Video Redaction & Presentation Techniques
Written by Stuart Boutell   

Until only a few years ago, redaction was only known as the process of obscuring or censoring part of a printed text document for legal or security reasons—for example, the release of sensitive documents by government agencies with parts redacted to protect individuals, or sensitive information. The same legal and security reasons for applying obfuscation now also exist with still images, and video from body-worn cameras, citizen smartphones, or CCTV. Therefore, the need to have the right tools to deliver this redaction in a timely manner has become critical to the successful adoption of new digital technologies within law enforcement and criminal justice in order to deliver digital evidence.

Unlike a text document, redaction of video, in particular, presents significant challenges to organizations: the redaction must be complete, accurate, and legitimate, while not obscuring the relevant remaining content, because in the same way a “broad-brush” redaction of text is unacceptable, taking a crude approach to video redaction is similarly likely to be unacceptable.

Also, unlike a text document, video consumes significant storage resources, and every copy of a video made with redaction applied consumes extra precious (and potentially costly) storage resources. Plus, every copy extracted from a video management platform in order to have redaction or edits applied is a data protection risk, both from loss and uncontrolled sharing or uploading, as well as a potential point-of-challenge in law. Is the redacted copy a faithful representation of the original, and can the audit-trail be proven from the source to court, even with the redaction applied?

In addition, unlike text documents, it is sometimes the case that the redaction of a video might need to remove the majority of the content (i.e. just leaving the relevant person or persons of interest, while protecting or obscuring the identity or location of the rest of the shot).

Until very recently, the combination of low-frequency need, combined with extensive training required in the relevant tools, had meant that evidential redactions of video and stills were the domain of specialist image or forensic processing units. The widespread introduction of video evidence, and of body-worn cameras in particular, challenges this. Every officer has the potential to generate evidential footage every day, along with a pressing requirement to apply redactions, clipping, and annotations while ensuring the resultant video is still relevant, legal, and evidential. This places an increasing burden on these specialists. It also has potential to introduce undesired delays to the legal process. These delays are especially important where redaction might be used as part of an ongoing investigation—for example, to make a rapid public release of video to garner information in response to an Amber alert.

Finally, the redacted evidential video is of no use if it can’t be presented easily in court. The output of the redaction process needs to be straightforward to share in a format that is compatible with court processes and procedures, as well as prosecution and defense legal teams. The presentation in court should also follow a cognitive model that is recognizable and familiar to the legal process—so handling of the digital evidence in a way analogous to the way that physical evidence is presented might be the most effective.

In summary, the requirements are:

  • Apply clipping, redaction and annotation to evidential or sensitive video in a straightforward way
  • Use tools that can be operated by non-specialist image processing units, without special software or special tools
  • Don’t invalidate the evidential chain-of-custody
  • Don’t make unnecessary copies that expose organizations to risk of loss, and storage costs
  • Have a solid audit trail for each change made to evidence
  • Do all this in a timely fashion
  • Make the redacted evidence available to the relevant actors in law-enforcement, criminal justice and the community

Redaction: Modern tools

Modern digital evidence management suites should include web-based tools to deliver easy-to-use clipping, redaction, and annotation workflows that are evidence-grade and straightforward to use. It’s important to consider browser-based tools first and foremost, because they have the lowest IT footprint and impact on your systems, and provide greatest security for your footage. By performing the manipulation within the browser, you present the lowest footprint for data loss should your laptop or workstation be lost or stolen.

Evidence processing is all about metadata. Therefore, a significant advantage is to be gained where your evidence management tools implement redactions, clipping, and annotation as metadata and do not generate copies of your original video, but instead apply this “metadata” at the point of playback (on the server before presenting footage in the browser) or (as a last resort) when exported to DVD or MP4 file for uncontrolled sharing or presentation. Metadata-led tools mean that it’s also easy to edit and modify a redaction later if it is deemed to be in error, and they also allow easier integration with other case-management or case-preparation tools.

Like a text redaction, it’s important that the recipient cannot easily un-apply the redaction, to reveal the original—both from a technical point of view of applying the redaction server-side rather than browser-side (so the original footage is never transmitted to the viewer and cannot be recovered) and from an algorithmic point of view that the redaction provides sufficient obscuration of the original. If in doubt, use solid-color to redact.

Clipping

The best redaction of all is to not share a video at all—at least, don’t share anything that is not relevant or required. To that end, the most powerful redaction is clipping—the process of extracting a clip from the longer recording. Again, a clip should be implemented as metadata within the digital evidence management store, and not generated as a copy. The second advantage of making a clip is that it also reduces the amount of video that needs to have redaction applied—so, always clip first.

