VIDEO MOTION DETECTION

Theory and Applications

The question that is often asked is; "Why video motion detection should be used in preference to many other technologies?" This document will outline the major advantages of Digital Video Motion Detection ("VMD"), and some of the most popular applications. Included is information that the user new to this technology must be aware of to make it effective. VMD is the digitization and analysis of a video picture generated, usually by a CCTV camera. Movement is detected as a change in the video signal in relation to a reference image of a specific location, video scene or part of a video scene. It is actually "Computer Vision" and is used in many scientific and industrial applications as well as security.

Advantages of a VMD

VMDs provide instant alarm verification by providing a video picture to security personnel and/or a video recording device, allowing for a quick and appropriate response.

VMDs are not bound to detection in a straight line, as are most other technologies such as IR beam barriers that is used in PIR and Microwave motion detectors. A single camera can protect a large area and can be selective in its detection in a given area. With the use of "masking" (selecting only a part or parts of the video scene that the VMD will protect, ignoring activity in the unmasked portions), the VMD can discriminate between multiple zones created on one camera view. Depending on the performance (and cost), the system can be extremely sensitive, down to a single pixel of video.

VMDs can operate remotely from a "danger zone". The camera can be placed in a safe location while protecting an area quite far away from it.

VMD systems can drastically reduce installation costs in some installations. Detection can occur on the other side of a road or river if needed without having to install cabling across it or add some type of wireless link.

Technology - Analog vs. Digital

Analog

Older generation VMDs were analog which are still in production today. Typically it provided very a limited ability to analyze an image and exclude false alarms, such as leaves on a tree, a paper bag caught on a fence in the detection zone or waves on a pool within the camera scene being protected. Nuisance alarm rates are high or the sensitivity must be set so low that detecting a true event is questionable.

Some analog VMDs which are still available are as simple as a small suction-cup with a photocell attached. This would be stuck over a small point on a video monitor and would alarm should that small spot change in intensity. While a good idea and in fact the basis of video motion detection, it is far from practical for most video monitoring applications.

Analog types with an adjustable window or zone actually integrates or averages the entire zone to produce one value for the intensity of the zone. When this overall value changes for any reason up or down from the reference setting beyond the range set for upper or lower limit, motion is then detected. While some models use On-screen displays and the look of digital while still using the above technique are still classified as analog video motion detectors.

This analog technology generally offers no resistance to camera vibration and normal light level changes throughout the day and must be used in strictly controlled environments.

Digital

Digital Video Motion Detectors are generally classified as units that use A/D (Analog to Digital) Converters to sample the incoming video signal and electronically convert it to a digital value. The internal processor of the VMD then determines video motion by processing the image in the digital domain.

Pixel Sampling / Resolution

The more resolution or higher sampling rate of the video signal gives greater accuracy of VMD and therefore higher performance. Digital VMDs range from sampling just a few pixels to 262,144 pixels for one captured video image. Each pixel will have a sampling value associated with it and this is normally referred to as the number of bits of the grayscale. If a 5 bits is used then each pixel would have 32 different grayscales available. If 8 bits was used then 256 different grayscales would be available. Some VMDs have even higher if they digitize color information separately.

Computing Resolution

High Resolution Motion Detection means further distances, wider field of view, and slower and smaller moving targets.

For a 1/3 " CCD Image Sensor

Field Of View (ft)

Pixel Width (in)

Velocity Requirement (in/sec)

Typical Lenses (mm Focal Length):

2.8, 3.6, 4, 6, 8, 12, 16, 25, 35, 50, 75, 100

Memory

The higher performance the VMD the more memory it will have. Video images are sampled into memory by one field and compared to the next field and so on. The more memory the fields or pictures can be stored and the better the video analysis between fields. Some modes of VMD hold a reference picture in memory continuously and compare all new images to the reference. For special trace, tracking and motion modes additional memory is required to these special viewing modes in additional to the VMD modes.

Processor

Generally the faster the processor, microprocessor or DSP the more pixels that can be processed per second. The faster this analysis the better the resolution and accuracy of the VMD. The speed of the processing is not determined only by the processor speed but how many images computations per second is actually accomplished. A PC based VMD may have a processor speed of 500 Mhz but actually running at 10 MIPs (Million Instructions per Second) of video computation. A stand alone unit may have dedicated DSP and/or dedicated microprocessor and/or a PLD running at 12Mhz each and have a combined processing speed of 52MIPs or higher.

Digital VMDs Advantages

Choosing a device

As with any security installation, most of the problems encountered with a VMD installation can be avoided by the correct selection of the device before the installation begins.

THE UNIT MUST HAVE A RELIABLE "OUTDOOR" SETTING or capability if it is going to be used with any cameras in a outdoor environment. This is important even if all cameras are indoor but some scenes contain 'virtual outdoors' such as glass doors or reflections from outdoors.

Applications

A common point of confusion is the difference between "Multiplexers"(Muxes) or Digital Video Recorders (DVRs) with their built in activity/motion detection and VMDs. Multiplexers and DVRs usually do not have true video motion detection but have activity detection that is intended to detect any type of motion in the video scene. This feature is used to record images with motion more frequently than images with no motion so the playback offers a seamless image of motion. When this is the purpose it is called "Activity Detection".

