CN109949136B - Image transaction method for protecting privacy - Google Patents
Image transaction method for protecting privacy Download PDFInfo
- Publication number
- CN109949136B CN109949136B CN201910222012.1A CN201910222012A CN109949136B CN 109949136 B CN109949136 B CN 109949136B CN 201910222012 A CN201910222012 A CN 201910222012A CN 109949136 B CN109949136 B CN 109949136B
- Authority
- CN
- China
- Prior art keywords
- image
- buyer
- cloud server
- seller
- features
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Abstract
The invention discloses an image transaction method for protecting privacy, which comprises the following steps: step 1, a buyer sends a purchase image demand to a cloud server; step 2, spreading the demand; step 3, uploading the characteristics; step 4, selecting an image; step 5, informing the buyer; step 6, paying by the buyer; step 7, notifying the seller; and 8, transmitting the encrypted image. The image transaction method has wide data sources, so that the data sources are not limited to the data published on the Internet; the buyer is guaranteed, the quantity and the quality of the image sets finally purchased by the buyer can be guaranteed through screening, and budget feasibility and user integrity can also be guaranteed; the seller is guaranteed, and the copyright and the privacy of the image can be protected by the method; the practicability is strong, centralized storage is not needed for the cloud server, only a small amount of calculation is needed, and calculation and communication overhead are low for the mobile equipment. The method of the present invention can also be used in various data paid collection and trading systems.
Description
Technical Field
The invention relates to the fields of crowdsourcing transaction, mobile application and privacy protection, in particular to an image transaction method for protecting privacy.
Background
In recent years, deep learning makes a major breakthrough in many fields, but a large number of high-quality training data sets are still a big bottleneck, and the existing data collection method (for example, capturing data by using crawler software) has many limitations, such as the quality of data cannot be guaranteed, the privacy of users may be violated, and the like; on the other hand, user devices continue to generate and store large amounts of data, which is not efficiently utilized. If the data can be collected in a certain way, the problem of lack of training data sets in the deep learning process and other data set requirement problems can be solved to a great extent.
Currently, some crowd-sourced image collection technologies either cannot measure the quality of an image or need to provide some specific metadata as a basis, and only have a limited application range for a specific object (such as a photo at a specific position). In addition, due to the complex structure of the image, the existing technology has not solved the privacy protection problem in the user image transaction.
Disclosure of Invention
Based on the problems in the prior art, the invention aims to provide an image transaction method for protecting privacy, which can be used for conveniently purchasing images of a plurality of sellers on the premise of protecting the privacy of the image content of a user.
The purpose of the invention is realized by the following technical scheme:
the embodiment of the invention provides an image transaction method for protecting privacy, which comprises the following steps:
step 1, a buyer sends a purchase image demand to a cloud server;
and step 3, uploading characteristics: after receiving the broadcasted image purchasing demand request, the potential seller searches whether an image meeting the requirement exists in local equipment of the seller, if so, extracts the characteristics of the image meeting the requirement from the local equipment, and uploads the extracted image characteristics and the quoted price to the cloud server;
According to the technical scheme provided by the invention, the privacy-protecting image transaction method provided by the embodiment of the invention has the beneficial effects that:
after the image to be purchased by the buyer is selected from the image characteristics provided by the plurality of sellers according to the request of the buyer through the cloud server, the encrypted image of the seller is transmitted to the buyer after the buyer pays, and the transaction process of the image purchased by the buyer and the seller is completed on the premise of effectively protecting the image content privacy of the seller.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings needed to be used in the description of the embodiments are briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings based on the drawings without creative efforts.
Fig. 1 is a flowchart of an image transaction method for protecting privacy according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention are clearly and completely described below with reference to the specific contents of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments of the present invention without making any creative effort, shall fall within the protection scope of the present invention. Details which are not described in detail in the embodiments of the invention belong to the prior art which is known to the person skilled in the art.
