We all know of the saying “A picture is worth a thousand words.” Now with modern digital photography, it could easily read “A picture is worth a thousand lies.”
What happens in the law enforcement world when a person can spend a couple of hundred dollars to buy a computer program allowing them to manipulate digital photographs for criminal purposes, or to create digital forgeries.
Photo and digital manipulation is difficult to detect, but unless the suspect is an expert in the field, forensic examiners will likely detect the digital manipulation.
Lucky for law enforcement, most who manipulate an image for criminal means leave enough traces to be caught.
In most cases, forgers use cut and paste methods to manipulate letters, numbers, or words to change the document. An individual will scan complete documents into their computers, and with programs like Photoshop, they manipulate the text and numbers.
Other scenarios would include accessing an authentic signature then scanning it in digitally, so it can be used on checks, wills, etc. To the unaided eye, the signature appears to be an original.
When looking at manipulation of documents, pixel color in the manipulated documents can be checked to locate differences, but visually, it’s hard to detect slight color changes. Other clues include: irregular spacing between letters, font differences in inserted words and letters, discrepancies in the size of inserted letters or words, and the crowding of letters and words. Lastly, the background surfaces of the manipulated documents will almost always show to be disturbed.
The forged document may also possess different noise level compared to that of its surroundings. The forged area will contain original and foreign pixels. Each of the original pixels in the area will be similar, while foreign pixels will show significantly different characteristics.
Let’s now look at doctoring photos. Photo altering is characterized as adding, changing, or deleting features from an image. There are three primary techniques used to forge an image; copy-move, image splicing, and image resampling.
Copy-Move – Images are copied from the original image, and paced in another place on the photo. Since the copy is from the original, its essential properties match, and that makes it hard to recognize when it has been altered.
Image Splicing – Image splicing uses cut-and-paste technology to take images from the original or from a different image, to create a fake entry on a photo.
Image Resampling – Essentially, you keep the image’s pixel dimensions but change the size of the image, on the original photo.
How are photo manipulations detected?
According to Adobe, the Photoshop developer; “Every time an image is manipulated, it leaves behind clues that can be studied to understand how it was altered”.
Let’s start looking at ways to detect manipulation by talking about digital compression. Basically, a suspect compresses the image when manipulation occurs, resulting in fewer bits being used than in the original image. Experts can use programs that show different levels of compression in a photograph. The unedited version would all be at the same basic compression, any part of it that has a different compression level indicates the image could be manipulated.
All manipulation techniques leave artifacts, such as strong contrast edges, deliberately smoothed areas, or different noise patterns. A competent user will leave artifacts, although they won’t be that visible to the unaided eye. Looking at the subject photo at the pixel level, or by applying special filters will allow many of the modifications to be seen.
Metadata is another way to detect manipulation. Metadata is information stored in the digital file of a photo, and typically includes data such as the date, time, camera model, geolocation, etc. In some cases, the metadata might also have information about the software that is used to edit or manipulate the image. Good stuff, except if a person manipulating the image has much skill, they can easily modify and hide relevant data.
Shadows are a big clue in photographs, so since many photos have scenery associated with them, experts look at the shadows and mismatched light patterns. Manipulating shadows is one of the hardest things to do, even for a Photoshop expert. Often, incorrect angles and incorrect shadow lengths are produced. Actual shadows become lighter and blurrier as they get further from the object, and that’s a very tricky task to get right for most users.
Image or pictures faked using Photoshop are often considerably larger in size to what the size would have been because the program works with layers. Added layers can be detected by an expert investigator.
There are also forged videos, goods ones are called “deep fakes”. Imagine cell phone video showing that you, an innocent person, committed a certain crime, and there it is, a video of you committing the act. It’s not far fetched; a deepfake video is made by modifying images or videos using an AI technique called a “generative adversarial network” (GAN). The software detects the way a person moves, by learning their movement from a source video, then it duplicates those movements, and creates the manipulated video. There are software programs that can identify the manipulated video, but many feel like since AI is involved, it will be an uphill battle to keep up with the forgeries.
In conclusion, if the suspect uses programs that are non AI, and they are not experienced, he or she will leave enough of a trail to be detected fairly easily. If the suspect is an expert, then the challenge to prove the fraud will be greatly enhanced, but well within a competent forensic experts abilities.