Generally the Muxes are designed to time base correct video images in real time (60 pictures per second) with inputs up to 16 cameras. This process requires high speed digital processing in multiport memory within the device. To achieve activity detection they generally use one detection circuit where each video channel is switched to it separately and video motion processed for a few fields and then one to the next. Other units sample at the same rate as the multiplexer samples the video which is faster than the first method. Both methods have a relatively slow response to activity detection and is not suitable for video motion detection.

Low End VMD devices of the analog type as described above suite the purpose of detecting any type of motion in the scene. For Mux's and DVRs this is acceptable since the activity detection is used to record more pictures per second to give the user more information in areas that have something of interest.

Low End VMD - low motion rates and/or selective cameras and times for alarms Mux's / DVRs - high motion rates on most cameras, watch or record all cameras at all times

Medium Performance VMDs allow detection in most 'normal' scenes, such as parking lots, underground garage areas, inside scenes, close-in outdoor perimeter, space and selective detection. These devices are generally used where the lighting is consistent and nature can not interfere.

Hi-End VMDs are generally very good at false alarm-rejection. Very small (pixel level) detection and direction of motion, for use in low contrast, slow movement, poor illumination areas. These devices can have special modes of motion detection for motion tracking, museum applications and high security areas.

Most applications being pursued today are:

High End VMD

Medium or High End VMD

Low End

APPLICATION NOTES

VMDs Single Channel

The vast majority of video motion detection is using a single camera or single channel. The low to medium end detectors are used where existing cameras are installed and a simple connection between the camera and the monitor can detect the presence of motion without the additional cost of wiring and installation. Some applications would be

The detection of someone coming in or out of door to a restricted zone.

The detection of any movement in an area to trigger a VCR to record this specific event or increase the number of pictures per second on a mulitplexer. When connected to a video switcher the areas of movement would presented to a guard and no view the other areas without motion.

For High End applications the major uses are high security areas where the detection of personnel movement triggers an alarm and is intended to give an immediate armed response. A typical application is watching the perimeter of a prison for motion the size of a man. You can view approximately 60-100m with one camera and accurately detect a man escaping while filtering out the motion common in nature. An airport application would be the surveillance of the a runway where one camera could view the entire strip and detect any large objects like airplanes or cars coming on to or off of the runway without alarming when normally landings and take offs are done. For a museum applications special video motion detection modes off protection of the removal of an art object but not motion in front of it like spectators or cleaning personnel.

Single Channel Multiple Zones

For single camera or single channel video motion detectors with multiple zones, say 8 zones, this offers the performance of 8 video motion detectors in one when applied correctly. Say for instance you wanted to protect a bonded warehouse where one large room was divided into several sections for different companies. On camera could be have the correct lens to view the entire area. The VMD would be set up that each zone was set to protect one company's goods. When any of the 8 zones was breached the alarm output for that particular zone could trigger a homing PTZ to view only that area with a close up image or alert the guard to where the bread was. A logging function could be viewed later to see which zones had activity and review the tape at that time.

VMDs Multiple Channel

For large installations with many cameras and many video motion detectors the rack of 16 VMDs or multiple racks controlled by a PC is the preferred choice. Each camera will then have a dedicated high end video motion detector to say protect all perimeter walls of a prison or the entire fence of a military installation. Since these areas are very large, it would normally take several persons to constantly viewing multiple cameras to watch for motion on the premises. With VMDs installed the amount of personnel is reduce to just a few who can monitor the scenes after video motion is detected.

Multiple Channel & Multiple Zones

For large scale high security areas with multiple entrances and exits using the multiple channel rack is a must be also each camera should have multiple zones. With this feature on camera can monitor several doors of an area since high end VMDs have up to 8 zones per channel. A separate zone for each door would be programmed. Also with multiple zones direction sensing can be accomplished. Using two zones per door one could detect when someone is going in a door but ignore when someone comes out. Two zones could be assigned to detect someone walking towards a building and ignore when the are walking away.

VMDs with Video Switchers

An example installation we would have 30 cameras connected to two "dual rack" sequential switchers each with 2 monitors. With a dwell time of 2 seconds/camera the security officer has 2 seconds/camera/minute to see any activity and determine if further viewing is required, usually by selecting "SPOT" for that camera. Realistically, the security officer must perform many other tasks such as qualifying guests, logging reports and dealing with other security equipment and cannot continuously watch the C.C.T.V. system. This dramatically reduces the 2 seconds/camera/minute mentioned above, making the C.C.T.V. system valuable mainly as a deterrent to crime but rather ineffective in actually viewing any criminal activity.

Two multi-camera VMDs will operate identically as sequential switchers until motion is detected on any of the cameras. Once motion is detected, on a camera, that camera would then be displayed on the "sequence" monitor until another camera has detected motion or for a preset time. If several cameras have motion unit will sequentially switch between only those cameras that have active motion. If only 3 cameras have something to watch on them the effective viewing time becomes 20 seconds/ camera/minute. Additionally, with the warning/alarming capabilities of the VMD the officer need not look at the C.C.T.V. system until alerted by the VMD that motion has been detected and there is something to view.

