Malviso

This project aims to leaverage Image Visualization techniques for malware detection in MS Office files.

The selected dataset is available at this link in Zenodo and is taken from this paper from Computers & Security 2021.

State of the art

We took inspiration from the structure of a similar project.

Here the state of our implementation:

• PREPROCESSING
  • Byte plot
  • Markov plot
• FEATURE EXTRACTION
  • Texture features
    • Gabor filter
    • LBP - Local Binary Patterns
    • Local Entropy
  • Keypoint Descriptors
    • SIFT - Scale Invariant Feature Transform (slower)
    • ORB - Oriented FAST and Rotated BRIEF (faster)
• CLASSIFICATION
  • RF - Random Forest
  • DT - Decision Tree
  • KNN - K-nearest neighbors
• RESULTS COMPARISON
  • Popular antiviruses
  • PDF image visualization based malware detection
  • ...