Unlock instant, AI-driven research and patent intelligence for your innovation.

Image processing method and system based on privacy protection, medium and equipment

An image processing and privacy protection technology, applied in digital data protection, electronic digital data processing, character and pattern recognition, etc., can solve the problems of long ciphertext length, large computing resources, and low system efficiency, and avoid low efficiency, Good compatibility, tolerating the effect of intrusion

Pending Publication Date: 2021-07-30
RENMIN UNIVERSITY OF CHINA
View PDF5 Cites 1 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] At present, the privacy protection method used for image processing is mainly fully homomorphic encryption method, but the main purpose of the statistical data security protection in the current image processing is to make the attacker still unable to identify personal data when he has the image database, and lacks the data itself. and the protection of the model itself
In addition, fully homomorphic encryption consumes a lot of computing resources, and the ciphertext length is long after using fully homomorphic encryption, so the system efficiency is low
Existing privacy protection methods only target a specific graphics processing method and are not compatible

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Image processing method and system based on privacy protection, medium and equipment
  • Image processing method and system based on privacy protection, medium and equipment
  • Image processing method and system based on privacy protection, medium and equipment

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0032] This embodiment discloses an image processing method based on privacy protection, such as figure 1 shown, including the following steps:

[0033] S1 randomly divides the obtained image data into several parts, and stores each part of data in an independent server, and the servers do not collude.

[0034] like figure 2 As shown, the image data is first preprocessed, and various image data are converted into pixel matrices according to requirements, and the image data is standardized so that the image data are all within the same set range. Eliminate image data that obviously does not meet the requirements.

[0035] In this embodiment, the image data is preferably randomly divided into two parts. The separation method is: randomly generate a random matrix A1 with the same dimension as the original image A, and then subtract A1 from A to obtain A2, thereby dividing the original image A into A1 and A1. A2 in two parts. Since A1 is randomly generated, the complete infor...

Embodiment 2

[0051] In this embodiment, the fingerprint image is taken as an example to further elaborate the technical solution in the first embodiment. like Image 6 As shown, the method of fingerprint image processing is:

[0052] Firstly, the fingerprint image data is preprocessed, the image data is converted into a standard format and the images that obviously do not meet the requirements are deleted. Then the image data is separated into two parts, and the content of each part is sent to a model training server separately, and there is no collusion between the two model training servers. After any model training server receives the uploaded data, it uses the sift algorithm for feature extraction, and the feature extraction results are saved to the cloud after semi-homomorphic encryption to construct a K-D tree for classification queries. Substitute the K-D tree into the trained image processing model for feature extraction, compare the extracted features with the existing fingerpri...

Embodiment 3

[0054] This embodiment takes the handwritten signature image as an example to further elaborate the technical solution in the first embodiment. like Figure 7 As shown, the method of handwritten signature image processing is:

[0055] First, the user selects a fully connected neural network as the image processing model, and then separates the handwritten signature image data into two parts, and sends each part to a data processor, and there is no collusion between the two data processors. Any data processor directly trains the image processing model, and after the model training is completed, the handwritten signature image is converted into a vector input and encrypted. Bring the encrypted handwritten signature image data into the trained model to obtain the encrypted result of the operation. Decrypt the returned encrypted result to obtain the image classification result.

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention belongs to the technical field of image processing, and relates to an image processing method and system based on privacy protection, a medium and equipment; the method comprises the following steps: S1, randomly dividing obtained image data into a plurality of parts, storing each part of data into an independent server, and enabling all the servers not to collude; s2, performing, by each server independently, feature extraction on the divided image data; s3, converting the image processing model into a 2PC training model, and training the 2PC training model; s4, substituting the features extracted in the step S2 into the model obtained in the step S3, generating a model training result, and decrypting the training result to obtain a final image processing result. According to the method, data are encrypted and stored in a plurality of non-collusive servers, so that the purposes of dispersing risks and tolerating intrusion are achieved, interaction among the servers is carried out through secure multi-party calculation, and feature extraction, model training and image classification are still safely executed in the absence of a trusted third party.

Description

technical field [0001] The invention relates to an image processing method, system, medium and equipment based on privacy protection, and belongs to the technical field of data processing, especially the technical field of image processing. Background technique [0002] With the continuous development of information technology, image processing technologies represented by handwritten signature recognition, face recognition and fingerprint recognition continue to evolve, and have been applied in many fields such as facial recognition payment and fingerprint unlocking. [0003] Usually in the image processing process, the data owner first sends the image data to the data processor, the data processor performs feature extraction and model training, and then returns the model to the data owner, so the data owner is also the model owner. Model users can apply to use the data owner's model for image classification. However, since the model user needs to upload the image data and ...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Applications(China)
IPC IPC(8): G06F21/60G06F21/62G06K9/46G06K9/62
CPCG06F21/602G06F21/6245G06V10/44G06V10/462G06F18/241
Inventor 秦波王李笑阳赵正朋冯宁轩陈政连迪迪
Owner RENMIN UNIVERSITY OF CHINA