结果r的图像分割执行服务器执行公式，获取到
Figure BDA0002754491850000056
公式为：For encrypted collections
Figure BDA0002754491850000054
each of the
Figure BDA0002754491850000055
The image segmentation execution server holding the comparison result r executes the formula and obtains
Figure BDA0002754491850000056
The formula is:

Figure BDA0002754491850000057
Figure BDA0002754491850000057

其中，α2为步骤S3中图像分割执行服务器生成的随机数，根据

Figure BDA0002754491850000058
图像分割执行服务器可获取水平分量
Figure BDA0002754491850000059
和垂直分量
Figure BDA00027544918500000510
随后根据公式Among them, α 2 is the random number generated by the image segmentation execution server in step S3, according to
Figure BDA0002754491850000058
Image segmentation execution server can obtain horizontal components
Figure BDA0002754491850000059
and the vertical component
Figure BDA00027544918500000510
Then according to the formula

Figure BDA00027544918500000511
Figure BDA00027544918500000511

构建Sobel滤波结果为加密图像

Figure BDA00027544918500000512
Construct Sobel filter result as encrypted image
Figure BDA00027544918500000512

优选地，步骤S3所述图像分割执行服务器与隐私服务提供商之间通过多方安全计算及乱码电路技术进行交互的过程还包括：获取加密图像

Figure BDA00027544918500000513
的加密阈值，步骤为：Preferably, the process of interacting between the image segmentation execution server and the privacy service provider in step S3 through multi-party secure computing and garbled circuit technology further includes: acquiring an encrypted image
Figure BDA00027544918500000513
encryption threshold, the steps are:

S301.设加密图像

Figure BDA00027544918500000514
的当前阈值为
Figure BDA00027544918500000515
图像分割执行服务首先设置加密图像
Figure BDA00027544918500000516
中最大像素值
Figure BDA00027544918500000517
和最小像素值
Figure BDA00027544918500000518
然后利用除法协议生成初始阈值
Figure BDA00027544918500000519
公式为：S301. Set encrypted image
Figure BDA00027544918500000514
The current threshold is
Figure BDA00027544918500000515
The image segmentation execution service first sets the encrypted image
Figure BDA00027544918500000516
medium maximum pixel value
Figure BDA00027544918500000517
and the minimum pixel value
Figure BDA00027544918500000518
An initial threshold is then generated using the division protocol
Figure BDA00027544918500000519
The formula is:

Figure BDA00027544918500000520
Figure BDA00027544918500000520

其中，SecDiv表示除法协议；Among them, SecDiv represents the division protocol;

S302.图像分割执行服务器根据当前阈值

Figure BDA00027544918500000521
将加密图像
Figure BDA00027544918500000522
分为前景部分和背景部分，其中，前景部分指像素的值大于Ti，背景部分指像素的值小于等于Ti；S302. The image segmentation executes the server according to the current threshold
Figure BDA00027544918500000521
will encrypt the image
Figure BDA00027544918500000522
It is divided into a foreground part and a background part, wherein the foreground part means that the value of the pixel is greater than T i , and the background part means that the value of the pixel is less than or equal to T i ;

S303.利用乱码电路对当前阈值

Figure BDA00027544918500000523
和加密像素值
Figure BDA00027544918500000524
进行比较，得到比较结果{λ}；S303. Use the garbled circuit to set the current threshold
Figure BDA00027544918500000523
and encrypted pixel values
Figure BDA00027544918500000524
Compare and get the comparison result {λ};

S304.设背景累计值为

Figure BDA00027544918500000525
前景累计值为
Figure BDA00027544918500000526
背景计数值为
Figure BDA00027544918500000527
前景技术值为
Figure BDA00027544918500000528
图像分割执行服务器初始化背景累计值
Figure BDA00027544918500000529
前景累计值
Figure BDA00027544918500000530
及背景计数值
Figure BDA00027544918500000531
前景计数值
Figure BDA00027544918500000532
并进行像素计算，公式为：S304. Set the background cumulative value to be
Figure BDA00027544918500000525
The prospect cumulative value is
Figure BDA00027544918500000526
The background count is
Figure BDA00027544918500000527
The future technical value is
Figure BDA00027544918500000528
Image segmentation execution server initializes background cumulative value
Figure BDA00027544918500000529
Prospect cumulative value
Figure BDA00027544918500000530
and background counts
Figure BDA00027544918500000531
Foreground count value
Figure BDA00027544918500000532
And perform pixel calculation, the formula is:

Figure BDA0002754491850000061
Figure BDA0002754491850000061

Figure BDA0002754491850000062
Figure BDA0002754491850000062

Figure BDA0002754491850000063
Figure BDA0002754491850000063

计算下一个阈值

Figure BDA0002754491850000064
公式为：Calculate the next threshold
Figure BDA0002754491850000064
The formula is:

Figure BDA0002754491850000065
Figure BDA0002754491850000065

S305.图像分割执行服务器通过与隐私服务提供商共同执行绝对值乱码电路比较，获取

Figure BDA0002754491850000066
图像分割执行服务器将
Figure BDA0002754491850000067
及Δ+r′给隐私服务提供商，r′为用于修饰的随机值，隐私服务提供商判断阈值计算是否达到阈值标准ε，若是，阈值迭代计算结束；否则，返回步骤S304。S305. The image segmentation execution server performs absolute value garbled circuit comparison with the privacy service provider to obtain
Figure BDA0002754491850000066
The image segmentation execution server will
Figure BDA0002754491850000067
and Δ+r' to the privacy service provider, where r' is a random value used for modification, the privacy service provider determines whether the threshold calculation reaches the threshold standard ε, if so, the threshold iteration calculation ends; otherwise, return to step S304.

优选地，步骤S303所述利用乱码电路对当前阈值

Figure BDA0002754491850000068
和加密像素值
Figure BDA0002754491850000069
进行比较的过程为：Preferably, in step S303, use the garbled circuit to control the current threshold
Figure BDA0002754491850000068
and encrypted pixel values
Figure BDA0002754491850000069
The comparison process is:

A.图像分割执行服务器生成随机数γ1和β∈{0，1}，基于随机数γ1和β∈{0，1}，计算

Figure BDA00027544918500000610
并将
Figure BDA00027544918500000611
发送至隐私服务提供商，公式为：A. The image segmentation execution server generates random numbers γ 1 and β ∈ {0, 1}, based on the random numbers γ 1 and β ∈ {0, 1}, calculates
Figure BDA00027544918500000610
and will
Figure BDA00027544918500000611
Sent to a privacy service provider with the formula:

Figure BDA00027544918500000612
Figure BDA00027544918500000612

B.隐私服务提供商将

Figure BDA00027544918500000613
得到t1，同时生成c∈{0，1}，持有t1和c的隐私服务提供商与持有随机数γ1的图像分割执行服务器执行带额外输入的比较乱码电路，隐私服务提供商获得修饰后的结果
Figure BDA00027544918500000614
B. The Privacy Service Provider will
Figure BDA00027544918500000613
get t 1 , and generate c ∈ {0, 1} at the same time, the privacy service provider holding t 1 and c and the image segmentation execution server holding the random number γ 1 perform a comparison garbled circuit with extra input, the privacy service provider Get modified results
Figure BDA00027544918500000614

