SiVMS® Video Analytics
With the help of the SiVMS® Video Analysis, the handling of complex video surveillance systems is only cost-effective. In order to reliably evaluate the large number of video signals, an enormous number of specialist personnel would be needed, which would be financially rarely affordable for a company. The analysis functions largely automate this evaluation and make a rough preselection.
SiVMS® has a comprehensive set of image analysis systems for almost all monitoring tasks. For high-performance applications, for example, optimized systems are available for use with thermal imaging cameras, which combine highest detection rates with large detection ranges, especially in the dark.

SiVMS® Video Analytics Service
The SiVMS® Analytics Service extends the SiVMS® and nestore® video servers with a set of internal algorithms and interfaces to third-party video analysis systems.
Functionality
The analysis service is a Windows system service without its own user interface. All configuration after setup is done through the video management system and through the SiNEO® administrator.
The analysis service communicates with the video server via a TCP / IP interface. For video analysis as well as for processing the image signals for third-party systems, the analysis service requires the video streams from the sources for which analysis algorithms have been defined. For this purpose, the service uses the so-called proxy channels of the video server.
The results of the video analysis are provided to the video servers and are available as events. Furthermore, the analysis service provides video streams with embedded data from the analysis, which can be displayed at the workstations.
Particularities
The analysis server can be used with all server video sources that provide MJPEG, H.264 and MPEG-4 video streams. Each video channel of the video server can be individually assigned to all algorithms provided in the service. However, the processing of the signal (decoding) takes place only once per video source, ie the streams received by the video server are received once and then distributed to the assigned algorithms.
The analysis server has its own decoder for MJPEG, H.264 and MPEG-4 video streams. To optimize server performance, image sizes and frame rates can be reduced separately for each video source. Furthermore, main frame filters are available for the H.264 and MPEG-4 formats, which only decode the first picture of a GOV (H.264) or GOP (MPEG-4) and thus enable a considerable reduction of the system load.
Interfaces to third-party analysis systems
In addition to the native algorithms, the analysis service SiVMS® video analysis service has interfaces to third-party analysis systems, which are seamlessly integrated into video management. In most cases, these analysis systems access or are already part of the video sources, but can also be supplied via the proxy channels of the video servers with MJPEG, H.264 and MPEG-4 image streams. At these interfaces, the analysis data are forwarded to the video system and in addition a video signal is generated, which is formed from the original stream and the analysis data shown in the image (meta data). The analysis results of the external systems are passed as events to the video management.
The analysis service differentiates between two methods for third-party analysis systems. The first method involves capturing the events of the analysis. Here, the alarms are fed to the video management and trigger directly functions. The second method extends the interface by processing the so-called metadata, which allows localization and determination of the detected objects. For example, with the second method it is possible to dynamically display the objects in site plans or to perform 3-dimensional camera controls.
The support of the respective method depends on the type of data that the third-party analysis system provides via the interfaces.
Supported Algorithms
SiVMS® Algorithm:
- SiVMS Motion Detection (free)
- SiVMS Tampering Detection (free)
- SiVMS Smoke Detection
- SiVMS Removed and Left Item Detection
- SiVMS License Plate Recognition
- SiVMS License Plate Recognition Pro
- SiVMS Object Tracking
- SiVMS Hot Spot Detection
SiVMS® 3rd Party algorithm:
- Bosch IVA Video Analytics
- iOmmniscient Video Analytics
- IPS Video Analytics Server
- IPS Intrusion Detection AXIS
- AXIS Crossline Detection
- Siemens Site IQ
- Siemens SiCore LPR
- NAVTECH Radar
- SICK Laser

Evaluation of the Analysis Results
The video analysis alarms are forwarded to the video server as events. For this purpose, the triggers can be individually assigned to the events of the respective video source. Within video management, these events are in the usual form e.g. for hook-ups or recordings and for forwarding to message management.
The analysis service also works as a video server that can be accessed via the workstations. The video streams to be retrieved correspond to the images of the original source with additional data shown in the image from the respective algorithm (OSD – On Screen Display). For algorithms that detect areas or objects within the image, this information is added to the video stream as additional metadata and is available for example for 3-dimensional and automated PTZ controls.
