460 likes | 653 Views
Forensic Imaging The History of Image Forgery Image Splicing. Yaniv Lefel Hagay Pollak . Cloning …. History. A Picture is worth a thousand words…. More. =. +. Is this real ? . Caribbean 2005. Virginia 2004. Look at the shadows … . Watch the lightning … . Fake or real ? . link.
E N D
Forensic ImagingThe History of Image ForgeryImage Splicing Yaniv Lefel Hagay Pollak
A Picture is worth a thousand words… More... = +
Is this real ? Caribbean 2005 Virginia 2004 Look at the shadows … Watch the lightning …
Fake or real ? link
Why? • Fake images are everywhere due to • Popularity of digital cameras. • Availability of desktop imaging software allows easy manipulation of images.
Image authenticity • A fake image can be defined as an image of an object or scene that wasn't captured as the image would imply. • The fake images that concern us most are those advertised as real.
Understanding faking • Imaging properties are far too complex for manually creating one - one pixel at a time. • Complex calculations are required in order to take into account the image physical properties. • High quality software isn’t accessible to the average PC user.
Understanding faking (cont’) • Common method of faking is by editing an existing image that was captured by a camera.
Origin of editing pictures • 19th century - remove wrinkles and blemishes. • Dark room tricks - adding and removing people from images (being unable to get an entire family together for a family portrait).
Change of context – Example 1 • The Surgeon's Photo, 1934, reportedly showing the Loch Ness Monster
Change of context – Example 2 • Cottingley Hoax, 1917, reportedly showing winged fairies
Embedding an image in another • All you need is a PC with image editing software. • The software allows the creator to modify the image to the appropriate size and rotation.
Detecting fake images using common sense • Our perception is the first line of defense at identifying fake images. • Example (man holding a cat): • The cat is obviously too big. • The man should have leaned backwards more to properly hold a cat of this weight.
Detecting fake images using common sense (cont’) • Our perception can fail to detect a fake image if there is no cause for suspicion. • TV Guide used Ann-Margret's bodyfor a picture of Oprah Winfrey.
Realistic computer generated images • Fake: Columbia disaster taken from a satellite • Real: A computer generated image from the movie Armageddon.
Detecting fake images An image has an encoded watermark that contains the edge information of original image, revealing an alteration that has been made to the original image.
The gray-level histogram may show signs that the image has been altered.
Inconsistent noise properties may be apparent in altered images.
Measuring the vanishing points reveals that a window has been added to this building. Perspectives lines converge to a single point.
Splicing and Blending • Composing multiple images into a single image. • Input: • Multiple images. • Output • Single composed (e.g. Panoramic) image.
Splicing and Blending (cont) • Splicing <-> composing • At the end all the images are composed into a single image so that the final image appears to show no traces of the composition. • The Challenge: • While taking the pictures (images) the scene might change – e.g: People moving around, cloud shadows move. • The images are not always identical in the areas the images overlap.
Splicing and Blending (cont) • Idea: • The technique extracts out parts of each of the individual images to construct the composed image (panorama). • The final result is not really a true wide angle snapshot of the busy scene, but it looks like it could have been.
Splicing and BlendingImage feathering • Idea: • Combine two images into one, by averaging the color values from the two images. • Problem: • When the scenes in the two images are different the result may contain an effect called “Ghosting” where an object appears blurred in the result image.
Splicing and BlendingComposing using “snakes” • Idea: • This splice technique computes a curved line from top to bottom in the common region , and assembles the composite by taking the part of the first image to the left of the line and places it adjacent to the part of the second image to the right of the line.
Splicing and BlendingComposing using “snakes” http://torina.fe.uni-lj.si/~tomo/ac/Snakes.cgi http://www.ecs.soton.ac.uk/~msn/book/new_demo/Snakes/ http://www.markschulze.net/snakes/index.html
Snakes • Define a set of points as a curved line – the snake. These points move to the lowest “energy” point in the local neighborhood, defined by the "Energy Function". This continues until the snakes stop moving. • This establish the problem as the minimization of some cost function. • Use established optimization techniques to find the optimal (minimum cost) solution.
Splicing and BlendingCreating tileable image texture tiles • The same technique can be used on a single image, to create a synthetically generated tiled larger image. • By cutting the single image in the middle, splicing the left and right parts (see next slide) into a new tile. • Repeat the process twice vertically and horizontally. • Then tiling the new generated tile.
What now ? Law and Order (proving authenticity) Journalism Scientific publications
What about this ? link
Well ? CG CG Real CG Real CG Real CG Real Real link