C.根据

Figure BDA00027544918500000615
和c，隐私服务提供商计算加密对t2，公式为：C. According to
Figure BDA00027544918500000615
and c, the privacy service provider calculates the encrypted pair t 2 with the formula:

Figure BDA00027544918500000616
Figure BDA00027544918500000616

隐私服务提供商将t2发送至图像分割执行服务器；The privacy service provider sends t 2 to the image segmentation execution server;

D.图像分割执行服务器根据β确认最终比较结果λ：若β＝1，图像分割执行服务器将t2的顺序调换再将其作为结果λ，若β＝0，图像分割执行服务器将t2作为结果λ。D. The image segmentation execution server confirms the final comparison result λ according to β: if β=1, the image segmentation execution server reverses the order of t 2 and takes it as the result λ; if β=0, the image segmentation execution server takes t 2 as the result λ.

优选地，步骤S305中，图像分割执行服务器通过与隐私服务提供商共同执行绝对值乱码电路时是与0进行比较的。Preferably, in step S305, the image segmentation execution server compares with 0 when the absolute value garbled code circuit is executed jointly with the privacy service provider.

优选地，图像分割执行服务器在拥有加密图像

Figure BDA0002754491850000071
及加密图像
Figure BDA0002754491850000072
的加密阈值Ti后，图像分割执行服务器与隐私服务提供商利用比较乱码电路进行加密图像
Figure BDA0002754491850000073
的二值化，得到加密分割图像
Figure BDA0002754491850000074
Preferably, the image segmentation execution server is in possession of the encrypted image
Figure BDA0002754491850000071
and encrypted images
Figure BDA0002754491850000072
After the encryption threshold T i , the image segmentation execution server and the privacy service provider use the comparison garbled circuit to encrypt the image
Figure BDA0002754491850000073
The binarization of , obtains the encrypted segmentation image
Figure BDA0002754491850000074

与现有技术相比，本发明技术方案的有益效果是：Compared with the prior art, the beneficial effects of the technical solution of the present invention are:

本发明提出一种基于数据打包技术的加密域图像分割优化方法，首先隐私服务提供商生成公钥和私钥发送至客户端，并将私钥发送给图像分割执行服务器，然后客户端对图像进行加密并进行数据打包，压缩加密图像尺寸，基于数据打包技术实现了计算复杂度及空间资源占用度的降低，提高图像分割的速度，图像分割执行服务器与隐私服务提供商之间通过多方安全计算及乱码电路技术进行交互，获取加密分割图像，图像分割执行服务器将加密分割图像发送给客户端解密，加密的图像仅能由图像拥有者进行解密，保证了图像分割执行服务器和客户端双方对隐私性的要求，确保在隐私安全的情况下进行图像分割。The present invention proposes an encryption domain image segmentation optimization method based on data packaging technology. First, a privacy service provider generates a public key and a private key and sends them to the client, and sends the private key to the image segmentation execution server, and then the client performs the image segmentation on the image. Encrypt and package data, compress and encrypt image size, reduce computational complexity and space resource occupancy based on data packaging technology, and improve the speed of image segmentation. The garbled circuit technology interacts to obtain the encrypted segmented image. The image segmentation execution server sends the encrypted segmented image to the client for decryption. The encrypted image can only be decrypted by the image owner, which ensures the privacy of both the image segmentation execution server and the client. requirements to ensure that image segmentation is performed in a privacy-safe manner.

附图说明Description of drawings

图1表示本发明实施例中提出的基于数据打包技术的加密域图像分割优化方法的流程示意图；Fig. 1 represents the schematic flow chart of the encryption domain image segmentation optimization method based on the data packing technology proposed in the embodiment of the present invention;

图2表示本发明实施例中提出的基本比较乱码电路的结构图；Fig. 2 shows the structure diagram of the basic comparison garbled circuit proposed in the embodiment of the present invention;

图3表示本发明实施例中提出的带额外输入的比较乱码电路的结构图；Fig. 3 shows the structure diagram of the comparison garbled circuit with extra input proposed in the embodiment of the present invention;

图4表示本发明实施例中提出的利用基于数据打包技术的加密域图像分割优化方法进行加密图像分割的实验结果示意图。FIG. 4 is a schematic diagram showing an experimental result of performing encrypted image segmentation by using the encryption domain image segmentation optimization method based on the data packing technology proposed in the embodiment of the present invention.

具体实施方式Detailed ways

附图仅用于示例性说明，不能理解为对本专利的限制；The accompanying drawings are for illustrative purposes only, and should not be construed as limitations on this patent;

为了更好地说明本实施例，附图某些部位会有省略、放大或缩小，并不代表实际尺寸；In order to better illustrate this embodiment, some parts of the drawings are omitted, enlarged or reduced, which do not represent the actual size;

对于本领域技术人员来说，附图中某些公知内容说明可能省略是可以理解的。For those skilled in the art, it is understandable that descriptions of certain well-known contents in the accompanying drawings may be omitted.

下面结合附图和实施例对本发明的技术方案做进一步的说明。The technical solutions of the present invention will be further described below with reference to the accompanying drawings and embodiments.

实施例1Example 1

如图1所示的基于数据打包技术的加密域图像分割优化方法的流程图，参见图1，所述基于数据打包技术的加密域图像分割优化方法的步骤包括：As shown in Figure 1, the flow chart of the encryption domain image segmentation optimization method based on data packaging technology, referring to Figure 1, the steps of the encryption domain image segmentation optimization method based on data packaging technology include:

S1.隐私服务提供商生成公钥pk和私钥sk，将公钥pk和私钥sk均发送至客户端，并将私钥sk发送给图像分割执行服务器；S1. The privacy service provider generates the public key pk and the private key sk, sends both the public key pk and the private key sk to the client, and sends the private key sk to the image segmentation execution server;

S2.客户端利用公钥pk对图像进行加密并进行数据打包，压缩加密图像尺寸，并将加密打包后的图像发送至图像分割执行服务器；S2. The client uses the public key pk to encrypt the image and package the data, compress the size of the encrypted image, and send the encrypted and packaged image to the image segmentation execution server;

S3.图像分割执行服务器与隐私服务提供商之间通过多方安全计算及乱码电路技术进行交互，获取加密分割图像；S3. The image segmentation execution server and the privacy service provider interact through multi-party secure computing and garbled circuit technology to obtain encrypted segmentation images;

S4.图像分割执行服务器将加密分割图像发送给客户端，客户端解密后得到最终边缘图像。S4. Image segmentation execution The server sends the encrypted segmented image to the client, and the client decrypts to obtain the final edge image.

