PhotoN
What is it?
PhotoN is a professional digital forensics software, specially designed for corporate clients. It processes any graphical data in JPEG format in order to reveal signs of digital modifications.
Who needs it?
All companies that interested in levelling up their digital security. This product mainly interesting for banks, insurance companies, financial organisations and all other companies who actively work with digital documents.
Why is it unique?
PhotoN analyses the inner structure of the image itself, which gives reliable fraud detection. PhotoN algorithms can reveal even most professional digital modifications of the image because it uses raw mathematics, but not unstable recognition patterns.
How does it work?
PhotoN analyses inner traces of possible image modifications. The analysis includes two levels and consists of 5 different tools. Here are these two levels of analysis:
- Metadata analysis. All photos have so called metadata – information about camera, resolution, diaphragm, date/time etc. This information erases during editing and it’s really easy to check.
- Advanced filter analysis. It is based on deep digital analysis of the image – for example searching of places with different compression levels (ELA) or searching for similar places of image code. This level of analysis allows to reveal fraud even if the image metadata was changed in HEX-editor.
Metadata analysis
First, most simple and informative method of analysis that we use is Metadata Analysis (MA). In is based if EXIF image standard which states that any .jpeg photo must have encoded information about:
- Image dimensions
- Camera model
- Graphical editor
- GPS coordinates
When any attempt of image editing is made, all these data are modified by graphical editor – it usually deletes all information about dimensions and GPS and always replaces “Camera model” with name of graphical editor. So, if image was modified, we can reliably detect it even at this, most simple level of analysis.
Advanced analysis
Unfortunately, there are advanced techniques for avoiding fraud detection by metadata analysis. In order to detect even these advanced techniques we developed four king of image analysis that deal with inner structure of the image, which allow to detect fraud even if metadata were replaced after the mage modification. These filters are:
- ELA
- CPM
- DCT
- Noise
