Hendon Publishing - Article Archive Details

Print Article Rate Comment Reprint Information

Improving Video Forensics with HDTV

Written by John Bartolac

If you’ve ever watched a high-definition television (HDTV), you have seen how the amazing clarity and vivid colors can enhance your appreciation of everything from sporting events to nature shows. Bringing that same level of vivid detail to forensic video could give law enforcement a significant investigative edge.

Unfortunately, the law enforcement community has always been slower to adopt and apply new video technology. This was especially true for the migration from analog-on-tape media and digital video recorders to IP-based video. With improvements in processors designed specifically to handle video from higher resolution HDTV image sensors, law enforcement can now take advantage of the same quality video we are used to seeing as consumers, in areas such as HD movies for home entertainment and high-quality video streaming over the Internet.

HDTV versus Traditional Megapixel

When considering a move to HDTV, it’s important to understand exactly what the technology brings to the table. One of the points of confusion is the difference between HDTV and traditional megapixel technology. Some people use the terms interchangeably when, in fact, they are not.

The Society of Motion Picture and Television Engineers (SMPTE) established a specific set of standards that characterize HDTV technology. These standards deal with resolution, image format and frames per second. Megapixel technology, on the other hand, merely concerns the number of pixels within a field of view and is not governed by any standards. Rather, it is an adaptation of the security industry’s best practices.

Regardless of the number of pixels within a megapixel image sensor, the delivery of that video stream to the viewing monitor cannot be called HDTV unless it meets all appropriate HDTV standards set forth by the SMPTE. In short, HDTV standards apply to all aspects and components of your video system.

If you’ve ever shopped for a high-definition television, you’re probably already familiar with the three primary standards: HDTV 720p, HDTV 1080i and HDTV 1080p. The HDTV 720p defines a resolution of 1280x720 pixels with high color fidelity in a 16:9 format (widescreen). This standard uses progressive scanning of either 25 to 30 frames per second or 50 to 60 frames per second, depending on country.

The HDTV 1080i (interlace) and HDTV 1080p (progressive scan) define a resolution of 1920x1080 pixels with high color fidelity in a 16:9 format. Like in HDTV 720p, these standards scan at a rate of either 25 to 30 frames per second or 50 to 60 frames per second, depending on country. More on the difference between interlace and progressive scan later.

Higher resolution creates huge amounts of image data. To reduce bandwidth consumption and storage requirements, many megapixel camera users choose to transmit at a much lower frame rate. The problem with this decision is that it provides too few frames of the scene to make the video usable for forensic purposes. As a result, the essential frame showing that split-second image from a video clip may not be there when you need it. Also, an essential part of the overall event may have been missed because it was outside the field of view as it relates to the image format.

In contrast, a network camera that complies with HDTV standards not only guarantees a certain resolution and widescreen format, but also a specific frame rate consistent with published international standards. This is extremely important in video forensics because it ensures a more accurate representation of the scene being captured, including any movement of individuals, objects and backgrounds. With full frame rates, you can more precisely investigate and evaluate an incident that encompasses fast moving objects.

By using SMPTE, you now have the ability to support the tools of your investigation that have the backing of an international standard. But, if you don’t need the full resolution and full frame rate associated with SMPTE standards for HDTV, rest assured that HDTV cameras can operate equally well at lower frame rates and at lower resolution. When your investigation does require a higher frame rate and resolution, an HDTV camera has the flexibility to deliver those options.

Rising Data Storage Costs

Because HDTV and megapixel solutions both run at higher bit rates, which increase bandwidth consumption and storage requirements, what are the best cost options? Historically, solutions using megapixel technology have forced users to choose between lowering the recorded video quality or lowering the frame rate because of the amount of data generated by higher resolution image sensors. Another option would be to reduce the retention time for video clips so that you could recycle storage. But given the length of time that often stretches between investigation and court proceedings, this probably isn’t a viable choice.

A better option is to employ advanced H.264 compression techniques for video storage. H.264 is an open, licensed standard that dramatically reduces the size of a digital video file without compromising frame rate and image quality, by more than 80 percent compared with the Motion JPEG format, and by as much as 50 percent compared with MPEG-4 Part 2. The resulting economy of bandwidth and storage allows you to affordably incorporate HDTV-quality forensic video into your policies and processes.

Better Resolution and Formatting

How many times have you viewed video footage where you couldn’t clearly identify the individual, determine the exact color of a subject’s clothing or vehicle, or distinguish an object someone was carrying? In an investigation, not only is time of the essence, but accuracy in detail is essential. HDTV video, from a forensic perspective, gives you the greatest opportunity to accurately identify and prosecute without giving the defense or media the opportunity to exploit gaps in the accuracy of the materials and tools you use to conduct your investigation.