Once an alarm has sounded and the camera scene is displayed on the "Spot Monitor" the officer must know where this camera is located in the building. There is rarely any camera identification available on existing switchers. Tracking movements throughout a building in this manner requires a very intimate knowledge of the premises. VMD switchers provide a camera number and/or a camera name to identify scenes and will automatically track movement on the camera system, displaying each new motion scene on the sequence monitor.

These VMDs can be setup to trigger the recorder to real-time mode when motion is detected or signal a video transmission system.

System Configurations

Multiple Channel Rack

This is a 19" rack mount system that contains 16 channels of Video Motion Detection. Each channel is programmed separately via the front panel keypad or remotely via a PC.

Quad System

This system uses four channels of video motion detection with the outputs going to a quad display unit for viewing. The alarm outputs trigger the quad to go full screen upon detection of motion in the respective camera.

Transmission System

This systems shows how the camera is first input into the video motion detection unit prior to the remote transmission system. This is necessary since transmission system have noise, loss of signal and other problems that will falsely trigger motion detection.

Multiplexer with Multiple Channel Rack

This system uses a multiple channel rack of video motion detectors in line before a multiplexer. The video motion detectors can be programmed for activity to activate the multiplexer to record more pictures per second of a respective camera if the multiplexer has no activity detection. In most applications the video motion detector is used to generate alarms to the multiplexer.

Matrix Switcher with Multiple Channel Rack

This systems show connection of the video motion detectors to alarm a matrix switcher to home to the camera with the motion for the operator to see or record. Multiple monitors can be used to view all motion scenes simultaneously.

This system uses a single unit medium range video motion detector that accepts 4 or 8 cameras. This VMD device switches internally between the cameras and checking for motion detection and then alarming a matrix switcher.

Medium Range VMD

This is the basic application of a medium range VMD with any number of camera inputs.

This system uses to medium range VMDs that handle 8 camera inputs. The alarm and video outputs are then connected to a 16 camera mutliplexers.

Single Unit VMD with 4 Camera Inputs and Quad

This systems use a standalone 4 camera VMD and alarms a quad when any motion is detected. This is a medium performance type VMD.

Medium Range 4 Camera VMD with Wireless Transmission System

This system shows the requirement of connecting the VMD unit of any quality before any type of transmission system.

RETROFITS:

A great opportunity for installation of VMD technology lies in retrofit of existing video installations, particularly in the condominium sector where multiple camera systems are generally connected to sequential switchers in both recorded and supervised (with guard personnel) settings.

Installation and use techniques

Good installation techniques are very important!

By far the largest problems, is cabling, connections (Twist-On BNCs!!) and grounding. We have found that in many installations twist-on BNCs are used extensively and improperly installed. Cables too large for the connector or the ground/shield not being folded back correctly, account for a great number of problems with noisy pictures and create ground loops. These may look fine on the monitor but close inspection can reveal a picture that varies sporadically in brightness as the ground level changes. PROPERLY INSTALLED CRIMP BNC's ARE HIGHLY RECOMMENDED!

Camera vibration, caused either from wind in outdoor settings or from general vibration at the camera, as in some industrial applications, must be minimized.

Autoiris lenses can cause alarms, even if the iris movement is caused by something in an "unmasked" area. For example a car head-light would be ignored by most VMDs, although an autoiris lens or AGC circuit in a camera may attempt to compensate for this change, causing an alarm even if the light was outside of the "masked" detection zone.

Illumination must be sufficient for the camera being installed. The size of object to be detected must be considered. Background noise can be a problem in very dark scenes requiring acceptance of either high nuisance alarm rates or reduced sensitivity.

Always place the VMD as the FIRST device in a video line, terminating at the VMD!

Some VMDs are "looping", however it is best to avoid this if at all possible unless the VMD terminates and has a separate output stage for the next device in the line. Any fault in the termination of other equipment will cause problems with the performance of the VMD when using normal looping types. These types of troubles are very difficult to trace and if the equipment is from different manufacturers there can be some confusion over who is responsible for what is thought to be bad equipment.

Some of the new telephone-line transmission devices do not allow you to set, or even verify, the termination from on site, only from the remote (off site) end!! These are best left as looping (unterminated), leaving the termination to the VMD.

Perimeter protection can be achieved by aiming a camera straight down a fence line with a high-mm lenses to allow alarming even on a medium-capability system in critical areas. Consideration must be given to outdoor cameras for protection from rain on the camera lens or snow buildup.

Summary

VMD has advanced from a false-alarm nightmare to a reliable and affordable security technology that is now widely used to solve many security protection problems.

The idea is a basic one - but it must be remembered that the system can only process the information it receives. If the cameras are of poor quality or poorly setup, if the cables are noisy or improperly connected or grounded, if the cameras are subject to excess vibration or lenses covered with water droplets then there WILL be performance problems.

No magic is needed, but good engineering is required throughout the installation process, not just during the specification of the VMD unit.