在本实施例中，步骤S2所述的数据打包，压缩加密图像尺寸的过程为：In the present embodiment, the data packaging described in step S2, the process of compressing the encrypted image size is:

客户端利用公钥pk对图像加密后形成加密图像

Figure BDA0002754491850000081
尺寸为sm×sn，将加密图像
Figure BDA0002754491850000082
分割成相同尺寸大小的L块图片，表示为：I0，I1，…，IL-1，执行公式：The client uses the public key pk to encrypt the image to form an encrypted image
Figure BDA0002754491850000081
of size s m ×s n , will encrypt the image
Figure BDA0002754491850000082
Divide the picture into L blocks of the same size, expressed as: I 0 , I 1 , ..., I L-1 , and execute the formula:

Figure BDA0002754491850000083
Figure BDA0002754491850000083

其中，

Figure BDA0002754491850000084
利用公钥pk对IP打包成尺寸为
Figure BDA0002754491850000085
的压缩图像
Figure BDA0002754491850000086
Figure BDA0002754491850000087
表示压缩图像
Figure BDA0002754491850000088
的尺寸参数，t，Q均表示打包参与参数，
Figure BDA0002754491850000089
sm、sn表示加密图像
Figure BDA00027544918500000810
的尺寸参数，Ik(i，j)表示第k块图片中索引为(i，j)的图像的像素值，在此，为了防止溢出，额外留出了2位，通过数据打包技术，压缩图像的尺寸，降低空间资源占用度，提高图像分割速度。in,
Figure BDA0002754491850000084
Use the public key pk to pack the IP into a size of
Figure BDA0002754491850000085
compressed image of
Figure BDA0002754491850000086
Figure BDA0002754491850000087
Represents a compressed image
Figure BDA0002754491850000088
The size parameters of , t, Q all represent the packaging participation parameters,
Figure BDA0002754491850000089
s m , s n represent encrypted images
Figure BDA00027544918500000810
The size parameter of , I k (i, j) represents the pixel value of the image with index (i, j) in the k-th block picture. Here, in order to prevent overflow, 2 additional bits are reserved. Through data packing technology, compression The size of the image can reduce the space resource occupancy and improve the image segmentation speed.

在本实施例中，步骤S2之后，步骤S3之前还包括：图像分割执行服务器将压缩图像

Figure BDA00027544918500000811
进行高斯滤波，高斯滤波后的结果表示为
Figure BDA00027544918500000812
满足：In this embodiment, after step S2 and before step S3, it further includes: the image segmentation execution server will compress the image
Figure BDA00027544918500000811
Gaussian filtering is performed, and the result after Gaussian filtering is expressed as
Figure BDA00027544918500000812
Satisfy:

Figure BDA00027544918500000813
保留乘法运算结果，表示为：
Figure BDA00027544918500000813
The result of the multiplication operation is preserved, expressed as:

Figure BDA00027544918500000814
Figure BDA00027544918500000814

其中，m＝0，1，…，sm；n＝0，1，…，sn，均表示图像尺寸参数，G为高斯滤波模板，Q为高斯滤波模板量化参数，sg为高斯滤波模板尺寸，即为了避免不必要的计算，减少除法运算带来的计算开销，降低计算复杂度，提高分割速度，在高斯滤波时没有进行除法运算，而是直接保留乘法运算结果。 Among them, m = 0 , 1, . Size, that is, in order to avoid unnecessary calculations, reduce the computational overhead caused by the division operation, reduce the computational complexity, and improve the segmentation speed, no division operation is performed during Gaussian filtering, but the multiplication result is directly retained.

在本实施例中，步骤S3所述图像分割执行服务器与隐私服务提供商之间通过多方安全计算及乱码电路技术进行交互的过程包括：对高斯滤波后的加密打包数据

Figure BDA00027544918500000815
进行比较，步骤为：In this embodiment, the process of interacting between the image segmentation execution server and the privacy service provider in step S3 through multi-party secure computing and garbled circuit technology includes: encrypting the encrypted package data after Gaussian filtering.
Figure BDA00027544918500000815
For comparison, the steps are:

S31.以水平方向的Sobel滤波模板Gx为例，图像分割执行服务器构建两个临时模板：模板系数为正的部分表示为

Figure BDA0002754491850000091
和模板系数为负的部分表示为
Figure BDA0002754491850000092
构建公式为：S31. Taking the Sobel filter template G x in the horizontal direction as an example, the image segmentation execution server constructs two temporary templates: the part with positive template coefficients is expressed as
Figure BDA0002754491850000091
and the part where the template coefficient is negative is expressed as
Figure BDA0002754491850000092
The build formula is:

Figure BDA0002754491850000093
Figure BDA0002754491850000093

Figure BDA0002754491850000094
Figure BDA0002754491850000094

其中，

Figure BDA0002754491850000095
和
Figure BDA0002754491850000096
分别是水平方向Sobel卷积核中系数为正和负的部分，in,
Figure BDA0002754491850000095
and
Figure BDA0002754491850000096
are the positive and negative coefficients of the Sobel convolution kernel in the horizontal direction, respectively.

Figure BDA0002754491850000097
Figure BDA0002754491850000097

S32.图像分割执行服务器生成两个随机数α+、α-，并执行公式对两个随机数打包：S32. The image segmentation execution server generates two random numbers α + , α - , and executes the formula to pack the two random numbers:

Figure BDA0002754491850000098
Figure BDA0002754491850000098

Figure BDA0002754491850000099
Figure BDA0002754491850000099

其中，

Figure BDA00027544918500000910
表示随机数α+进行数据打包后的值；
Figure BDA00027544918500000911
表示随机数α_进行数据打包后的值，基于打包后的随机数，图像分割执行服务器生成中间密文
Figure BDA00027544918500000912
及垂直
Figure BDA00027544918500000913
分别满足：in,
Figure BDA00027544918500000910
Represents the value of random number α+ after data packing;
Figure BDA00027544918500000911
Indicates the value of the random number α_ after data packaging. Based on the packaged random number, the image segmentation execution server generates an intermediate ciphertext
Figure BDA00027544918500000912
and vertical
Figure BDA00027544918500000913
respectively satisfy:

Figure BDA00027544918500000914
Figure BDA00027544918500000914

Figure BDA00027544918500000915
Figure BDA00027544918500000915

图像分割执行服务器将中间密文

Figure BDA00027544918500000916
及
Figure BDA00027544918500000917
发送至隐私服务提供商；The image segmentation execution server converts the intermediate ciphertext
Figure BDA00027544918500000916
and
Figure BDA00027544918500000917
to a privacy service provider;

S33.隐私服务提供商利用私钥sk对中间密文

Figure BDA00027544918500000918
及
Figure BDA00027544918500000919
进行解密，分段得到
Figure BDA00027544918500000920
和
Figure BDA00027544918500000921
并利用随机数α1生成集合{δ}，集合{δ}中任意一个值δi满足：S33. The privacy service provider uses the private key sk to pair the intermediate ciphertext
Figure BDA00027544918500000918
and
Figure BDA00027544918500000919
Decrypt, segment to get
Figure BDA00027544918500000920
and
Figure BDA00027544918500000921
And use the random number α 1 to generate the set {δ}, any value δ i in the set {δ} satisfies:

