Perceptual Hashing on Image-based Malware Detection

Abstract

Information delivery through communications in the digital age is aided by visuals. Malicious actors also can utilize images to convey malevolent communication. This study attempts to assess the viability of detecting image-based malware by utilizing perceptual hashes: average hashing, difference hashing, perceptive hashing, and wavelet hashing. The proposed method allows the determination of the aforementioned perceptual hashes’ effectiveness against malware embedded in the images. The approach used to achieve this study objective is methodological experiments through comparisons of 26 images and 6 malware variants against selected perceptual hashing. The proposed method involves adding malware in pictures before and after the EOF bytes. Our research shows that perceptual hashing methods are vulnerable to malware in images, urging the development of new, more effective methods for malware detection.