Video motion detection (VMD) is rapidly changing how security professionals use video. The term "motion detection" as it is applied to digital video is ambiguous. It can refer to capabilities ranging from simple activity detection to the search of massive databases to pre-empt serious incidents. Therefore, if you desire to implement a VMD system, you have much to consider.
How Video Motion Detection Works
When a video image is converted to data in a digital format, the image information becomes a stored digital value. This digital value changes as the video image, the source of the data, changes. Complex algorithms (a series of analytical steps) analyze the changing digital values to recognize patterns. Some vendors call this video content analysis.
Since these algorithms are a software function, they are programmed into chips and boards that can be installed in cameras, stand-alone modules, digital video recorders and dedicated computer processors. VMD is also available as software for installation in off-the-shelf computers.
The complexity of these products varies greatly. The IP cameras provide a separate output on basic activity detection, while the PC-based software and modules provide graphic identification of the identified movement, user-selectable monitored areas, compensation for environmental movement, and a host of other features.
The table on page 28 lists several products and the systems on which they operate. It is not intended to compare apples to apples; it simply shows the variety of equipment upon which VMD can reside.
Available VMD Features
Because VMD terminology is not standardized, the vendor-supplied product information can be less than concise. Selecting the appropriate VMD approach requires understanding the features available, and understanding requires study.
Basic motion detection typically recognizes any type of motion in the video field. A single output then activates automatic call-up to the monitor screens of surveillance personnel or initiates automatic DVR recording. The video call-up is no longer limited to cabled CCTV systems, but can be transmitted via the Internet or wirelessly. Many basic DVRs can search and retrieve records of movement or activity on their stored hard drives. These features are often found on off-the-shelf equipment, are economical, and have limited applications. Advanced VMD products enhance the concepts of basic motion detection and can, when properly applied and operated, provide innovative, effective solutions to security issues. Most of these features result from elaborate algorithms that search out detailed movement patterns and only activate a system response under very specific conditions. Capabilities include:
- Intruder Identification: Identifying unauthorized humans in specified areas of the field of view.
- Environmental Compensation: Recognizing and ignoring wind-blown debris, animals, background traffic, etc.
- Counting: Recognizing a quantity of a particular object moving or activity performed.
- Directional Identification: Ignoring objects moving in one direction, while alarming for objects moving in unauthorized directions.
- Item Recognition: Activating when specific user-selected items are removed from, placed in, or passed through the field of view.
- Subject Tracking: Highlighting and following a specific person or item as it moves about the field of view, or from the field of view of one camera to another.
- Multiple Subject Tracking: Highlighting and following multiple persons or items simultaneously as they move about the field of view, or from the field of view of one camera to another.
Advanced programming for "specific act recognition" is just beginning to emerge from development. This is motion detection that recognizes unique and complicated motions associated with undesirable acts or phenomena. Recognized acts can be the typical movement of shoplifters, acts of physical violence or phenomena such as fire. Video smoke detection programs are currently being marketed. Also emerging is the coupling of motion detection with character and biometric recognition. Systems that will spot a moving license plate or vehicle signage, process the number or characters remotely, and activate a response on a wanted vehicle should not be far in the future. Similarly, recognition of faces of note moving in a crowd will be finding their way to market.
Versatility of these systems continues to expand. Some systems can now learn new responsibilities by adjusting for the repetitive motions and routine activities. Other products are often capable of environmental adaptation, so as to be unaffected by changing light or weather conditions.
Virtually all advanced motion detection can be customized to some extent by the end-user. Programming can accommodate shifting operational requirements according to time of day, field of view, and changes in environmental motion. Most multi-camera, user-programmable systems permit the user to customize the visual response or the automatic response to any recognized movement pattern. The operator interface screens and devices can usually be modified standard computer operating system toolbox techniques.
What VMD Features Will Benefit My Application?
Applying the correct VMD equipment and features can solve numerous security challenges. The most important step, however, is not to look at the available equipment until gaining a thorough understanding of, and applying imagination to, the security and operational needs of the facility.
The security director, who understands his facility, and the design professional, who understands the options and applications, must begin by asking themselves the important questions: What can move? and What do we want to know about its movement? As simple as these questions appear, the answers must be detailed, accurate and complete.
The first objective is to identify what can move. This determines the surveillance areas to be covered by the cameras and begins to define the VMD product required. The answer to What can move? includes items of interest and any moving background items that may distract the system.
The items of interest can be items that are typically in motion and therefore either pass through the field of view, stop in the field of view, require identification, or must be followed by the surveillance apparatus. Some of these include:
- Vehicles moving through entrances or on prescribed traffic routes.
- Routine entry and exit of authorized personnel.
- Baggage left unattended.
- Personal property that is carried by the public.
- Suspect individuals.
- Employee work methods or handling of assets.
Conversely, the concern can be for items that are typically stationary and should not be moved. This is often applied to protecting critical or easily negotiable assets from theft or misuse. Items of interest may also be unusual circumstances or phenomena that are characterized by a particular motion. This list includes, but is not limited to:
- Intruders or unauthorized personnel in an area or perimeter.
- Leaks/mechanical failures.
- Smoke, fire or flame.
- Violent or erratic behavior.
- Counter-flow directional movement.
Once the security director and the design professional have decided upon the items of interest, they must determine the best coverage areas and propose a layout. Often the location of the existing CCTV cameras provides a good starting point and the modifications can be minimal. Other times a completely new layout is necessary. The presence of background movement and environmental conditions can be a major consideration in the layout. As mentioned before, systems exist that can screen background movement and adapt to environmental changes. These capabilities, however, are not standard on many VMD systems, and should be specified by the design personnel if the layout determines that they are necessary.
Now that the type of movement is understood, the next criteria affecting design and selection should be what to do when the motion of interest occurs. If the item of interest requires immediate response, active surveillance personnel must receive the image and understand what they are seeing. Most motion detection has alarm states capable of seizing an active monitor. Many have enhancement graphics that will highlight the movement in question. This can be of enormous benefit to dispatchers communicating to responding personnel. If subject or object tracking is required, digital pan-tilt, mechanical pan-tilt, or panoramic cameras may need to be activated and under the control of the motion detection processor.
If a primary purpose of the video is documentation for prosecution or litigation, changes of the field of view to accommodate the movement should be minimal, and more cameras should be implemented to confirm the events. Also, the graphic enhancement of the VMD, the storage methods of the video, and the signal compression methods must be more closely scrutinized when selected to minimize controversy. If the detected movement requires identification or authorization, the VMD processor must be capable of interaction with the corrected database. Sometimes this requires the integration of video with access control or other file servers. The VMD may also be required to translate the required video information for comparison to the database information.
Recommendations
The considerable range of products and their complex capabilities bring uncertainty to the implementation of VMD systems. What is certain, however, is that VMD will continue to develop new capabilities for application to security issues. Therefore, for any CCTV system that now has, or may have, need of any but the simplest VMD products, the following should be considered:
- Apply imagination. If you think it can be done, it probably can.
- Plan for continual upgrades to the system and to the functions it performs. Phasing of VMD projects can be beneficial.
- Don?t skimp. Off the shelf is fine for basic needs, but complex solutions require study, design and funds.
Though the complexities bring uncertainty, the innovations offered by VMD create an exciting new potential to provide safer and better managed environments. Keeping informed is, therefore, the best tool for VMD.
Kenneth L. Gentile, P.E. is a senior consulting engineer with the Houston, TX, office of Rolf Jensen & Associates Inc. To learn more about RJA, visit its Web site at www.rjainc.com.
