As Axis Communications' Security Industry Liaison, Steve Surfaro consults with a number of industry associations, including ASIS, BICSI, SIA and NBFAA/ESA, on physical security technology innovation and best practice adoption.
Recorded video and related feature data of events is one of the first and most important resources for incident review. But what is available in this data, and what are the opportunities for industrial security, first responders, operations management, safety and loss-prevention professionals? To understand this, we first need to define what is included with this “forensic” video data and then apply a process for its use.
Digital Multimedia Content — more than just video data
Digital Multimedia Content (DMC) is also known as digital video data, IP video content, or Digital Multimedia Evidence (DME). The purpose of referring to this evidence, as “multimedia” is essential in understanding the different digital data categories included. Video content, audio content, metadata-incorporating object characteristics such as color, size, trajectory, location-based information, relevant IP addresses, recording time and system time may be attached or associated with a digital video file. DMC may be compressed or uncompressed and may also be referred to as original, copied, local or virtual. Compressed DMC is the most common video data available that has been transcoded from the original DMC in an industry-standard file format, resulting in a reduced size and network bandwidth required to represent the original data set. Advances in h.264 video compression, the ability to store DMC within the camera or video-encoding device itself and virtualized or cloud computing have dramatically improved the volume and duration of video data available to investigations.
Uncompressed DMC or a copy of the original DMC with no further compression or loss of information that is in an industry-standard file format — although desirable by professional video-evidence examiners — is often unavailable and can be an unreasonable expectation due to the far larger storage requirements. Given the choice between having compressed video evidence that can be used together with other data, still-image photography is what most security professionals prefer.
The act of applying forensic video technology to this DMC defines “forensic video review.” Some of these review tasks include playback and analysis of DMC, together with applying a scientific methodology of forensic video analysis.
Also important is the use of DMC evidence in the legal setting, performing data recovery as needed, performing forensic image comparison and the development of a visual presentation of evidence.
DMC authentication and tamper detection are examples of maintaining the chain of custody for DMC evidence as specified under Law Enforcement and Emergency Services Video Association “Guidelines for the Best Practice in the Forensic Analysis of Video Evidence.”
Applying an understanding of the effect of light on the scene can improve the image quality of the video content. Advances in camera technology that produce usable color or “find the light” in dark- or low-illumination scenes are improving forensic video content.
The design of the video solution to provide maximum coverage is of great importance for systems used for forensic review. Using standards-based, high-image quality sources like HDTV IP cameras and technologies to accommodate difficult lighting will improve the recorded image quality.
Video content analysis
Applications that use video analytics can perform complex repetitive functions such as object detection and recognition simultaneously on many channels of video. These tools can provide improved searches, based on object characteristics and behavior.
Designers consider this where:
- The system uses a large quantity of cameras that require monitoring for specific conditions or behaviors that are capable of being recognized.
- The setup and installation are relatively simple for the video analytics subsystem, which has high, sustained accuracy for the types of behaviors and objects recognized. With video synopsis or summarization, a condensed clip of all motion for selected criteria is continuously generated and stored, allowing an “instant review” of a readily available “video synopsis.” It is possible to summarize a 24-hour period of event entries in as little as 15 minutes, reducing incident-review time by at least 50 percent. Video analytics offering abnormal scene detection allows the user to set specific object criteria and direction. The scene is analyzed continuously, and abnormal behavior differing from the majority of the scene content is detected and annunciated or marked for later review.
- Video analytics embedded in the network camera represents a growing segment where applications run and values or decisions based on recognition are available with the “edge” network camera and minimal software.
One popular example in retail and quick-service establishments is the “people counter” where the network camera and built-in app return the number of people passing into a zone, through a boundary, or into the field of view. This can provide criteria on which to increase camera frame rate and stored resolution during the time of highest traffic.
Another popular video-recognition solution that runs either as an embedded network camera application or in the Video Management System is fixed License Plate Recognition and Capture (LPR/LPC). This specialized app captures license plate information for immediate processing by LPR software. The software may run in a rapid-acquisition mode and compare plates later against an approved list or perform the recognition sequentially as the vehicles pass within the camera field of view. In either case, LPR is a mature application embraced by law enforcement, electronic-toll collection, and parking management organizations; the trend to embed this function reduces cost and allows greater flexibility.
“Heat” activity mapping provides a visual color-coded summary showing how people have moved in the camera scene for a fixed duration. Useful in retail environments where “business intelligence” data is needed, this type of video content analysis can improve safety by analyzing the flow of pedestrian and vehicular traffic in a facility. Understanding personnel traffic flow will often help camera placement and ultimately the video forensic-review process.
As Axis Communications’ Security Industry Liaison, Steve Surfaro consults with a number of industry associations, including ASIS, BICSI, SIA and NBFAA/ESA, on physical security technology innovation and best practice adoption.