Figure BDA00027544918500000922
Figure BDA00027544918500000922

其中，i＝0，1，…，L-1，c为额外变量，c∈{0，1}，为隐私服务提供商随机挑选；Among them, i=0, 1, ..., L-1, c is an additional variable, c∈{0, 1}, which is randomly selected by the privacy service provider;

S34.根据随机数α+、α_，图像分割执行服务器计算中间参数σ0及σ1，满足：σ0＝α+-α-，σ1＝α--α+，然后通过随机数α2修饰σi，公式为：S34. According to the random numbers α + , α _ , the image segmentation execution server calculates the intermediate parameters σ 0 and σ 1 , satisfying: σ 0 =α + -α - , σ 1 =α - -α + , and then pass the random number α 2 Modified σ i , the formula is:

Figure BDA00027544918500000923
Figure BDA00027544918500000923

其中，

Figure BDA0002754491850000101
表示修饰后的σi，i＝0，1；图像分割执行服务器与隐私服务提供商之间执行茫然传输后，图像分割执行服务器持有
Figure BDA0002754491850000102
隐私服务提供商持有δi；in,
Figure BDA0002754491850000101
Represents the modified σ i , i=0, 1; after the ambiguity transmission is performed between the image segmentation execution server and the privacy service provider, the image segmentation execution server holds the
Figure BDA0002754491850000102
The privacy service provider holds δ i ;

S35.由图像分割执行服务器创建乱码电路，生成一个对应带额外输入的比较电路C的加密乱码电路表GCT，另一方对乱码电路表GCT解码，以持有值

Figure BDA0002754491850000103
及δi计算最终电路的输出
Figure BDA0002754491850000104
基于比较乱码电路完成加密打包数据
Figure BDA0002754491850000105
的比较，图像分割执行服务器获取比较结果r∈{0，1}。图像分割执行服务器与隐私服务提供商之间执行茫然传输的过程中，图像分割执行服务器持有
Figure BDA0002754491850000106
和
Figure BDA0002754491850000107
隐私服务提供商持有c∈{0，1}，图像分割执行服务器无法获得c的值，隐私服务提供商也无法获得
Figure BDA0002754491850000108
隐私服务提供商生成
Figure BDA0002754491850000109
并将
Figure BDA00027544918500001010
发送给图像分割执行服务器，图像分割执行服务器通过减去α2获取
Figure BDA00027544918500001011
保证了隐私安全性。S35. Create a garbled circuit by the image segmentation execution server, generate an encrypted garbled circuit table GCT corresponding to the comparison circuit C with extra input, and the other party decodes the garbled circuit table GCT to hold the value
Figure BDA0002754491850000103
and δ i to calculate the output of the final circuit
Figure BDA0002754491850000104
Encrypted and packaged data based on comparison garbled circuit
Figure BDA0002754491850000105
For comparison, the image segmentation execution server obtains the comparison result r∈{0,1}. In the process of performing dazed transmission between the image segmentation execution server and the privacy service provider, the image segmentation execution server holds the
Figure BDA0002754491850000106
and
Figure BDA0002754491850000107
The privacy service provider holds c ∈ {0, 1}, the image segmentation execution server cannot obtain the value of c, and the privacy service provider cannot obtain
Figure BDA0002754491850000108
Privacy Service Provider Generated
Figure BDA0002754491850000109
and will
Figure BDA00027544918500001010
Send to the image segmentation execution server, and the image segmentation execution server obtains by subtracting α 2
Figure BDA00027544918500001011
Privacy security is guaranteed.

在本实施例中，步骤S3所述图像分割执行服务器与隐私服务提供商之间通过多方安全计算及乱码电路技术进行交互的过程还包括：对高斯滤波后的加密打包数据

Figure BDA00027544918500001012
进行Sobel滤波边缘检测，得到加密图像
Figure BDA00027544918500001013
步骤为：In this embodiment, the process of interacting between the image segmentation execution server and the privacy service provider in step S3 through multi-party secure computing and garbled circuit technology further includes: encrypting the encrypted packaged data after Gaussian filtering.
Figure BDA00027544918500001012
Perform Sobel filter edge detection to get encrypted image
Figure BDA00027544918500001013
The steps are:

隐私服务提供商对集合{δ}加密得到加密集合

Figure BDA00027544918500001014
并发送至图像分割执行服务器；The privacy service provider encrypts the set {δ} to get the encrypted set
Figure BDA00027544918500001014
And send it to the image segmentation execution server;

对于加密集合

Figure BDA00027544918500001015
中的每一个
Figure BDA00027544918500001016
持有比较结果r的图像分割执行服务器执行公式，获取到
Figure BDA00027544918500001017
公式为：For encrypted collections
Figure BDA00027544918500001015
each of the
Figure BDA00027544918500001016
The image segmentation execution server holding the comparison result r executes the formula and obtains
Figure BDA00027544918500001017
The formula is:

Figure BDA00027544918500001018
Figure BDA00027544918500001018

其中，α2为步骤S3中图像分割执行服务器生成的随机数，根据

Figure BDA00027544918500001019
图像分割执行服务器可获取水平分量
Figure BDA00027544918500001020
和垂直分量
Figure BDA00027544918500001021
随后根据公式Among them, α 2 is the random number generated by the image segmentation execution server in step S3, according to
Figure BDA00027544918500001019
Image segmentation execution server can obtain horizontal components
Figure BDA00027544918500001020
and the vertical component
Figure BDA00027544918500001021
Then according to the formula

Figure BDA00027544918500001022
Figure BDA00027544918500001022

构建Sobel滤波结果为加密图像

Figure BDA00027544918500001023
Construct Sobel filter result as encrypted image
Figure BDA00027544918500001023

在本实施例中，步骤S3所述图像分割执行服务器与隐私服务提供商之间通过多方安全计算及乱码电路技术进行交互的过程还包括：获取加密图像

Figure BDA00027544918500001024
的加密阈值，步骤为：In this embodiment, the process of interacting between the image segmentation execution server and the privacy service provider in step S3 through multi-party secure computing and garbled circuit technology further includes: acquiring an encrypted image
Figure BDA00027544918500001024
encryption threshold, the steps are:

S301.设加密图像

Figure BDA00027544918500001025
的当前阈值为
Figure BDA00027544918500001026
图像分割执行服务首先设置加密图像
Figure BDA00027544918500001027
和最大像素值
Figure BDA00027544918500001028
和最小像素值
Figure BDA00027544918500001029
然后利用除法协议生成初始阈值
Figure BDA00027544918500001030
公式为：S301. Set encrypted image
Figure BDA00027544918500001025
The current threshold is
Figure BDA00027544918500001026
The image segmentation execution service first sets the encrypted image
Figure BDA00027544918500001027
and the maximum pixel value
Figure BDA00027544918500001028
and the minimum pixel value
Figure BDA00027544918500001029
An initial threshold is then generated using the division protocol
Figure BDA00027544918500001030
The formula is:

Figure BDA0002754491850000111
Figure BDA0002754491850000111

其中，SecDiv表示除法协议；Among them, SecDiv represents the division protocol;

S302.图像分割执行服务器根据当前阈值

Figure BDA0002754491850000112
将加密图像
Figure BDA0002754491850000113
分为前景部分和背景部分，其中，前景部分指像素的值大于Ti，背景部分指像素的值小于等于Ti；S302. The image segmentation executes the server according to the current threshold
Figure BDA0002754491850000112
will encrypt the image
Figure BDA0002754491850000113
It is divided into a foreground part and a background part, wherein the foreground part means that the value of the pixel is greater than T i , and the background part means that the value of the pixel is less than or equal to T i ;

S303.利用乱码电路对当前阈值

Figure BDA0002754491850000114
和加密像素值
Figure BDA0002754491850000115
进行比较，得到比较结果{λ}；所述利用乱码电路对当前阈值
Figure BDA0002754491850000116
和加密像素值
Figure BDA0002754491850000117
进行比较的过程为：S303. Use the garbled circuit to set the current threshold
Figure BDA0002754491850000114
and encrypted pixel values
Figure BDA0002754491850000115
Compare to obtain the comparison result {λ}; the use of the garbled circuit to compare the current threshold
Figure BDA0002754491850000116
and encrypted pixel values
Figure BDA0002754491850000117
The comparison process is:

A.图像分割执行服务器生成随机数γ1和β∈{0，1}，基于随机数γ1和β∈{0，1}，计算

Figure BDA0002754491850000118
并将
Figure BDA0002754491850000119
发送至隐私服务提供商，公式为：A. The image segmentation execution server generates random numbers γ 1 and β ∈ {0, 1}, based on the random numbers γ 1 and β ∈ {0, 1}, calculates
Figure BDA0002754491850000118
and will
Figure BDA0002754491850000119
Sent to a privacy service provider with the formula:

Figure BDA00027544918500001110
Figure BDA00027544918500001110

B.隐私服务提供商将

Figure BDA00027544918500001111
得到t1，同时生成c∈{0，1}，持有t1和c的隐私服务提供商与持有随机数γ1的图像分割执行服务器执行带额外输入的比较乱码电路，隐私服务提供商获得修饰后的结果
Figure BDA00027544918500001112
B. The Privacy Service Provider will
Figure BDA00027544918500001111
get t 1 , and generate c ∈ {0, 1} at the same time, the privacy service provider holding t 1 and c and the image segmentation execution server holding the random number γ 1 perform a comparison garbled circuit with extra input, the privacy service provider Get modified results
Figure BDA00027544918500001112

C.根据

Figure BDA00027544918500001113
和c，隐私服务提供商计算加密对t2，公式为：C. According to
Figure BDA00027544918500001113
and c, the privacy service provider calculates the encrypted pair t 2 with the formula:

Figure BDA00027544918500001114
Figure BDA00027544918500001114

隐私服务提供商将t2发送至图像分割执行服务器；The privacy service provider sends t 2 to the image segmentation execution server;

D.图像分割执行服务器根据β确认最终比较结果λ：若β＝1，图像分割执行服务器将t2的顺序调换再将其作为结果λ，若β＝0，图像分割执行服务器将t2作为结果λ；D. The image segmentation execution server confirms the final comparison result λ according to β: if β=1, the image segmentation execution server reverses the order of t 2 and takes it as the result λ, if β=0, the image segmentation execution server takes t 2 as the result λ;

针对上述带额外输入的比较乱码电路，作进一步的说明，设两个l位长的整数x(l)和y(l)进行比较的乱码电路如图2所示，参见图2，客户端持有输入x(l)(x1-xl)，图像分割执行服务器持有输入y(l)(y1-yl)，为了隐藏图2中CMP的输入，图像分割执行服务器在电路中引入一个额外变量c，满足：For the above comparison garbled circuit with extra input, for further description, set the garbled circuit for comparing two 1-bit integers x (l) and y (l) as shown in Figure 2, referring to Figure 2, the client holds With input x (l) (x 1 -x l ), the image segmentation execution server holds the input y (l) (y 1 -y l ). In order to hide the input of the CMP in Figure 2, the image segmentation execution server is introduced in the circuit An extra variable c that satisfies:

Figure BDA00027544918500001115
此时，电路的构造如图3所示，
Figure BDA00027544918500001116
表示带额外输入的乱码比较电路的输出。
Figure BDA00027544918500001115
At this time, the structure of the circuit is shown in Figure 3,
Figure BDA00027544918500001116
Represents the output of a garbled compare circuit with extra inputs.

S304.设背景累计值为

Figure BDA0002754491850000121
前景累计值为
Figure BDA0002754491850000122
背景计数值为
Figure BDA0002754491850000123
前景技术值为
Figure BDA0002754491850000124
图像分割执行服务器初始化背景累计值
Figure BDA0002754491850000125
前景累计值
Figure BDA0002754491850000126
及背景计数值
Figure BDA0002754491850000127
前景计数值
Figure BDA0002754491850000128
并进行像素计算，公式为：S304. Set the background cumulative value to be
Figure BDA0002754491850000121
The prospect cumulative value is
Figure BDA0002754491850000122
The background count is
Figure BDA0002754491850000123
The future technical value is
Figure BDA0002754491850000124
Image segmentation execution server initializes background cumulative value
Figure BDA0002754491850000125
Prospect cumulative value
Figure BDA0002754491850000126
and background counts
Figure BDA0002754491850000127
Foreground count value
Figure BDA0002754491850000128
And perform pixel calculation, the formula is:

Figure BDA0002754491850000129
Figure BDA0002754491850000129

Figure BDA00027544918500001210
Figure BDA00027544918500001210

Figure BDA00027544918500001211
Figure BDA00027544918500001211

计算下一个阈值

Figure BDA00027544918500001212
公式为：Calculate the next threshold
Figure BDA00027544918500001212
The formula is:

Figure BDA00027544918500001213
Figure BDA00027544918500001213

S305.图像分割执行服务器通过与隐私服务提供商共同执行绝对值乱码电路比较，获取

Figure BDA00027544918500001214
图像分割执行服务器将
Figure BDA00027544918500001215
及Δ+r′给隐私服务提供商，r′为用于修饰的随机值，隐私服务提供商判断阈值计算是否达到阈值标准ε，若是，阈值迭代计算结束；否则，返回步骤S304，另外，图像分割执行服务器通过与隐私服务提供商共同执行绝对值乱码电路时是与0进行比较的。S305. The image segmentation execution server performs absolute value garbled circuit comparison with the privacy service provider to obtain
Figure BDA00027544918500001214
The image segmentation execution server will
Figure BDA00027544918500001215
and Δ+r' to the privacy service provider, r' is a random value for modification, the privacy service provider judges whether the threshold calculation reaches the threshold standard ε, if so, the threshold iteration calculation ends; otherwise, return to step S304, in addition, the image When the segmentation execution server executes the absolute value garbled circuit together with the privacy service provider, it is compared with 0.