The embodiment of the invention provides an image transaction method for protecting privacy, which comprises the following steps:
step 1, a buyer sends a purchase image demand to a cloud server;
and step 3, uploading characteristics: after receiving the broadcasted image purchasing demand request, the potential seller searches whether an image meeting the requirement exists in local equipment of the seller, if so, extracts the characteristics of the image meeting the requirement from the local equipment, and uploads the extracted image characteristics and the quoted price to the cloud server;
The method further comprises, after step 8:
In the above method, in the cloud arbitration step, after verifying that the image confirms which of the buyer and the seller is dishonest, the party punishing dishonest is:
the image is verified to be dishonest by the buyer or dishonest by the seller, if the seller is dishonest, the cloud server sends the payment corresponding to the image to the seller, and if the seller is dishonest, the cloud server returns the unpaid payment to the buyer.
In the above method, before step 9, the method further includes: buyer verification step: after receiving the feature sets of all the image ciphertexts and images sent by the cloud server, the buyer uses the own private key SkDecrypting all the image ciphertexts to obtain all the images;
and verifying all images by using the feature set of the images, and confirming whether all the images are matched with the corresponding features.
In the buyer verifying step of the above method, the method further includes:
and the buyer extracts the features of the image from the unmatched image, compares the features with the corresponding features in the feature set of the image, and if the features are not matched, submits the features to the cloud server for arbitration.
In step 1 of the above method, the image purchasing requirement of the buyer includes:
text description, image proof, budget range, data quality standard, and buyer public key.
In step 4 of the method, the step of screening out a subset of images within the budget given by the buyer to perform a transaction includes:
and screening out the images meeting the requirements, and screening out a high-quality image forming image subset according to the requirements of the buyer.
In step 8 of the method, after receiving the image ciphertexts, the cloud server generates a hash value for verification for each image ciphertext.
In step 3 of the method, in extracting the features of the image meeting the requirements from the local device, the features of the image meeting the requirements are extracted from the local device by adopting a feature extraction mode of deep neural network image recognition. Specifically, the deep neural network can extract features in the following ways: firstly, extracting 1000-dimensional features of an fc8 layer of an image by adopting a Vgg16 deep neural network, and then converting the 1000-dimensional features into 2-dimensional features by using a pre-trained automatic coding machine. The automatic coding machine comprises three full-connected layers and two softplus layers, and approximately 200000 pictures in an Imagenet data set are used for training. Specifically, the first level of the auto-encoder is a 1000 × 256 fully-connected level followed by a softplus level, followed by a 256 × 64 fully-connected level followed by a softplus level, and finally a 64 × 2 fully-connected level.
According to the transaction method, a buyer can directly send an image request (including text description, image samples and the like) purchased by the buyer to the cloud server, then the cloud server broadcasts the image request to potential sellers, the sellers search whether photos meeting the requirements are available for sale in local equipment after receiving the image request, if yes, specified information is uploaded, and finally the cloud server selects which images are purchased according to given rules. The mode changes the mode of data transaction at present, and a fixed data center is not needed for storing and selling the data sets which are transacted in a centralized way, but all the data sets are stored in the own equipment of the user in a scattered way.
The transaction method of the invention utilizes crowdsourcing technology to collect mass image data in user equipment in a paid mode to obtain an image set meeting specified requirements. The method is flexible and convenient, can ensure the quantity and quality of the finally purchased data sets, and also ensures the budget feasibility and the user integrity; in addition, different from the mode that the original image needs to be uploaded in the existing data transaction mode (although the copyright protection modes such as watermarking and the like may exist, the image content cannot be protected), the method and the device do not need to upload the original image, and therefore the copyright and the privacy of each user image are fundamentally protected. The invention enables mass mobile equipment users to participate in the image transaction process, thereby greatly widening the data sources.
The embodiments of the present invention are described in further detail below.