Furthermore, the analysis service has an optional OPC-DA server, which allows external access to certain events and conditions of the video analysis. It accesses the server service via the SiNEO © data model and adjusts the data automatically.


SiVMS® Motion Detection
SiVMS® Motion Detection
The Motion Detection (Motion Detection) algorithm is used to detect changes in image content. For this purpose, the current picture is compared with a previously stored picture and an alarm is triggered when a threshold value is reached. The threshold is defined as a percentage image change within an adjustable image zone.
Like all other video analysis algorithms of the SiVMS video analysis service, the motion detection algorithm can be integrated into a timetable and therefore limited to specific time periods within a weekly schedule. Likewise, the image areas to be detected can be restricted.
The analysis result indicates the areas of the detected movement by image contents marked in red. Furthermore, the percentage image change as well as the trigger threshold is shown.
SiVMS® Tampering Detection
SiVMS® Tampering Detection
The “Tampering Detection” allows a comprehensive check of the functionality of a video source. The algorithm includes a check for twist, a noise level measurement, and a control of the camera’s coverage and glare. Each component of this check can be configured separately and placed on different or shared events.
Like all other video analysis algorithms of the SiVMS video analysis service, the tampering detection algorithm can be integrated into a timetable and therefore limited to specific time periods within a weekly schedule. Likewise, the image areas to be detected of the algorithm for the noise level measurement and the manipulation measurement can be limited.
The analysis result indicates the areas of the detection by means of image contents marked in red. Furthermore, the detection ranges for the manipulation level (blue) and the noise measurement (red) are displayed. The lower part of the screen shows the measured values and trip limits.
SiVMS® Smoke Detection
SiVMS® Smoke Detection
The Smoke Detection (Smoke Detection) is used to detect the occurrence of smoke in defined image areas. The algorithm is based on a measurement of the standard deviation, e. g. on a loss of contrasts. The analysis should refer to non-moving image areas with as strong edges as possible in order to achieve a high detection rate.
Like all other video analysis algorithms of the SiVMS video analysis service, the smoke detection algorithm can be integrated into a timetable and thus limited to specific time periods within a weekly schedule. Likewise, the image areas to be detected for the noise level measurement can be limited.
The analysis result displays the red marked detection areas. In the lower part of the screen, the measured values for the image brightness and the smoke level with the trigger limit are displayed.
SiVMS® Removed and Left Item Detection
SiVMS® Removed and Left Item Detection
The Removed and Left Item Detection is used to detect long-term image changes, such as theft, leftover objects, or indoor and outdoor spills. An adjustable storage depth of up to 65535 images results in a separation of the image background and a subsequent detection of a change in the image content.
Notes on the algorithm
The algorithm can be used in both quiet and moving scenes. It should be noted that the separation of the image into the foreground and background takes a certain amount of time for moving scenes. Each movement in the image leads to a recalculation in the movement area, whereby for each pixel, a separate processing is performed.
The algorithm automatically compensates for short-term brightness fluctuations below the difference limit and long-term changes such as daytime and nighttime. Sudden image changes, such as reflections, however, cause an alarm if the image area is changed within the set object boundaries and beyond the memory depth.
When setting the memory depth, make sure that it is larger than the dwell time of objects not to be detected, such as persons or vehicles in one place.
To start the analysis, the detection is blocked until the memory depth is reached for the first time. This is indicated by a display. The blocking time is calculated from the memory depth divided by the frame rate. The startup phase is called again when a picture change by more than 50% occurs, as can be done for example by changes in the illumination.
Like all other video analysis algorithms of the SiVMS video analysis service, the Removed and Left Item Detection algorithm can be integrated into a timetable and thus limited to specific time periods within a weekly schedule. Likewise, the image areas to be detected can be restricted.
SiVMS® License Plate Recognition
SiVMS® License Plate Recognition
The analysis algorithm “SiVMS License Plate Recognition” is used to recognize license plates and to process them by, for example, matching the identified license plate with a list (Whitelist). If the indicator is found in the list, the route is approved by opening a barrier.