在本实施例中，图像分割执行服务器在拥有加密图像

Figure BDA00027544918500001216
及加密图像
Figure BDA00027544918500001217
的加密阈值Ti后，图像分割执行服务器与隐私服务提供商利用比较乱码电路进行加密图像
Figure BDA00027544918500001218
的二值化，得到加密分割图像
Figure BDA00027544918500001219
In this embodiment, the image segmentation execution server owns the encrypted image
Figure BDA00027544918500001216
and encrypted images
Figure BDA00027544918500001217
After the encryption threshold T i , the image segmentation execution server and the privacy service provider use the comparison garbled circuit to encrypt the image
Figure BDA00027544918500001218
The binarization of , obtains the encrypted segmentation image
Figure BDA00027544918500001219

为进一步验证本发明所提方法的有效性，以图4所示的加密图像分割实验效果作为说明，参见图4，其中，a列为原图像，b列为明文域图像分割结果，c、d、e列是基于本发明所提方法，利用数据打包技术，打包尺寸分别为4×4、8×8和16×16的加密域图像分割结果，通过图4可以看出，利用本发明所提出的方法进行图像分割时，b、c、d列加密域图像分割的结果与明文域图像分割结果基本一样，加密域分割保证了隐私安全，而利用数据打包技术又能降低空间资源的占用度和计算复杂度，做到了对当前图像分割方法的综合优化。In order to further verify the effectiveness of the method proposed in the present invention, the experimental effect of the encrypted image segmentation shown in Fig. 4 is used as an illustration, see Fig. 4, where a is the original image, b is the plaintext domain image segmentation result, c, d , column e is based on the method proposed in the present invention, using data packaging technology, the packaging size is 4 × 4, 8 × 8 and 16 × 16 encryption domain image segmentation results, as can be seen from Figure 4, using the proposed method of the present invention. When image segmentation is performed using the method of , the results of image segmentation in columns b, c, and d in the encrypted domain are basically the same as those in the plaintext domain. Computational complexity achieves comprehensive optimization of current image segmentation methods.

附图中描述位置关系的用于仅用于示例性说明，不能理解为对本专利的限制；显然，本发明的上述实施例仅是为清楚地说明本发明所作的举例，而并非是对本发明的实施方式的限定。对于所属领域的普通技术人员来说，在上述说明的基础上还可以做出其它不同形式的变化或变动。这里无需也无法对所有的实施方式予以穷举。凡在本发明的精神和原则之内所作的任何修改、等同替换和改进等，均应包含在本发明权利要求的保护范围之内。The positional relationships described in the accompanying drawings are for illustrative purposes only, and should not be construed as limitations on this patent; obviously, the above-mentioned embodiments of the present invention are only examples for clearly illustrating the present invention, rather than for the present invention. Limitations of Embodiments. For those of ordinary skill in the art, changes or modifications in other different forms can also be made on the basis of the above description. There is no need and cannot be exhaustive of all implementations here. Any modifications, equivalent replacements and improvements made within the spirit and principle of the present invention shall be included within the protection scope of the claims of the present invention.

Claims (10)

