Method and system for identifying fraudulent websites based on image instance-level features

Abstract

The invention discloses a method and system for identifying fraudulent websites based on image instance-level features, which belong to the technical field of image processing. Aiming at the existing technical solutions, the identification mode for fraudulent websites is single, the identification effect is not high, and false positives are prone to occur. reported situation. The present invention extracts the global feature vector of the picture by performing a global feature model on the screenshot of the homepage of the website that needs to be identified, then searches the extracted global feature vector of the picture in the feature vector database, and calculates the global feature vector of the picture and the feature vector database. The Euclidean distance of the global feature vectors of all pictures, and returns the global feature vectors of the first N pictures with the closest distance and the type of fraud involved, and finally calculates the maximum number of feature points of instance objects similar to the N pictures respectively, and obtains the type of website to be identified .

Description

technical field [0001] The invention belongs to the technical field of image processing, and in particular relates to a fraudulent website identification method and system based on image instance-level features. Background technique [0002] With the development of Internet technology, the transfer of traditional illegal crimes to non-contact crimes using telecommunications and the Internet as the medium is accelerating, and various new types of websites suspected of cybercrime emerge in endlessly. New types of cyber crimes are being implemented in various links through new technologies such as artificial intelligence, machine learning, and big data, forming an intricate "black and gray industrial chain" and a combination of criminal interests, seriously damaging the legitimate rights and interests of the people and social security and stability. [0003] The existing fraudulent website identification methods mainly use content-based matching technology to identify fraudulen...

AI Technical Summary

Problems solved by technology

[0005] 1. Identify fraudulent websites by matching text keywords and classifying pictures. The pattern is relatively simple, the recognition effect is not high, and false positives are prone to occur.

[0006] 2. In the current situation where all kinds of new websites suspected of cybercrime emerge in endlessly, there are various fraudulent websites of the same type, and the existing methods cannot effectively identify the same type of fraudulent websites

[0007] 3. In the face of the increase of fraudulent websites, the existing method adopts re-data labeling and model training of the model, which is not efficient and does not meet the current situation of rapid growth of fraudulent websites that requires real-time identification

Images