The image transaction method for protecting privacy provided by the embodiment of the invention is based on crowdsourcing technology, and can collect mass image data in user equipment in a paid manner on the premise of protecting privacy to obtain an image set meeting specified requirements and provide the image set to a demand side.
The image transaction method can be divided into two privacy protection levels, and in the first privacy protection level, the original image of a seller and the personal identity of the seller are protected; in the second privacy protection level, the invention also prevents the image characteristics of the seller from being leaked on the basis of the first privacy protection level.
In the first privacy protection level, original images of the seller are protected from being leaked to any party who does not pay a reward, and meanwhile, the personal identity of the seller is guaranteed not to be leaked to a buyer; in the privacy protection class, the invention comprises the steps of:
step 1, a buyer sends the own image purchasing requirement to a cloud server, wherein the image purchasing requirement comprises text description, image sample sheets, budget limitation, a data quality standard, a buyer public key and the like;
and step 3, uploading characteristics: after receiving a broadcast request, a potential seller checks whether an image meeting the requirement exists in local equipment, if so, the characteristic of the image is extracted locally in the equipment by using a characteristic extraction algorithm of deep neural network picture recognition, and the characteristic of the image and a quoted price are uploaded to a cloud server; the characteristic extraction algorithm for deep neural network picture identification specifically comprises the following steps: firstly, extracting 1000-dimensional features of an fc8 layer of an image by adopting a Vgg16 deep neural network, and then converting the 1000-dimensional features into 2-dimensional features by using a pre-trained automatic coding machine. The automatic coding machine comprises three full-connection layers and two softplus layers, and is trained by using approximately 200000 pictures in an Imagenet data set; specifically, the first layer of the automatic coding machine is a 1000 × 256 full-link layer, followed by a softplus layer, followed by a 256 × 64 full-link layer, followed by a softplus layer, and finally followed by a 64 × 2 full-link layer;
the set mechanism adopts three measurement standards of quantity, matching degree and diversity, and for the quantity, the adopted mechanism is to sort the quotations of users from small to large, such as c _1< ═ c _2< ═ … < ═ c _ n, and then find the maximum subscript k meeting c _ k < ═ B/k; for the users with low price of the former k, the reward is min { B/k, c _ { k +1} }, and the rewards of the other users are 0; for the matching degree and diversity, a Budget feasible mechanism (Budget feasible mechanism) is used (see the method proposed in the a Budget objective input mechanism for weighted conversion in mobile crowdsensing article specifically).
step 11, after verification and arbitration, the cloud server gives consideration to honest sellers; if a dishonest seller exists, the cloud server returns the unpaid payment to the buyer.
Secondly, since some existing technologies can extract a large amount of information from image features, it is necessary to protect the features of the image. On the basis of the first privacy protection level, in the second privacy protection level, the image characteristics of the seller are prevented from being leaked; in the privacy protection class, the invention comprises the steps of:
step 21, the buyer sends the image requirement of the buyer to a cloud server, wherein the image requirement comprises text description, image sample sheets, budget limit, data quality standard, a buyer public key and the like;
step 22, propagating the demand: the cloud server sends the demands of the buyers, such as text description, image samples and the like, to the potential sellers;
and step 23, uploading the processed characteristics. After receiving a broadcast request, a potential seller checks whether an image meeting the requirement exists in local equipment, if so, the characteristics of the image are extracted and processed locally in the equipment, and then the processed characteristics and the quotation of the image are uploaded to a cloud server;
