Detecting Fake Photos
Don't worry, it's not real
(Image used under: Fair Use doctrine)
MIT is working on software that analyzes the surfaces in photos to see if the light patterns all match in order to detect fakes.
Many fraudulent images are created by combining parts of two or more photographs into a single image. When the parts are combined, the combination can sometimes be spotted by variations in the lighting conditions within the image. An observant person might notice such variations, Johnson says; however, "people are pretty insensitive to lighting."Another technique is to check the grain of the photo for irregularities. When an object is placed within a photo, the new object will have differing grain from the original (though there are ways to reduce and eliminate this).
If you're wondering why this matters, check out these examples of how fake photos can be drastically harmful:
Share This
|
|
|
|
|
|
|
||
