Image security identification system and method in network environment

Abstract

The invention provides an image security identification system and method in a network environment, and the method comprises the following steps: 1, enabling a front end to collect a picture, and enabling the front end to comprise an image identification module, an image security identification module and an image security authentication module; 2, the front end performs feature extraction and binaryzation on the acquired picture; step 3, carrying out paillier cryptographic algorithm encryption on the front-end feature data; 4, binding a camera to the DOM element; and step 5, front-end and back-end interaction: the front end submits the data to the background, then the background processes the data and returns a result to the front end, and the front end receives the data sent back by thebackground, analyzes the data and renders a front-end page according to the data of the background. According to the method, the image recognition technology is reasonably and correctly applied on thepremise of safety to a certain extent; privacy information between participants is protected from being leaked to each other in the whole process, third-party lawbreakers are prevented from attackingthe image recognition model by using countermeasure samples to achieve the illegal purpose of the third-party lawbreakers, and the method can be used for safety recognition and classification of electronic files.

Description

technical field [0001] The invention relates to an image security recognition system and method in a network environment, belonging to the field of visual security. Background technique [0002] The research on image recognition technology can be originated in the 1950s, but in the early years, because the accuracy rate was very low in the actual environment, it has not been practical. In recent years, with the rapid development of big data and deep neural network technology, deep neural network has achieved good results in various computer vision classification tasks. Deep neural network technology has begun to be applied to image recognition tasks and has achieved good results. The effect of image recognition has reached more than 99% on many data sets. It has applications in public security cases, enterprise attendance, property security and so on. [0003] With the emergence of cloud services, image recognition technology combined with cloud computing makes it more pra...

AI Technical Summary

Problems solved by technology

The impact of adversarial samples on major deep neural network models is huge, making deep neural network models vulnerable to third-party attacks using adversarial samples in practical applications, resulting in wrong results, reduced recognition security, and immeasurable losses

Images