step 24, selecting an image: and after receiving the processed characteristics and the offers uploaded by the sellers, the cloud server screens a subset in a budget allowable range to carry out transaction. The screening comprises the following steps: screening out images meeting requirements, screening out a high-quality image set according to user requirements, and using a proper mechanism to ensure budget feasibility and integrity of seller quoted prices;
step 25, notify buyer: the cloud server informs the buyer of the selected set and the total price to be paid;
step 26, payment by the buyer: the buyer pays the total price to the cloud server by using a certain protocol agreed in advance;
step 27, notifying the seller: after receiving the payment of the buyer, the cloud server transmits the public key P of the buyerkSending the images to all selected sellers and informing the sellers to upload the selected images;
step 28, transmitting the encrypted image: the seller receiving the notification would select the image PkAnd uploading the encrypted data to a cloud server. And after receiving the image data, the cloud server generates a hash value for the ciphertext of each image for later verification. Then the cloud server packs all the ciphertexts and sends the selected processed characteristics to the buyer;
step 29, buyer verification (optional step): the buyer uses his own private key SkAll images are decrypted. In order to prevent the dishonest seller from uploading the image which is not matched with the selected processed features, the buyer can re-extract the image features and compare the image features with the selected processed features; if not, submitting the data to a cloud server and arbitrating by the cloud server;
step 30, cloud arbitration: if the buyer reports that the image uploaded by the seller is inconsistent with the selected processed characteristics, the cloud server carries out arbitration to determine whether the buyer is dishonest or the seller is dishonest; the buyer can refuse to pay for the legal image by forging the verification result, and the seller can obtain illegal benefit by uploading the image which is not matched with the selected related information; the cloud server carries out arbitration and gives a certain penalty to a dishonest buyer or seller according to the fact;
and step 31, after verification and arbitration, the cloud server gives consideration to honest sellers. If a dishonest seller exists, the cloud server returns the unpaid payment to the buyer.
The image transaction method has wide data source, so that mass mobile equipment users can participate in the image transaction process, and the data source is not limited to the data published on the Internet; the buyer is guaranteed, and through screening, the quantity and the quality of the image sets finally purchased by the buyer can be guaranteed, and budget feasibility and user integrity can also be guaranteed; the seller is guaranteed, and the copyright and the privacy of the image can be protected by the method; the practicability is strong, centralized storage is not needed for the cloud server, only a small amount of calculation is needed, and calculation and communication overhead are low for the mobile equipment. The method of the present invention can also be used in various data paid collection and trading systems.
Examples
The embodiment provides an image transaction method for protecting privacy, which is divided into two privacy protection levels, specifically as follows:
in the first privacy protection level, the image purchasing system for protecting the privacy, which is provided by the invention, protects the original image of the seller from being leaked to any party without paying, and simultaneously ensures that the personal identity of the seller is not leaked to the buyer. In the privacy protection class, the method specifically includes the steps of:
step 1, a buyer sends a self image purchasing demand (namely an image demand) to a cloud server;
preferably, the image purchasing requirement specifically includes:
the description of the characters and the image samples are used for measuring whether the images to be sold by the seller and the samples belong to the same class; budget limits, i.e., the total cost of the last purchased image set does not exceed the budget;
the data quality standard comprises three screening modes (quantity is first, similarity is first, and diversity is first), and a buyer designates one of the three screening modes according to the actual situation of the buyer; the three data quality standards are illustrated in the following table:
and the buyer public key is used for finally encrypting the purchased original image.