The enhanced clarity of HDTV video can aid law enforcement in quickly pinpointing details that can ensure swift identification of individuals and actions within a scene. This clarity, combined with the HDTV standard 16:9 viewing format, is especially critical in monitoring large crowds where others in the scene may be crucial to your investigation, such as accomplices to a crime, or where secondary incidents are occurring simultaneously within the same field of view.

HDTV Frame Rate Standards

Subjects of an investigation are rarely at a standstill. They are moving away in cars, running away from the scene, shooting, stabbing, stealing and dropping evidence—all in a very quick manner. That is why identifying events in motion is essential for video forensics. While frame rate is an overarching component, let’s first be sure we understand the difference between two underlying scanning options: interlaced and progressive scan.

Interlaced video scanning was originally introduced as a way to improve the image quality of a video signal without consuming additional bandwidth. Interlacing splits each frame into two fields. The scanning starts at the top-left corner and sweeps all the way to the bottom-right corner, skipping every other row along the way to cut bandwidth in half.

The downside of interlaced video occurs when objects or subjects are moving so fast that they appear in different positions when each individual field is captured. This may cause something called “motion artifacts” or ghost images when viewed in slow motion, which is something obviously not good in forensic investigation. Another potential problem with interlaced video is called “interline twitter.” It is an effect that shows up when the subject in the scene contains very fine vertical details that approach the horizontal resolution of the video format.

Progressive scan video avoids the limitations of interlaced video by capturing, transmitting and displaying all lines in the image in a single frame. Scanning is done line by line, from top to bottom, instead of split into alternate fields so there is virtually no “flickering” effect. In a surveillance application, this can be critical for viewing details within a moving image—such as a person running, a vehicle in motion, a slight of hand or a stabbing/fighting incident. In fact, a single progress scan frame can be reproduced in almost photographic quality, which can serve as a crucial piece of evidence in a court of law.

Even with progressive scan technology, a sufficiently high frame rate is necessary for forensic analysis of video of scenes involving objects or people in motion. It is critical that you have enough frames of the incident to accurately ascertain whether an object was tossed by an individual or which suspect actually used the weapon. If a vehicle is spotted fleeing, did you get a frame showing the suspect in the vehicle as it turned and sped off? Or, did you capture a frame that clearly showed a suspect’s tattoo or other identifying artifact? Having enough frames of the event for in-depth forensic analysis can break a case wide open and lead to the actual perpetrator being brought to justice swiftly.

Oftentimes, spotting critical details in a crime scene video involves playing back the video in slow motion, similar to the replay of a questionable call in a baseball game. With the HDTV standard, you are guaranteed a very high frame rate (25/30 fps, depending on country) which ensures capturing the necessary level of forensic detail, as well as documenting the true flow of the incident as it played out.

While one, two and three megapixel cameras might support these same high frame rates, as the amount of pixels increases to five and beyond, most processors today do not have sufficient capacity to both process all the data coming from all the pixels in the image sensor, and provide a full 30 fps frame rate. As a result, higher resolution cameras generally do not provide sufficient frame rates to effectively use the video in a forensic manner. Some could argue that the cameras themselves can be adjusted to achieve the higher frame rates comparable to the HDTV standard, but this is only achieved by reducing the resolution of the camera, effectively providing a lower megapixel resolution than expected.

Pixels per Foot

It is important to determine how you plan to use the video for forensic detail. Quite simply, do you merely want to monitor an area and detect that something or someone was there, or do you need to recognize precise details from the scene and identify colors and other subject particulars pertinent to your forensic investigation?

In the United States, there are no publicly published standards from any law enforcement agency that govern standards for the number of pixels per foot required for evidentiary video. When dealing with forensic video, most focus and legal standards are based upon the handling of the evidence or chain of custody. In some European countries, however, law enforcement agencies are already setting minimum pixel counts necessary for accurate identification of people and objects for evidentiary purposes.

As an example, for simple observation, 20 pixels per foot or 640x480 would be adequate. For forensic review, the number should increase to 40 pixels per foot or 1024x 768. But for recognition, resolution should be around 80 pixels per foot or 1280x960 for maximum clarity.

While no hard and fast standard exists here in the United States regarding the number of pixels per foot required to make a positive identification, using the European guidelines could help you in setting up a surveillance system that reflects your desired use for the video. Will you be using the video for detection, recognition or identification? Your requirements will determine the level of clarity you need.

Close More Cases

Applying HDTV standards to video used for forensic analysis provides the basis for acquiring evidentiary-quality images you can use for investigative purposes. HDTV performs flawlessly in real-time and slow motion playback. Now that the technology has become mainstream and more affordable to deploy because of advanced H.264 compression, using HDTV for video forensics in order to close cases is well within your department’s budget. If challenged, you will have confidence that you took the necessary steps to ensure you used the absolute best tools possible—the same tools supported by international standards set by SMPTE for High Definition Video.

John Bartolac has more than 20 years of experience in enterprise-class IP video solutions and access control, 12 years focused on the government sector. Bartolac is currently Government Business Development and Programs Manager for Axis Communication, www.axis.com.

