MPEG-7: Security industry's best kept secret?

Wednesday, August 1, 2007

Undoubtedly, digital video is an irreversible trend in the video surveillance industry. The benefits of handling video digitally seem to be well accepted.
Amidst a rush of widespread adoption, can we identify the "killer" reason for the security industry's move to digital video? It seems that even the biggest security industry players are hard pressed to clearly answer this question. I believe this is due to the fact that the standards being touted and adopted are irrelevant to the security industry. Benefits derived from video compression standards such as H.264 and MPEG-4 are limited for the security industry, because these standards are specific to motion picture and video conferencing applications, and do not take into account specific needs of security surveillance video.
To the best of my knowledge, no security industry experts have participated in defining either MPEG or International Telecommunications Union (ITU) standards. However, even if the security industry had produced its own experts ready with wisdom and insight specific to the security industry's needs for digital video standards, it couldn't have affected a better standard to address security video surveillance than MPEG-7.
You may intelligently question why the Google version of video search still doesn't exist. It's for good reason; it is still very difficult for software or machines to automatically generate meaningful tags and metadata that represent video content and then use the metadata to search for the real thing. The main issue is the complexity of analyzing video as scenes continuously change and cameras move. However, for a security camera that is always stationary, or can be assumed to be stationary for long periods of time, the complexity is greatly reduced. Software exists today for automatic MPEG-7 metadata generation of video surveillance footage. This security breakthrough for our industry should be fully recognized, promoted and adopted as the standard for security surveillance cameras.
A standard overlooked by Hollywood, MPEG-7 turns out to be uncompromisingly tuned to the surveillance industry's needs. With MPEG-7, the security video industry can start immediately to make sense of the millions and billions of volumes of recorded video footage and do so at lower bandwidth, storage volumes and costs.
Though the motion picture industry originally sought video search capabilities, it has all but ignored ISO's 2001 ratification of the MPEG-7 standard, too distracted by piracy and digital rights management concerns. The result of years of research by experts within the motion picture industry, but mostly from universities and research centers, MPEG-7 provides digital video with the same search ability that exists for digital text. Unlike better known video standards MPEG-4 and H.264, MPEG-7 is not about compression, image reproduction, or pixels. By chance, this standard's inherent ability to define and focus on events and behaviors within reams of video footage is critical to security surveillance.
The MPEG-7 standard is about creating and then searching metadata of particular relevance to security video. Since the goal in video surveillance is to call out and identify suspicious behaviors and security breaches or events, details and fine recognition in viewing isn't emphasized, though available as needed. Because MPEG-7 inherently searches for pre-identified behaviors on which to focus, it is particularly suited for surveillance security analytics. Instead of investing time and bandwidth processing H.264 or MPEG-4, which require 1-2 Mbit/sec to transmit or store streams at decent resolution and frame rate, let's refocus on how to standardize on MPEG-7 for our industry.
At once, we would realize security's compelling reason to adopt a digital format; accomplish much faster results; use much less bandwidth at 1-2 Kbit/sec rates; and realize great economy in transmission, storage and infrastructure.
While several million video surveillance cameras already are installed on our streets and well into the fabric of our societies, remember that most of these cameras are used actually only to record video, and that video is viewed mostly after the fact. The glut of video recorders available offers no means for searching all the recorded video, thus nothing to address the need. As such, we must look to standards to define effective means to search from hard drives through the enormous amount of video generated by all these cameras!
If we are serious about security and determined to exceed superficial prevention, standards must address both video surveillance applications for "record and rewind" search, as well as those for active crime prevention and live monitoring.
Again, MPEG-7 brings superlative advantage. By augmenting MPEG-4 or H.264 synthetic images with a scene reconstructed with MPEG-7 metadata, bandwidth and storage can be reduced by three orders of magnitude. At the same time, objects, events and anomalies are extracted from any scene. In so doing, security monitoring becomes correctly and immediately focused on salient security events rather than across a scene or limited to only people in a scene. Security is raised both by minimizing human bias and focusing attention precisely on suspicious events, such as trespassing, tailgating, abandoned object, removed object, direction flow, loitering, or crowd formation.
Clearly, MPEG-7 inherently addresses the Security industry's needs in video surveillance. Let's insist on this standard for our industry, resist marketing hype and trends, and fulfill the waiting world's expectations for secure surveillance.