and step 3, uploading characteristics: after receiving a broadcast request, a potential seller checks whether an image meeting the requirement exists in local equipment, if so, the characteristic of the image is extracted locally in the equipment by using a characteristic extraction algorithm of deep neural network picture recognition, and the characteristic of the image and a quoted price are uploaded to a cloud server; the features are extracted by using a trained Vgg16 deep neural network, and a trained automatic coding machine (3 layers of full-connected layers and 2 layers of softplus) is connected behind an fc8 layer of the features to obtain a two-dimensional vector as the features of the image; the quoted price uploaded by the seller is not necessarily equal to the real price of the image, but the real price of the image is only known by the seller, and the integrity of the quoted price of the seller is ensured by setting a proper mechanism; the set mechanism adopts three measurement standards of quantity, matching degree and diversity, and for the quantity, the adopted mechanism is to sort the quotations of users from small to large, such as c _1< ═ c _2< ═ … < ═ c _ n, and then find the maximum subscript k meeting c _ k < ═ B/k; for the users with low price of the former k, the reward is min { B/k, c _ { k +1} }, and the rewards of the other users are 0; for the matching degree and diversity, a Budget feasible mechanism (Budget feasible mechanism) is used (see the method proposed in the A Budget intensive mechanism for weighted conversion in mobile crowdsensing article specifically);
the specific treatment of the step 4 is as follows:
step 41 the cloud server extracts features from the image samples provided by the sellerTaking the characteristics as centers to obtain a series of centers O1，O2，…OmThen for each feature F offered by the selleriJudging whether a circle center O existsjSo that 0 < Fi-OjR, | ≦ R, if present, adding FiAdd candidate set, otherwise consider FiMismatch is not achieved;
in the candidate set, a trading image set is selected for a given budget B, step 42, using one of three options:
a. number-first screening selection: the features in the candidate set are arranged from small to large according to quotation to obtain c1≤c2≤…≤cnLet k satisfy ckMaximum subscript of ≦ B/k; the first k lowest priced images are selected and the payment for each image is min { B/k, ck+1}；
b. A limited-similarity priority screening mode: for each feature in the candidate set, its similarity is calculated:
here, the features in the candidate set all satisfy O < min { | Fi-OjR is less than or equal to R, so that the similarity of all the characteristics in the candidate set is a positive number between 0 and 1; then, selecting by using a BEACON algorithm within a budget range;
c. diversity-first screening: for each feature F in the candidate setiSetting its availability as FiThe area of a square with a center, a side length of d, and each side parallel to a coordinate axis, the total availability of a set is the area of the union of squares of all the features in the set, and it is noted that the area of the overlapped region can be calculated only once, and the availability of the set is calculated by using a line segment tree and a line scanning method. (ii) a So each feature FiWith respect to set P (F)iOut of P) is defined as FiAvailability to join in set PSubtracting the availability of P, in particular, when P is empty, the marginal value of all features is its own availability. (ii) a And finally, selecting the subset with the maximum availability in the budget range by adopting a BEACON algorithm.
step 11, after verification and arbitration, the cloud server gives consideration to honest sellers; if a dishonest seller exists, the cloud server returns the unpaid payment to the buyer.
(II) because some existing technologies can extract a large amount of information from image features, the image features are required to be protected, and on the basis of the first privacy protection level, the image features of the seller are prevented from being leaked in the second privacy protection level of the method; in the privacy protection class, the method specifically includes the steps of:
step 21, the buyer sends the image purchasing requirement (namely the image requirement) to the cloud server;
preferably, the image purchasing requirement specifically includes:
the description of the characters and the image samples are used for measuring whether the images to be sold by the seller and the samples belong to the same class; budget limits, i.e., the total cost of the last purchased image set does not exceed the budget;
the data quality standard adopts three screening modes (quantity is first, similarity is first, diversity is first), and a buyer designates one of the three modes according to the actual situation of the buyer;
and the buyer public key is used for finally encrypting the purchased original image.