Redaction

Identify background or foreground redaction—Once the smallest clip has been made, identify which background redaction applies best to the use case. Depending on the tool, this might be easiest if applied last (so that you can easily work with the full image up to the point it is ready), but it is important to know up front if you are redacting to highlight part of an image, or hide part of an image.

Select objects to obscure (or highlight)—Starting at the beginning, select each object that requires redaction (see Figure 1), and use the tools provided with your DEM to move the redaction with your footage. Most modern DEM should include a “tweening”, meaning that big motions every few frames convert automatically into small motions on each frame. Advance through the clip a few frames at a time, adjusting the location of the redaction. Select the redaction type (blur, solid etc.) depending on the reason for redaction. If in doubt, use a block-color.


Figure 1—To begin, select each object that requires redaction.

Apply the object redactions, and then the background redaction—Finally, select the object and background redactions depending on the use case. To protect a law enforcement officer or a minor or child, use a blur or solid color with a clear background (see Figure 2). To emphasize a person of interest, blur or darken the background, and use a clear redaction to highlight the individual concerned (see Figure 3).


Figure 2—To protect sensitive subjects (such as children), use a blur or solid color with a clear background


Figure 3—Use a clear redaction to highlight an individual

Consider using zoom—The ability to zoom in on part of a video (e.g. to zoom in on a specific individual) is also a powerful redaction, as it removes the non-relevant sections from the frame—but use of zoom should be sparing, since totally removing content (by clipping out using a zoom) might be viewed as an unacceptable manipulation.

If you think using zoom helps to present the necessary detail, it might be wise to make two redacted versions of footage, and present them sequentially in court—first the non-zoomed version, to provide context, then the zoomed version, to provide the detail.

This can be viewed as analogous to showing bagged physical evidence in court, then showing a fingerprint extract and database match; you set the context of evidence, and then zoom in on the important detail.

Annotation

Lastly, use any built-in tools to apply text or other annotations (e.g. textual metadata such as recording time and date, camera serial number, or highlighting shapes to draw attention to persons of interest). This will help when the redacted footage is presented, to draw the viewer to the significant items or events (see Figure 4).


Figure 4—Use built-in tools to add text or other annotations

Burnt-in signatures or other metadata will also help track down footage should it be used in an unauthorized way.

Presentation

Once you have ingathered your digital evidence, clipped it to length, and applied redactions and annotations, the next step is to prepare it for presentation.

Web-based playback—If your criminal justice partners support it, and you can provide secure authenticated access to your digital evidence manager (see Figure 5), then using web-based sharing is the best option. It has the advantages of continuing to provide playback audit data into the same evidence manager platform, and reduces the risk of the generation of uncontrolled copies—especially if your shared playbacks include on-screen annotations identifying the recipient—so even if a bad actor uses a smartphone or screen capture application, you should be able to use the in-frame watermark to identify them, or at least deter them.


Figure 5—If supported, provide secure authenticated access to your digital evidence manager

The only disadvantage of web-based playback, especially in court, is the requirement to have high-quality, high-reliability networking in the court.

Web-based export—When interacting with agencies or individuals with whom you cannot establish an authentication scheme, the next best option is to provide web-based exports of traditional media (e.g. MP4 files or DVD images). This obviates the need to generate physical media and gives you a “first download” audit trail when the media is accessed. However, once this media has left your platform, you have no further controls available to its dissemination or use. This is the best option when sharing footage with media outlets (e.g. for Amber alerts).

Physical media generation—Practically, this is the way that evidence will continue to be shared with some organizations for some time to come, so your digital evidence management and redaction platform must support a generation of legacy export formats (e.g. DVD images or MP4 files), and you will need to handle them securely in the usual ways to mitigate against physical media loss.

About the Author

This e-mail address is being protected from spam bots, you need JavaScript enabled to view it brings a quarter century’s experience to the challenge of delivering secure, robust, scalable, wearable video products, and their associated digital evidence-management systems. He co-founded Edesix in 2002, having previously worked for 3Com, Cisco, and Tenor networks. Boutell is responsible for the product management of the Edesix VideoBadge body-worn camera range, and the Edesix VideoManager and Edesix AssetManager digital asset management platforms.



 

 
< Prev   Next >






Interview with an Expert

One of the more specialized areas of crime-scene investigation has to do with searching for evidence of arson. To get some background in this area, we spoke with an individual who has had more than 46 years in fire service, 24 of which have focused specifically on fire/arson investigation.

Read more...