1. An encrypted domain image segmentation optimization method based on a data packaging technology is characterized by comprising the following steps:
s1, a privacy service provider generates a public key pk and a private key sk, sends the public key pk and the private key sk to a client, and sends the private key sk to an image segmentation execution server;
s2, the client encrypts the image by using the public key pk, packages the data, compresses the size of the encrypted image, and sends the encrypted and packaged image to an image segmentation execution server;
s3, the image segmentation execution server and the privacy service provider interact through a multi-party security calculation and code messing circuit technology to obtain an encrypted segmentation image;
and S4, the image segmentation execution server sends the encrypted segmented image to the client, and the client decrypts the image to obtain a final edge image.
2. The method for optimizing image segmentation in encrypted domain based on data packing technology according to claim 1, wherein in the step S2, the data packing procedure for compressing the size of the encrypted image is as follows:
the client encrypts the image by using the public key pk to form an encrypted image
Figure FDA0002754491840000011
Size of sm×snTo encrypt the image
Figure FDA0002754491840000012
L-block pictures divided into the same size, denoted as: i is0,I1,…,IL-1Executing the formula:
Figure FDA0002754491840000013
wherein ,
Figure FDA0002754491840000014
using the public key pk pair IPPacked to the size of
Figure FDA0002754491840000015
Compressed image of
Figure FDA0002754491840000016
Figure FDA0002754491840000017
Representing compressed images
Figure FDA0002754491840000018
The size parameter of (a), t, Q all represent the packing participation parameter,
Figure FDA0002754491840000019
Ik(i, j) represents the pixel value of the image with index (i, j) in the k-th block picture.
3. The method for optimizing image segmentation in encrypted domain based on data packing technology according to claim 2, wherein after step S2, step S3 further includes: image segmentation execution server compresses image
Figure FDA00027544918400000110
Performing Gaussian filtering, after Gaussian filteringThe results are expressed as
Figure FDA00027544918400000111
Satisfies the following conditions:
Figure FDA00027544918400000112
the result of the multiplication is retained, expressed as:
Figure FDA00027544918400000113
wherein m is 0,1, …, sm；n＝0,1,…,snAll represent image size parameters, G is a Gaussian filter template, Q is a quantization parameter of the Gaussian filter template, sgIs the gaussian filter template size.
4. The encryption domain image segmentation optimization method based on data packaging technology as claimed in claim 3, wherein the step S3 of interacting between the image segmentation execution server and the privacy service provider through multi-party secure computation and scrambling code circuit technology includes: for the encrypted packed data after Gaussian filtering
Figure FDA0002754491840000021
And comparing, comprising the following steps:
s31, using a horizontal Sobel filtering template GxFor example, the image segmentation execution server constructs two temporary templates: the portion where the template coefficient is positive is represented as
Figure FDA0002754491840000022
The portion where the sum template coefficient is negative is represented as
Figure FDA0002754491840000023
The construction formula is as follows:
Figure FDA0002754491840000024
Figure FDA0002754491840000025
wherein ,
Figure FDA0002754491840000026
and
Figure FDA0002754491840000027
respectively, the positive and negative parts of the coefficients in the horizontal Sobel convolution kernel,
Figure FDA0002754491840000028
s32, the image segmentation execution server generates two random numbers alpha+、α-And a formula is executed to pack two random numbers:
Figure FDA0002754491840000029
Figure FDA00027544918400000210
wherein ,
Figure FDA00027544918400000211
representing a random number alpha+Values after data packing;
Figure FDA00027544918400000212
representing a random number alpha-Performing data packing to obtain a value, and generating an intermediate ciphertext by an image segmentation execution server based on the packed random number
Figure FDA00027544918400000213
And is perpendicular to
Figure FDA00027544918400000214
Respectively satisfy:
Figure FDA00027544918400000215
Figure FDA00027544918400000216
the image segmentation execution server divides the intermediate ciphertext
Figure FDA00027544918400000217
And
Figure FDA00027544918400000218
sending to a privacy service provider;
s33, the privacy service provider utilizes the private key sk to pair the intermediate ciphertext
Figure FDA00027544918400000219
And
Figure FDA00027544918400000220
decrypting and obtaining by segmentation
Figure FDA00027544918400000221
And
Figure FDA00027544918400000222
and using a random number alpha1Generating any value delta in the set delta and the set deltaiSatisfies the following conditions:
Figure FDA00027544918400000223
wherein i is 0,1, …, L-1, c is an additional variable, c belongs to {0,1}, and is randomly chosen for the privacy service provider;
s34, according to the random number alpha+、α-The image segmentation execution server calculates an intermediate parameter σ0And sigma1And satisfies the following conditions: sigma0＝α+-α-，σ1＝α--α+Then by a random number alpha2Modifying sigmaiThe formula is as follows:
Figure FDA0002754491840000031
wherein ,
Figure FDA0002754491840000032
denotes the modified σiI is 0, 1; the image segmentation execution server holds the image segmentation execution server after performing the lossy transmission between the image segmentation execution server and the privacy service provider
Figure FDA0002754491840000033
Privacy service provider holds deltai；
S35, creating a garbled circuit by the image segmentation execution server, generating an encrypted garbled circuit table GCT corresponding to the comparison circuit C with additional input, and decoding the garbled circuit table GCT by the other party to hold a value
Figure FDA0002754491840000034
And deltaiCalculating the output of the final circuit
Figure FDA0002754491840000035
Data encryption and packaging based on comparison messy code circuit
Figure FDA0002754491840000036
The image segmentation execution server obtains the comparison resultr∈{0,1}。
5. The encryption domain image segmentation optimization method based on the data packaging technology as claimed in claim 4, wherein the image segmentation execution server holds the image segmentation execution server in the process of performing the lost transmission between the image segmentation execution server and the privacy service provider
Figure FDA0002754491840000037
And
Figure FDA0002754491840000038
the privacy service provider holds c ∈ {0,1}, the image segmentation execution server cannot obtain the value of c, and the privacy service provider cannot obtain the value of c
Figure FDA0002754491840000039
Privacy service provider generation
Figure FDA00027544918400000310
And will be
Figure FDA00027544918400000311
Sending to an image segmentation execution server which performs the segmentation by subtracting alpha2Obtaining
Figure FDA00027544918400000312
6. The encryption domain image segmentation optimization method based on data packaging technology as claimed in claim 5, wherein the step S3 of interacting between the image segmentation execution server and the privacy service provider through multi-party secure computation and scrambling code circuit technology further includes: for the encrypted packed data after Gaussian filtering
Figure FDA00027544918400000313
Sobel filtering edge detection is carried out to obtain an encrypted image
Figure FDA00027544918400000314
The method comprises the following steps:
the privacy service provider encrypts the set { delta } to obtain an encrypted set
Figure FDA00027544918400000315
And sending to an image segmentation execution server;
for encrypted collections
Figure FDA00027544918400000316
Each of which is
Figure FDA00027544918400000317
The image segmentation execution server with the comparison result r executes a formula to obtain
Figure FDA00027544918400000318
The formula is as follows:
Figure FDA00027544918400000319
wherein ,α2Random numbers generated for the image segmentation execution server in step S3, based on
Figure FDA00027544918400000320
The image segmentation execution server may acquire the horizontal component
Figure FDA00027544918400000321
And the perpendicular component
Figure FDA00027544918400000322
Then according to the formula
Figure FDA00027544918400000323
Constructing Sobel filtering result as encrypted image
Figure FDA00027544918400000324
7. The method for optimizing image segmentation of encryption domain based on data packaging technology as claimed in claim 6, wherein the step S3 of interacting between the image segmentation execution server and the privacy service provider through multi-party secure computing and scrambling circuit technology further includes: obtaining an encrypted image
Figure FDA0002754491840000041
The encryption threshold of (2) comprises the steps of:
s301, setting an encrypted image
Figure FDA0002754491840000042
Has a current threshold of
Figure FDA0002754491840000043
The image segmentation execution service first sets an encrypted image
Figure FDA0002754491840000044
Middle maximum pixel value
Figure FDA0002754491840000045
And minimum pixel value
Figure FDA0002754491840000046
An initial threshold is then generated using a division protocol
Figure FDA0002754491840000047
The formula is as follows:
Figure FDA0002754491840000048
wherein SecDiv represents a division protocol;
s302, the image segmentation execution server performs segmentation according to the current threshold value
Figure FDA0002754491840000049
To encrypt an image
Figure FDA00027544918400000410
Is divided into a foreground part and a background part, wherein the foreground part refers to that the value of a pixel is more than TiThe background portion refers to a pixel having a value of T or lessi；
S303, utilizing a messy code circuit to compare the current threshold value
Figure FDA00027544918400000411
And encrypting the pixel values
Figure FDA00027544918400000412
Comparing to obtain a comparison result { };
s304, setting the background cumulative value as
Figure FDA00027544918400000413
The accumulated value of the foreground is
Figure FDA00027544918400000414
The background count value is
Figure FDA00027544918400000415
A foreground technical value of
Figure FDA00027544918400000416
Image segmentation execution server initialization background integrated value
Figure FDA00027544918400000417
Accumulated value of foreground
Figure FDA00027544918400000418
And background count value
Figure FDA00027544918400000419
Foreground count value
Figure FDA00027544918400000420
And pixel calculation is carried out, and the formula is as follows:
Figure FDA00027544918400000421
Figure FDA00027544918400000422
Figure FDA00027544918400000423
calculating the next threshold
Figure FDA00027544918400000424
The formula is as follows:
Figure FDA00027544918400000425
s305, the image segmentation execution server obtains the absolute value scrambled code through carrying out absolute value scrambled code circuit comparison with the privacy service provider
Figure FDA00027544918400000426
The image segmentation execution server will
Figure FDA00027544918400000427
And delta + r 'to a privacy service provider, wherein r' is a random value used for modification, the privacy service provider judges whether the threshold calculation reaches a threshold standard epsilon, and if yes, the threshold iteration meterFinishing the calculation; otherwise, return to step S304.
8. The method of claim 7, wherein the step S303 utilizes a scrambling circuit to optimize the current threshold value
Figure FDA0002754491840000051
And encrypting the pixel values
Figure FDA0002754491840000052
The process of comparison was:
A. image segmentation execution server generates random number gamma1And β ∈ {0,1}, based on the random number γ1And beta is belonged to {0,1}, and calculation is carried out
Figure FDA0002754491840000053
And will be
Figure FDA0002754491840000054
And sending the information to the privacy service provider, wherein the formula is as follows:
Figure FDA0002754491840000055
B. privacy service provider will
Figure FDA0002754491840000056
To obtain t1Simultaneously generating c ∈ {0,1}, holding t1And c privacy service provider and holding random number gamma1The image segmentation execution server executes the comparison scrambling code circuit with additional input, and the privacy service provider obtains the modified result
Figure FDA0002754491840000057
C. According to
Figure FDA0002754491840000058
And c, the privacy service provider calculates the encrypted pair t2The formula is as follows:
Figure FDA0002754491840000059
privacy service provider will t2Sending the image to an image segmentation execution server;
D. the image segmentation execution server confirms the final comparison result lambda according to beta: if β is 1, the image division execution server divides t2The image division execution server changes the order of the image division units to the order of the image division units, and then takes the result as λ, and if β is 0, the image division execution server divides t into2As a result λ.
9. The encryption domain image segmentation optimization method based on the data packaging technology as claimed in claim 8, wherein in step S305, the image segmentation execution server compares the absolute value scrambling code circuit with 0 by executing the absolute value scrambling code circuit together with the privacy service provider.
10. The method of claim 9, wherein the image segmentation execution server possesses the encrypted image
Figure FDA00027544918400000510
And encrypting the image
Figure FDA00027544918400000511
Encryption threshold value T ofiThen, the image segmentation execution server and the privacy service provider encrypt the image by using the comparison scrambling code circuit
Figure FDA00027544918400000512
Obtaining an encrypted segmented image
Figure FDA00027544918400000513
CN202011197779.2A 2020-10-30 2020-10-30 An encryption domain image segmentation optimization method based on data packaging technology Active CN112288757B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011197779.2A CN112288757B (en) 2020-10-30 2020-10-30 An encryption domain image segmentation optimization method based on data packaging technology