Step 22, propagating the demand: the cloud server sends the demands of the buyers, such as text description, image samples and the like, to the potential sellers;
and step 23, uploading the processed characteristics: after receiving the broadcast request, the potential seller checks whether an image meeting the requirement exists in the local equipment, if so, the characteristics of the image are extracted and processed locally in the equipment, and then the processed characteristics and the quotation of the image are uploaded to the cloud server. Preferably, a trained Vgg16 deep neural network is used for picture recognition to extract features, a trained automatic coding machine (comprising 3 layers of full-connected layers and 2 layers of softplus layers) is connected behind an fc8 layer of the image to obtain a two-dimensional vector as the features of the image, then k points are randomly selected from a square which takes the features as the center and has the side length of x and each side parallel to a certain coordinate axis, and the average value of the k points is taken as the processed features; preferably, the quoted price uploaded by the seller is not necessarily equal to the true price of the image, but the true price of the image is only known by the seller, and the integrity of the quoted price of the seller is ensured through a proper mechanism;
step 24, selecting an image: after receiving the processed characteristics and offers uploaded by a plurality of sellers, the cloud server screens a subset in a budget allowable range for transaction;
the specific processing of step 24 is:
in step 241, the cloud server extracts processed features from the image sample provided by the seller, and uses the processed features as a center to obtain a series of centers O1, O2, … On, and then obtains a processed feature F for each feature provided by the selleriJudging whether a circle center O existsjSo that 0 < | Fi-OjR, | ≦ R, if present, adding FiAdd candidate set, otherwise consider FiMismatch is not achieved;
step 242, in the candidate set, for a given budget B, a selected transaction image set using one of the following three screening methods is used, specifically:
a. number-first screening method: the features in the candidate set are arranged from small to large according to the quotation to obtain c2 which is more than or equal to c1 and is more than or equal to … and cn, and k is satisfied with ckMaximum subscript of ≦ B/k; the first k lowest priced images are selected and the payment for each image is min { B/k, ck+1}；
b. A similarity priority screening mode: for each processed feature in the candidate set, its similarity is calculated:
here, all the processed features in the candidate set satisfy 0 < min { | Fi-OjR is less than or equal to | }, so the candidate setThe similarity of all the processed characteristics is a positive number between 0 and 1; then, selecting by using a BEACON algorithm within a budget range;
c. diversity-first screening: for each processed feature F in the candidate setiSetting its availability as FiThe area of a square which is central, has the side length of d and each side parallel to a certain coordinate axis, the total availability of a set is the area of a union set of squares in which all processed features are positioned in the set, and the area of an overlapped area can be calculated only once, wherein the availability of the set is calculated by using a line segment tree and a line scanning method; so each processed feature FiWith respect to set P (F)iOut of P) is defined as FiThe availability added to the set P minus the availability of P, in particular, when P is empty, the marginal value of all processed features is its own availability; and finally, selecting the subset with the maximum availability in the budget range by adopting a BEACON algorithm.
Step 24, notifying the buyer: the cloud server informs the buyer of the selected image set and the total price to be paid;
step 25, payment by the buyer: the buyer pays the total price to the cloud server by using a certain protocol agreed in advance;
step 26, notifying the seller: after receiving the payment of the buyer, the cloud server transmits the public key P of the buyerkSending the images to all selected sellers, and informing all the selected sellers to upload the selected images;
step 27, the encrypted image is transmitted. The seller receiving the notification uses the selected image PkAfter being encrypted, the data are uploaded to a cloud server; after the cloud server receives the image data, a hash value is generated for the ciphertext of each image for later verification. Then the cloud server packs all the ciphertexts and sends the selected set to the buyer;
buyer verification (optional step) step 28: the buyer uses his own private key SkDecrypting all images; to prevent a dishonest seller from uploading and selecting processed featuresIf the image is not matched, the buyer can extract the features of the image again and compare the features with the selected processed features, and the distance between the two dimensions is smaller than x; if not, submitting the data to the cloud end and arbitrating by the cloud end;
step 29, cloud arbitration: if the image uploaded by the seller reported by the buyer is inconsistent with the processed characteristics, the cloud server carries out arbitration to determine whether the buyer is dishonest or the seller is dishonest. Buyers may refuse to pay for legal images by forging the verification result, and sellers may obtain illegal benefits by uploading images which are not matched with the selected and processed characteristics; the cloud server carries out arbitration, if the hash value obtained after the image reported by the buyer is re-encrypted is the same as the reserved hash value of the cloud, and the distance between each dimension of the image after the characteristic is re-extracted and the selected processed characteristic is greater than x, the seller is not honest, otherwise, the buyer is not honest; the cloud server gives a certain penalty to the dishonest buyer or seller according to the arbitration result;
and step 30, after verification and arbitration, the cloud server gives consideration to honest sellers. If a dishonest seller exists, the cloud server returns the unpaid payment to the buyer.
