Images can be misled by adding, removing, or changing objects in the scene. With a few simple clicks, many photo editing programs include tools to remove objects from their surrounding. This technique is called inpainting.
One early method consisted of filling the gap left by the missing image using a smooth interpolation based on the pixels around the edge. Other techniques include sewing carving and content-aware resizing, where an algorithm identifies the important areas of the image in order to remove sections or expand them around them without affecting its subject.
Copy the area (or image) that you want to fill in the gap. It is also possible to use this technique to copy objects, such as Iranian missiles or North Korean hovercrafts. However, the edges may need to blend into the background in order to look convincing.
The content, and thus the interpretation of an image, can be drastically altered. It is harder than you think to create convincing images. The direction and intensity of the lighting must match the rest of the image and the area being altered.
Unmasking forgeries
Digital image forensics is a science that identifies manipulated areas in images and connects them to the devices or cameras that produced them.
There are generally two types of forgeries: those that can be detected by the inconsistencies within an image’s composition and those that have been altered.
Even the untrained will notice a poorly constructed composite photograph. It will have gross inconsistencies with lighting and perspective. To detect more sophisticated forgeries, an analysis may be needed to determine the geometry of reflections and shadows. Researchers at Dartmouth College in the US have developed a new method that involves superimposing lines onto an image to connect objects with their shadows to indicate the location of a source of light within the scene. The shading of objects inserted in the scene will likely be inconsistent with the expected shade, given the location of the light source.
Photographs can show the effect of lateral chromatic Aberration when white light is passed through a lens. The intensity of chromatic distortion depends on both the lens properties and the distance between the focal point of the lens as well as the objects in the photograph. Any elements added to the composition from a picture taken with a different lens will have a detectable difference in chromatic distortion.
A composite image of foreground objects and background objects. Stuart Gibson is the Author.
Our research team developed software that can detect differences in noise levels to identify the edited area. Stuart Gibson, Author, provides
The digital imaging process causes artifacts and can be seen in all photographs. They are usually invisible to the naked eye, but they play a vital role in digital image forensics.
A color image is created, for example, by applying a filter that alternates red, green, and blue to the pixels on a camera sensor. Each pixel absorbs one color. Demosaicing renders the information into a full-color image but leaves behind a regular pattern. This pattern is disrupted by tampering.