SiVMS® Hot Spot Detection
SiVMS® Object Tracking
SiVMS® Object Tracking
This algorithm is used to capture moving objects and track them until they have left the video image area again. Each object is individually registered, placed into a frame, numbered and tracked independently of the other objects. Optionally, the omitted distance of each object can be represented as a line.
AXIS Crossline Detection
AXIS Crossline Detection
IPS Intrusion Detection (AXIS)
IPS Intrusion Detection (AXIS)
IPS Intrusion Detection is an intelligent video analysis for real-time alarms when entering protected areas. The analysis is designed as an add-on module for certain AXIS cameras and encoders and runs directly on these systems. The analysis service offers an embedding of this module into the video management and an analysis of the analysis results. In addition to the alarm status, the algorithm also provides object positions, which can be used for automatic PTZ traces and position map settings via the command versions of the VMS.
Like all other video analysis algorithms in the SiVMS video analysis service, IPS Intrusion Detection (AXIS) can be integrated into a timetable and therefore limited to specific time periods within a weekly schedule. Likewise, the image areas to be detected can be restricted.
IPS Analytics Server
IPS Analytics Server
Via the interface “IPS Analytics Server“, it is possible to connect to an IPS analyzer server and react to various analysis results. The SiVMS receives the metadata of all active analyzes, which can then be processed separately.
Like all other video analysis algorithms in the SiVMS video analysis service, IPS Analytics Server can be integrated into a timetable and therefore limited to specific time periods within a weekly schedule. Likewise, the image areas to be detected can be restricted.
iOmniscient Video Analytics
iOmniscient Video Analytics
The iOmniscient video analysis is a server-based software with a variety of algorithms for detection in indoor and outdoor areas. The analysis service offers an embedding of this software into the video management and an analysis of the analysis results. In addition to the alarm status, the algorithm also provides object positions, which can be used for automatic PTZ traces and position map settings via the command versions of the VMS.
Like all other video analysis algorithms in the SiVMS video analysis service, iOmniscient video analysis can be integrated into a timetable and therefore limited to specific time periods within a weekly schedule. Likewise, the image areas to be detected can be restricted.
Siemens SiteIQ Analytics
Siemens SiteIQ Analytics
The analysis algorithm “Siemens SiteIQ Analytics” is used to capture moving objects and track them until they have left the video image area again. Each object is individually registered, placed into a frame, numbered and tracked independently of the other objects. This information is forwarded as metadata to the interface of the SiVMS and can then be processed. Optionally, the omitted distance of each object can be represented as a line.
Like all other video analysis algorithms in the SiVMS video analysis service, Siemens SiteIQ Analytics can be integrated into a timetable and therefore limited to specific time periods within a weekly schedule. Likewise, the image areas to be detected can be restricted.
Bosch IVA
Bosch IVA
The analysis algorithm “Bosch IVA” is designed as an additional module for certain Bosch cameras and is executed directly on these systems. It serves to capture moving objects and track them until they have left the video image area again. Each object is individually registered, placed into a frame, numbered and tracked independently of the other objects. This information is forwarded as metadata to the interface of the SiVMS and can then be processed. Optionally, the omitted distance of each object can be represented as a line.
Like all other video analysis algorithms in the SiVMS video analysis service, Siemens SiteIQ Analytics can be integrated into a timetable and therefore limited to specific time periods within a weekly schedule. Likewise, the image areas to be detected can be restricted.
NAVTECH Radar
NAVTECH Radar
The “NAVTECH Radar” analysis service is used to record moving objects and track them until they have left the radar area again. Each object is registered individually, captured in a red frame and tracked independently of the other objects. The interface provides the GEO coordinates of the tracked objects as metadata, which can then be evaluated and displayed by the SiVMS analysis service.
Like all other video analysis algorithms in the SiVMS video analysis service, NAVTECH Radar analytics can be integrated into a timetable and therefore limited to specific time periods within a weekly schedule. Likewise, the image areas to be detected can be restricted.
Follow us on Social Media