The above description is only for the preferred embodiment of the present invention, but the scope of the present invention is not limited thereto, and any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope of the present invention are included in the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims (8)
1. A privacy-preserving image transaction method, comprising:
step 1, a buyer sends a purchase image demand to a cloud server;
step 2, spreading requirements: the cloud server broadcasts the purchase image demand to potential sellers;
and step 3, uploading characteristics: after receiving the broadcasted image purchasing demand request, the potential seller searches whether an image meeting the requirement exists in local equipment of the seller, if so, extracts the characteristics of the image meeting the requirement from the local equipment, and uploads the extracted image characteristics and the quoted price to the cloud server;
step 4, image selection: after receiving the image characteristics and offers uploaded by a plurality of sellers, the cloud server screens out an image subset in a budget range given by a buyer for transaction;
step 5, informing the buyer: the cloud server informs the buyer of the screened image subset and the total price to be paid;
step 6, payment of the buyer: the buyer pays the total price needing to be paid to the cloud server;
step 7, notifying the seller: after receiving the total price paid by the buyer, the cloud server sends the public key of the buyer to all sellers of the selected image and informs all sellers of uploading the selected image;
step 8, transmitting the encrypted image: the seller receiving the notification encrypts the selected image into an image ciphertext by using the public key and uploads the image ciphertext to the cloud server, and the cloud server packages all the image ciphertexts and sends the image ciphertexts and the feature set of the selected image to the buyer;
further comprising: step 9, cloud arbitration: after receiving the image ciphertext and the feature set of the image, the buyer reports to the cloud server that the image uploaded by the seller is inconsistent with the feature, and the cloud server performs arbitration and punishs a dishonest party after verifying that the image confirms whether the buyer or the seller is dishonest.
2. The privacy-preserving image transaction method according to claim 1, wherein in the cloud arbitration step, after verifying whether the image confirms the buyer or the seller is dishonest, the dishonest party is punished as follows:
the image is verified to be dishonest by a buyer or dishonest by a seller, if the seller is dishonest, the cloud server sends the reward corresponding to the image to the seller, and if the seller is dishonest, the cloud server refunds the unpaid reward corresponding to the image to the buyer.
3. The privacy-preserving image transaction method according to claim 1, further comprising, before step 9: buyer verification step: after receiving the feature sets of all the image ciphertexts and images sent by the cloud server, the buyer uses the own private key SkDecrypting all the image ciphertexts to obtain all the images;
and verifying all images by using the feature set of the images, and confirming whether all the images are matched with the corresponding features.
4. The privacy-preserving image transaction method according to claim 3, wherein in the buyer verifying step, the method further comprises:
and the buyer extracts the features of the image from the unmatched image, compares the features with the corresponding features in the feature set of the image, and if the features are not matched, submits the features to the cloud server for arbitration.
5. The privacy-preserving image transaction method according to any one of claims 1 to 4, wherein in step 1 of the method, the buyer's purchase image request includes:
text description, image proof, budget range, data quality standard, and buyer public key.
6. The privacy-preserving image transaction method according to any one of claims 1 to 4, wherein in step 4 of the method, the step of screening out a subset of images within the budget given by the buyer for transaction comprises:
and screening out the images meeting the requirements, and screening out a high-quality image forming image subset according to the requirements of the buyer.
7. The privacy-preserving image transaction method according to any one of claims 1 to 4, wherein in step 8 of the method, the cloud server generates a hash value for verification for each image ciphertext after receiving the image ciphertext.
