Tuesday , 13 November 2018

Cool Izitru service can determine fake photos

izitru-logoIzitru can work out for you if a photo has been adjusted, using its simple method, you will be able to work out if the target photo is a fake or not.

Computers have long made it possible to create highly realistic fake images. Now, thanks to Izitru — a new service launched by photo forensics firm Fourandsix — those very same computers can help us automatically detect those fakes. Izitru builds on the pioneering work that imaging guru Hany Farid has done in detecting phony photos. Unlike previous systems, Izitru is completely automated, relying on a set of six tests to help verify uploaded JPEG images.

Izitru (pronounced “is it true”) builds on the JPEG signature analysis first launched in Fourandsix’s FourMatch plug-in for Photoshop. This test uses the fact that nearly every model of digital camera uses slightly different coefficients for doing the encoding and compression of its JPEG images. Photoshop uses a different, but consistent, set. Using its database of camera models and their coefficients — aka signatures — Izitru can verify what camera is most likely to have created the image.

Of course, now that JPEG signature analysis is a known technology, it can be spoofed with some work, so Izitru runs other tests. One test looks for signs that the JPEG has been saved more than once (perhaps with the second save designed to fake a camera signature). Another is designed to look at the structure of the JPEG encoding to determine if it is likely to have come from camera firmware or from computer software.

Final tests look to see evidence of the camera’s original Bayer sensor pattern, and also whether the image may have been saved at multiple quality settings — called JPEG ghost detection. The ghost detection works by re-saving the image at a variety of quality settings and looking to see which ones are most similar to the suspect image.

See The Original Story click here

Leave a Reply