Malicious PNG Image Recognition Method Based on Machine Learning
In order to prevent network virus from spreading malicious code by means of Portable Network Graphics (PNG), this paper uses machine learning technology to propose a set of effective malicious PNG image recognition methods. This recognition method uses dynamic analysis to evaluate PNG images and loader behavior, and combines machine learning algorithms based on traditional call features to classify and identify malicious PNG images. The results show that the malicious PNG image recognition method proposed in this paper has high reliability and feasibility, can ensure the recognition effect of a variety of malicious PNG images, and can avoid the user steganization of healthy image information, which fully meets the application requirements of image watermarking covert transmission.
万方数据
维普