Photos courtesy of Axis Communications.

Published in Law and Order, Jun 2010

Rating : 5.5


Comments

Comment on This Article

RE: A few good points, but....response continued

By John Bartolac

4 - Cont. - But the purpose of the section was to focus on the number of pixels in an image such that the final data revealed in the image provides a clearer, less pixelated rendering of the recorded image. Maybe we can expand upon this section in a future article allowing more detail. 5 - To address this comment, lets go back to comment number 2 where the issue of cost was brought up. Higher resolution video creates more data, A LOT of data. And in a digital format, that data must be transmitted across the netowrk and it has traditionally required a lot of very expensive storage media. Not to mention the additional rack space required in an IT closet to house the storage units. The advent of H.264 significantly reduces the amount of data transferred and stored H.264 is 80 less vs. MJPEG, and 50 less vs. MPEG4. This therefore significantly reduces the costs associated with transfer and storage of the data, thus facilitating the use of higher resolution video on a broader scale. Therefore, H.264 could be the holy grail in allowing Law Enforcement to deploy more higher resolution video using limited budgets and IT real estate. Final Comment - Unfortunately publication limitations dont allow the ability to expand in detail on my experience. So, for your benefit it breaks down like this: 4 years in enterprise level intrusion detection, 9 years in enterprise level access control siolutions and 7 years in enterprise level IP video solutions. Thanks again for reading the article, providing comments and encouraging further discussion. Best Regards, John

Submitted Oct 21 at 10:23 AM

RE: A few good points, but...

By John Bartolac

Thanks for your comments. In response: 1 - In the security industry, when dealing with video and how it is used for evidentiary purposes, the term commonly used is video forensics as the video from CCTV systems is used and analyzed to extract as much forensic detail as possible. While yes the term forensic can be applied in a scientific way, when addressing how video is used, we refer to Websters definition of forensic as belonging to, used in, or suitable to courts of judicature or to public discussion and debate. 2 - Agreed, cost has been a big hindrance to law enforcement actually being able to implement and deploy newer technology on a larger scale. So while the actual end users of the technology would like to adopt the technology, the fact remains that it is not being used in a broader scale. There are countless antiquated technologies still deployed and being used today by those in the Law Enforcement community which could be improved upon. However, you are correct in that the slow adoption has historically been atributed to budgetary issues. And its unfortunate that sometimes it takes a large media event, after a crime is committed and or lives have been lost, to force a government agency or business to make the technology upgrade. With more proactive planning and education, maybe the ROI can be explained up front and help ease some of the budget burdens. 3 - Im very familiar with SWIGIT. However, SWIGIT is again, as you stated, best practices. SWIGIT is not an adopted standard here in the United States that can be pointed to and used in a court of law again, key word here standard. The SWIGIT guidelines are a good and effective set of best practices, but not a published standard like SMPTE. 4 - Yes, focal length, FOV and lighting all come into play in achieving a good quality image. But the purpose of the section was to focus on the number of pixels in an image such that the final data can reveal a clearer, less pixelated rendering of the record

Submitted Oct 21 at 10:06 AM

A few good points, but...

By Anonymous

1 This article has little to do with video forensics, and more to do with evidence produced by a CCTV system. They are not the same. For instance, a finger print is not the same as the forensic science of analyzing fingerprints. 2 Law Enforcement has not been slow to adopt video technology, quite the contrary. Businesses implementing CCTV have been slow to upgrade systems though, predominantly because of costs. 3 The US law enforcement community has had published BEST PRACTICES regarding the use of commercial CCTV systems for several years. Visit the Scientific Working Group on Imaging Technologys site to learn more - www.swgit.org 4 Your Pixels per Foot section is meaningless. Simply specifying an image raster size does not equate to an acceptable standard for identification. What about the objects percentage of that raster size? Field of View, Focal length, lighting, etc...? 5 H.264 is not the holy grail. A capable family of standards yes, but there are others just as capable. Law enforcement in the US has been trying for several years to get the hundreds literally of DVR/NVR manufacturers to standardize digital recording technologies and devices...or at the very least provide their proprietary players and codecs to law enforcement. And finally, I see that the author has been doing enterprise-class IP video solutions for longer than IP video has existed...in fact, prior to the Internet being a public network. Even the military was not using IP for video 20 years ago...not even for testing purposes they were using X.25. While I appreciate the effort to educate readers on some of the problems with CCTV, Id encourage the author to do a little more homework prior to publishing a national article next time.

Submitted Jul 2 at 4:19 PM

Related Companies

Axis Communications
 

Related Products

High-Definition Television TechnologyVideo Forensics
 

Article Images

Click to enlarge images.

 
Events and Tradeshows: LAOPFMTRPSIT
Latest News: LAOPFMTRPSIT
 
Close ...