8. The privacy-preserving image transaction method according to any one of claims 1 to 4, wherein in step 3 of the method, in extracting the features satisfying the required image from the local device, the features satisfying the required image are extracted from the local device by a feature extraction method of deep neural network picture recognition.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910222012.1A CN109949136B (en) | 2019-03-22 | 2019-03-22 | Image transaction method for protecting privacy |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910222012.1A CN109949136B (en) | 2019-03-22 | 2019-03-22 | Image transaction method for protecting privacy |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109949136A CN109949136A (en) | 2019-06-28 |
| CN109949136B true CN109949136B (en) | 2022-04-19 |
Family
ID=67010600
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910222012.1A Active CN109949136B (en) | 2019-03-22 | 2019-03-22 | Image transaction method for protecting privacy |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109949136B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPWO2023139803A1 (en) * | 2022-01-18 | 2023-07-27 |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CA3023294A1 (en) * | 2015-05-06 | 2016-11-10 | Paydatum Co. | Digital receipt processing and analytics system |
| CN108898464A (en) * | 2018-06-29 | 2018-11-27 | 合肥微商圈信息科技有限公司 | A kind of e-commerce platform marketing method |
| CN109376504A (en) * | 2018-09-26 | 2019-02-22 | 福州大学 | A kind of picture method for secret protection based on block chain technology |
-
2019
- 2019-03-22 CN CN201910222012.1A patent/CN109949136B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CA3023294A1 (en) * | 2015-05-06 | 2016-11-10 | Paydatum Co. | Digital receipt processing and analytics system |
| CN108898464A (en) * | 2018-06-29 | 2018-11-27 | 合肥微商圈信息科技有限公司 | A kind of e-commerce platform marketing method |
| CN109376504A (en) * | 2018-09-26 | 2019-02-22 | 福州大学 | A kind of picture method for secret protection based on block chain technology |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109949136A (en) | 2019-06-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| AU2018100150A4 (en) | Proof of delivery on the blockchain | |
| CN107292150B (en) | User identity confirmation method and device in security information processing | |
| Zhang et al. | Crowdbuy: Privacy-friendly image dataset purchasing via crowdsourcing | |
| CN116091068B (en) | Method for providing physical asset original authentication service by using decentralised identifier and heterogeneous tokens | |
| CN112884554B (en) | Auction method of Internet of things data auction system based on alliance chain | |
| WO2000033271A2 (en) | Method and apparatus for facilitating buyer-driven purchase orders on a commercial network system | |
| JP2000511672A (en) | Method and apparatus for a cryptographically assisted commercial network system to promote and support expert-based commerce | |
| US20190228429A1 (en) | Safety index for the calculation of a rating based on user generated reports or actions and rewards system therefor | |
| US20200274714A1 (en) | System for, method of, and server computer system for implementing transformation of an original entity into a verifiably authenticable entity in a heterogeneous communications network environment | |
| KR20210105066A (en) | Real estate dealing platform server having advenced reliability and search comvenience, and dealing method thereof | |
| KR20210105060A (en) | Real estate dealing platform server capable of custom-made article search and dealing method thereof | |
| CN111046078A (en) | Block chain-based credit investigation query method and device and electronic equipment | |
| CN105095691B (en) | A kind of method and apparatus sending digital publication | |
| CN109949136B (en) | Image transaction method for protecting privacy | |
| JP2013045460A (en) | E-commerce transaction method for intangible merchandise | |
| CN116503070B (en) | Digital asset right-determining and trading method, device, equipment and storage medium | |
| Wang et al. | A new quantum sealed-bid auction protocol with secret order in post-confirmation | |
| CN112927060B (en) | Land flow stable and safe transaction system and method | |
| US20220067808A1 (en) | Computer systems for peer-to-peer onboarding to an online marketplace | |
| JP2020134958A (en) | Program, information processing method, and information processing device | |
| KR20140055480A (en) | Method for registration and certification of real estate rights through networks | |
| JP2005352786A (en) | Electronic ticket vending method, electronic ticket vending/transferring method, server device, client device, program, and recording medium | |
| CN105913248B (en) | Online payment system based on mobile internet service application | |
| CN112464178B (en) | Data transaction copyright protection method based on blockchain and homomorphic encryption | |
| WO2018051139A1 (en) | Transaction validation |
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 |