Tag: image watermarking

Article

Qualcomm will work towards authenticating photos taken by smartphones or other devices using its ARM silicon at the point of capture

One of the strongest ways to fight misinformation will soon be right in your phone | FastCompany

My Comments

The rise of deepfaked and doctored imagery surfacing on the Web and being used to corroborate lies has started an arms race to  verify the authenticity of audio and visual asset

It was encouraged by the Trusted News Initiative which is a group of leading newsrooms who want to set standards regarding the authenticity of news imagery and introduce watermarks for this purpose.

TruePic, an image authentication service, are partnering with Qualcomm to develop hardware-based authentication of images as they are being taken. Qualcomm has become the first manufacturer of choice because of themselves being involved with ARM-based silicon for most Android smartphones and the Windows 10 ARM platform.

This will use actual time and date, data gained from various device sensors and the image itself as it is taken to attach a certificate of authenticity to that image or video footage. This will be used to guarantee the authenticity of the photos or vision before they leave the user’s phone.

TruePic primarily implements this technology in industries like banking, insurance, warranty provision and law enforcement to work against fraudulent images being used to corroborate claims or to where imagery has to be of high forensic standards. But at the moment, Truepic implements this technology as an additional app that users have to install.

The partnership with Qualcomm is to integrate the functionality in to the smartphone’s camera firmware so that the software becomes more tamper-evident and this kind of authentication applies to all images captured by that sensor at the user’s discretion.

The fact that TruePic is partnering with Qualcomm at the moment is because most of the amateur photos are being taken with smartphones which use this kind of silicon. Once they have worked with Qualcomm, other camera chipmakers including Apple would need to collaborate with them to build in authenticated image technology in to their camera technology.

It can then appeal to implementation within standalone camera devices like traditional digital cameras, videosurveillance equipment, dashcams and the like. For example, it can be easier to verify media footage shot on pro gear as being authentic or to have videosurveillance footage being offered as evidence verified as being forensically accurate. But in these cases, there may be calls for the devices to be able to have access to highly-accurate time and location references for this to work.

The watermark generated by this technology will be intended to be machine-readable and packaged with the image file. This will make it easier for software to show whether the image is authentic or not and such software could be part of the Trusted News Initiative to authenticate amateur, stringer or other imagery or footage that comes in to a newsroom’s workflow. Or it could be used by eBay, Facebook or Tinder to indicate whether images or vision are a genuine representation of the goods for sale or the p

But this technology needs to also apply to images captured by dedicated digital cameras like this Canon PowerShot G1 X

rofile holder.

The idea of providing this function would be to offer it as an opt-in manner, typically as a shooting “mode” within a camera application. This allows the photographer to preserve their privacy. But the use of authenticated photos won’t allow users to digitally adjust their original photos to make them look better. This same situation may also apply to the use of digital zoom which effectively crops photos and videos at the time they are taken.

There is the idea of implementing distributed-ledger technology to track edits made to a photo. This can be used to track at what point the photo was edited and what kind of editing work took place. This kind of ledger technology could also apply to copies of that photo, which will be of importance where people save a copy of the image when they save any edits. This will also apply where a derivative work is created from the source file like a still image or a short clip is obtained from a longer file of existing footage.

A question that will then come about is how the time of day is recorded in these certificates, including the currently-effective time zone and whether the time is obtained from a highly-accurate reference. Such factors may put in to doubt the forensic accuracy of these certificates as far as when the photo or footage was actually taken.

For most of us, it could come in to its own when combatting deepfake and doctored images used to destabilise society. Those of us who use online dating or social-network platforms may use this to verify the authenticity of a person who is on that platform, thus working against catfishing. Similarly, the use of image authentication at the point of capture may come in to its own when we supply images or video to the media or to corroborate transactions.

Articles

The news services that appear on the “screen of respect” that is main TV screen like the ABC are often seen as being “of respect” and all the screen text is part of their identity

TNI steps up fight against disinformation  | Advanced Television

News outlets will digitally watermark content to limit misinformation | Engadget

News Organizations Will Start Using Digital Watermarks To Combat Fake News |Ubergizmo

My Comments

The Trusted New Initiative are a recently formed group of global news and tech organisations, mostly household names in these fields, who are working together to stop the spread of disinformation where it poses a risk of real-world harm. It also includes flagging misinformation that undermines trust the the TNI’s partner news providers like the BBC. Here, the online platforms can review the content that comes in, perhaps red-flagging questionable content, and newsrooms avoid blindly republishing it.

.. as well as their online presence – they will benefit from having their imagery authenticated by a TNI watermark

One of their efforts is to agree on and establish an early-warning system to combat the spread of fake news and disinformation. It is being established in the months leading up to the polling day for the US Presidential Election 2020 and is flagging disinformation were there is an immediate threat to life or election integrity.

It is based on efforts to tackle disinformation associated with the 2019 UK general election, the Taiwan 2020 general election, and the COVID-19 coronavirus plague.

Another tactic is Project Origin, which this article is primarily about.

An issue often associated with fake news and disinformation is the use of imagery and graphics to make the news look credible and from a trusted source.

Typically this involves altered or synthesised images and vision that is overlaid with the logos and other trade dress associated with BBC, CNN or another newsroom of respect. This conveys to people who view this online or on TV that the news is for real and is from a respected source.

Project Origin is about creating a watermark for imagery and vision that comes from a particular authentic content creator. This will degrade whenever the content is manipulated. It will be based around open standards overseen by TNI that relate to authenticating visual content thus avoiding the need to reinvent the wheel when it comes to developing any software for this to work.

One question I would have is whether it is only readable by computer equipment or if there is a human-visible element like the so-called logo “bug” that appears in the corner of video content you see on TV. If this is machine-readable only, will there be the ability for a news publisher or broadcaster to overlay a graphic or message that states the authenticity at the point of publication. Similarly, would a Web browser or native client for an online service have extra logic to indicate the authenticity of an image or video footage.

I would also like to see the ability to indicate the date of the actual image or footage being part of the watermark. This is because some fake news tends to be corroborated with older lookalike imagery like crowd footage from a similar but prior event to convince the viewer. Some of us may also look at the idea of embedding the actual or approximate location of the image or footage in the watermark.

There is also the issue of newsrooms importing images and footage from other sources whose equipment they don’t have control over. For example, an increasing amount of amateur and videosurveillance imagery is used in the news usually because the amateur photographer or the videosurveillance setup has the “first images” of the news event. Then there is reliance on stock-image libraries and image archives for extra or historical footage; along with newsrooms and news / PR agencies sharing imagery with each other. Let’s not forget media companies who engage “stringers” (freelance photographers and videographers) who supply images and vision taken with their own equipment.

The question with all this, especially with amateur / videosurveillance / stringer footage taken with equipment that media organisations don’t have control over is how such imagery can be authenticated by a newsroom. This is more so where the image just came off a source like someone’s smartphone or the DVR equipment within a premises’ security room. There is also the factor that one source could tender the same imagery to multiple media outlets, whether through a media-relations team or simply offering it around.

At least Project Origin will be useful as a method to allow the audience to know the authenticity and provenance of imagery that is purported to corroborate a newsworthy event.