A Malicious PNG Image Recognition Method Based on Machine Learning

Abstract

The invention discloses a machine learning-based malicious PNG image recognition method, and belongs to the technical field of network space safety. The method comprises the following steps of: firstly establishing a PNG image feature library and a digital steganography recognition model; investigating a picture file uploading request at a server, carrying out feature matching recognition according to the PNG image feature library, preliminarily recognizing whether a PNG picture is legal or not, if the PNG picture is legal, calling the digital steganography recognition model to mine whether the PNG picture has information hiding or not, and if the PNG picture is illegal and has information hiding, rejecting uploading; and monitoring PNG picture format file data in a webpage transmission process at a client, carrying out feature matching recognition according to the PNG image feature library, if the PNG picture is legal, calling the digital steganography recognition model to mine whether the PNG picture has information hiding, and if the PNG picture is illegal or has information hiding, forbidding accessing the picture resource. The method is capable of forbidding uploading of illegal pictures at the server and forbidding access of illegal pictures at the client, so that the network safety is strengthened.

Description

technical field [0001] The invention belongs to the technical field of cyberspace security, and in particular relates to a machine learning-based malicious PNG image recognition method. Background technique [0002] With the rapid popularization and application of the Internet and the rapid development of digital technology, the issue of cyberspace security has gradually entered people's field of vision and has attracted more and more attention. [0003] On the one hand, browsers are the main medium for people to obtain Internet information, and their security issues cannot be underestimated. In recent years, more and more webpages have been implanted with all kinds of webpage advertisements due to the slack review of JavaScript and other reasons. At least they induce users to click on malicious links, and at worst, they pass malicious software, malicious dynamic link library files (DynamicLink Libraries, DLL) are attached to web page images, bypassing computer and network ...

AI Technical Summary

Problems solved by technology

Since every website relies on various multimedia resources, such as audio, video, and images, attackers can use digital steganography to hide malicious software and malicious attack payloads in multimedia, and can easily bypass Anti-malware detection, which poses a greater potential threat

Images