We encounter the video compression standard MPEG everywhere particularly in the consumer sector. Internet, DVD, digital television or digital camcorders, motion images of any kind are compressed using the MPEG standard. Consumer applications mainly focus on replay and transmission of MPEG-compressed video films. The calculating time needed for generating a high-quality and strongly compressed video sequence is irrelevant. When rewinding or displaying freeze images, time lags or jumping’s are accepted. Solely a high compression rate allowing quickly transmitting and replaying the video film is of importance.
Just different for applications in video surveillance systems (CCTV).
The main focus is placed on real-time compression, high picture quality with preferably large compression rate, low bandwidths and storage costs, short latencies when transmitting live images and ideal replay mechanisms.
The currently available multimedia encoders are not suited for such applications.
The DSP-based GeViScope real-time encoder developed by GEUTEBRÜCK offers ideal performance features for CCTV-applications:
For each GeViScope media channel the requested picture format can be adjusted separately for permanent or event-triggered recordings:
704 (H) x 288 (V) Picture elements (field, 2CIF),
352 (H) x 288 (V) Picture elements (CIF),
176 (H) x 144 (V) Picture elements (QCIF)
704 (H) x 576 (V) Picture elements (frame, 4CIF) on request
A smaller picture format results into a proportionally smaller picture file size after compression.
DCS – Dual Channel Streaming
Parallel recording and network matrix.
One camera picture (frame) consists of two fields. At any one time GeViScope records one field and transmits the second field into the network. This separate processing allows live recordings with up to 25 fps (fields per second) and parallel live transmissions in the network with up to 25 fps per video channel (network matrix). The latency time for MPEG4CCTV transmissions is very low and amounts to < 150 ms in undisturbed networks and is therefore identical to M-JPEG.
VFR – Variable Frame Rate
The picture rate for permanent recordings and event-triggered recordings can be adjusted variably. The adjustment range goes from live (25 fps) up to time-lapse recordings with for example 1 picture / 10 s.
GeViScope Dual Sensor – Licence/GSC/Dual-Sensor
The GeViScope Dual Sensor (GSC/Dual-Sensor) combines Video Motion Detection (GSC/VMD) with intelligent Video Analytics (GSC/VA-Class) in one system
GSC/VMD – The proved 3-D Video Motion Detection evaluates in adjustable intervals (measure cycles) a measurement (measurement value) for the average brightness in each VMD zone individually set in the video picture (max. 128). The evaluation algorithm of the VMD compares these measurement values in relation to the set sensitivity and diversifies between global and local changes, i.e. whether there are changes to brightness in many VMD zones or in only one VMD zone. Local changes will be considered as alarm, global changes will be ignored (adjustable).
VMD zones can be connected to chains in order to detect direction-dependent or speed-dependent motions in the video picture.
The most important features of GSC/VMD:
128 VMD zones are freely definable in function, position, size and sensitivity
Measure cycles of up to 40 ms to 10 s for real time detection of very fast and very slow motions
Three-dimensional field adaptation via special perspective function
Algorithm for diversification between local and global changes in the picture
Collection and evaluation of direction and speed
Vector function can be switched on
SC/VA-Class – Intelligent Video Analytics permanently analyses the pixels in the complete video picture and generates a background model of the scene in a self-learning mode. The background model is continuously updated in order to take over global picture changes. Objects to be detected are defined in the setup according to size, speed, motion direction and status. Up to 64 areas of interest can be set for the video picture. These trigger different kinds of alarm behavior upon detection of an object. If a group of pixels in the updated picture changes in comparison to the background model the system calculates whether this group of pixels corresponds to the characteristics of a pre-defined object and reports an alarm according to the settings for the areas of interest.
The most important features of GSC/VA-Class:
Intelligent real-time analysis of video data
64 areas of interest / areas of non-interest
Classification of persons and cars plus their characteristics as for example speed, size, direction and
status (fast, slow, in motion)
Three-dimensional detection of object size by parameterizable reference lines (definition of real distances in the picture)
Detection of abnormal activities and security-relevant situation
Automatic tracking of moving persons
Video Motion Detection (GSC/VMD) and Intelligent Video Analytics (GSC/VA-Class) run on one video channel each. Each system is parameterized separately and reports separately any activity in the video picture to the Dual Sensor analytics. A Dual-Sensor alarm is only generated if both systems (AND gating) report a detected motion simultaneously (in a time period to be set).