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011197779.2A CN112288757B (en) 2020-10-30 2020-10-30 An encryption domain image segmentation optimization method based on data packaging technology

Publications (2)

Publication Number Publication Date
CN112288757A true CN112288757A (en) 2021-01-29
CN112288757B CN112288757B (en) 2023-09-19

Family

ID=74353418

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011197779.2A Active CN112288757B (en) 2020-10-30 2020-10-30 An encryption domain image segmentation optimization method based on data packaging technology

Country Status (1)

Country Link
CN (1) CN112288757B (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113159316A (en) * 2021-04-08 2021-07-23 支付宝(杭州)信息技术有限公司 Model training method, method and device for predicting business
WO2025001930A1 (en) * 2023-06-25 2025-01-02 中兴通讯股份有限公司 Image filtering methods and apparatus

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101409831A (en) * 2008-07-10 2009-04-15 浙江师范大学 Method for processing multimedia video object
CN103347170A (en) * 2013-06-27 2013-10-09 郑永春 Image processing method used for intelligent monitoring and high-resolution camera applied in image processing method
CN105959770A (en) * 2016-07-21 2016-09-21 李勇 Communication method using data line
CN110008717A (en) * 2019-02-26 2019-07-12 东北大学 Support the decision tree classification service system and method for secret protection
CN110298363A (en) * 2019-06-27 2019-10-01 西安电子科技大学 A kind of secret protection LBP feature extracting method on encrypted image
CN111526148A (en) * 2020-04-26 2020-08-11 中山大学 A secure denoising system and method for encrypted audio in cloud computing environment
CN111522973A (en) * 2020-04-16 2020-08-11 重庆邮电大学 A Privacy-Preserving Image Retrieval Method Based on Compressed Sensing

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101409831A (en) * 2008-07-10 2009-04-15 浙江师范大学 Method for processing multimedia video object
CN103347170A (en) * 2013-06-27 2013-10-09 郑永春 Image processing method used for intelligent monitoring and high-resolution camera applied in image processing method
CN105959770A (en) * 2016-07-21 2016-09-21 李勇 Communication method using data line
CN110008717A (en) * 2019-02-26 2019-07-12 东北大学 Support the decision tree classification service system and method for secret protection
CN110298363A (en) * 2019-06-27 2019-10-01 西安电子科技大学 A kind of secret protection LBP feature extracting method on encrypted image
CN111522973A (en) * 2020-04-16 2020-08-11 重庆邮电大学 A Privacy-Preserving Image Retrieval Method Based on Compressed Sensing
CN111526148A (en) * 2020-04-26 2020-08-11 中山大学 A secure denoising system and method for encrypted audio in cloud computing environment

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
江南;张帆;刘文予;: "适合于网络传输的二次置乱图像加密算法", 中国图象图形学报, no. 05, pages 1 - 4 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113159316A (en) * 2021-04-08 2021-07-23 支付宝(杭州)信息技术有限公司 Model training method, method and device for predicting business
WO2025001930A1 (en) * 2023-06-25 2025-01-02 中兴通讯股份有限公司 Image filtering methods and apparatus

Also Published As

Publication number Publication date
CN112288757B (en) 2023-09-19

Similar Documents

Publication Publication Date Title
Hossain et al. Cloud-assisted secure video transmission and sharing framework for smart cities
Singh et al. Towards integrating image encryption with compression: A survey
US10320557B2 (en) Encoder, decoder and methods employing partial data encryption
Patel et al. Lazy wavelet transform based steganography in video
Amalarethinam et al. Image encryption and decryption in public key cryptography based on MR
CN106941402B (en) A Reversible Hiding Method in Homomorphic Ciphertext Domain
Soleymani et al. A survey on principal aspects of secure image transmission
CN112288757A (en) An optimization method for image segmentation in encrypted domain based on data packaging technology
Das et al. An image secret sharing technique with block based image coding
CN107146192A (en) Image security protection method and device
Bhadra et al. Securing data using elliptic curve cryptography and least significant bit steganography
Zhang et al. Computation outsourcing meets lossy channel: Secure sparse robustness decoding service in multi-clouds
WO2021196960A1 (en) Encrypted video call method and apparatus, and device and storage medium
CN112330551B (en) Remote sensing image outsourcing noise reduction method based on secret sharing
Shah et al. Paillier cryptosystem based mean value computation for encrypted domain image processing operations
Bezerra et al. Fast chaotic image encryption with simultaneous permutation and diffusion for IoT applications
Zhou et al. Image encryption based on edge information
Addepalli et al. A hybrid security framework for medical data in IoT applications
Shankar et al. Image encryption for mobile devices
Mahasree et al. An improved reversible data hiding for encrypted images over cloud
Kaur et al. Hiding Data in Video Sequences using LSB with Elliptic Curve Cryptography
Britto et al. Gaussian noise analysis in elliptic curve encrypted images
Tripathi et al. Review on Metamorphic Cryptography: A Combined Approach of Cryptography & Steganography Techniques
Malathi et al. Enhanced image steganography using AES & SPIHT compression
Ali et al. Implementation of a cryptograpgy algorithm for image transmission

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