Surveillance Systems

or by third parties. Some manufacturers also provide a full SDK for developing custom applica-tions in various programming languages. For developing smart camera applicaapplica-tions the following limitations should be considered:

• Limited electrical power: The power coming from the power supply over the Ethernet cable is limited to maximum 15.4 W (or 25 in case of the PoE+ standard). This limitation signi-cantly restricts the number of devices (e.g. memory modules) and the computing performance of the processors used in smart cameras.

• Real-time processing: The image processing and machine vision algorithms require real-time performance. However, the processors of smart cameras have less processing power and functionalities (e.g. parallel computing) than modern PC architectures (e.g. multi-core CPU, parallel GPU) can provide.

• Constrained memory: The amount of memory of a smart camera is limited, that is storing large numbers of images and image processing results in the memory may lead to out of memory errors.

1.1.5 Surveillance Systems

Early systems used only analog techniques for communication and data storage. In the rst systems the output of the analog cameras was connected by a coaxial cable to a VCR for recording the camera images. The raw video signal was recorded on the tape without any compression. In case of multiple cameras a multiplexer or a quad was integrated into the system. The quad creates one output image from the images taken from e.g. four cameras by simply merging the downscaled images. A multiplexer takes the images from several cameras and combines them into a multiplexed video stream, which is usually performed by using time-division multiplexing (TDM) to output only one video channel at a time. A fully analog system is demonstrated in Fig. 1.1.

As a result of technical improvements, the analog VCR of the surveillance system was changed to a digital storage solution, hence partly digital systems appeared. The Digital Video Recorder (DVR) is a device capable of recording the images of analog cameras in digital format.

Instead of a videotape, DVRs use hard drives to store the video data. Therefore, the analog video signal is digitized, and then compressed to achieve ecient storage. Most DVRs have multiple video inputs and have multiplexing or quad functions integrated.

Improvements of network technology aected the surveillance systems and Network Video Recorders (NVR) have become commercially available. These devices have all the functionalities

1.1.5.1. Visual Management Systems 13

Multiplexer/Quad

Analog monitor

VCR

Analog cameras

Analog coaxial cable Analog coaxial cable

Figure 1.1: Architecture of a fully analog surveillance system.

of a traditional DVR, but they are additionally equipped with an Ethernet port for transporting the stored recordings or the live view over the IP network. Usually a custom client is provided by the manufacturer for accessing the video, but many systems can be used with standard web browsers. Fig. 1.2 demonstrates a partly digital surveillance system.

Recent technological improvements have enabled manufacturers to build fully digital surveil-lance systems. In such systems the communication is IP-based and ecient compression techniques are used for video data transmission and storage. The main advantage of a fully digital system is that standard network devices can be used to build the infrastructure and an ordinary PC can be used both for managing the camera network and for storing video recordings. Analog cameras can be easily integrated into the system by using video servers (see Sec. 1.1.2.4). Fig. 1.3 presents the architecture of a fully digital system.

1.1.5.1 Visual Management Systems

The large amount of video and audio data produced by the cameras of a modern IP-based surveil-lance network requires a video management system to store, process and retrieve/playback the data automatically. Most of these systems are either built on PC platform or by using an NVR.

Analog cameras

Analog coaxial cable

DVR

IP network IP network

Switch

PC

Figure 1.2: Architecture of a partly digital surveillance system.

1.1.5.1. Visual Management Systems 14

IP cameras IP network

Switch

PC

NVR IP network

Analog cameras

Analog coaxial cable

IP network

Video server

Figure 1.3: Architecture of a fully digital surveillance system.

PC platforms involve PC servers and other computer data storage solutions (e.g. Network-attached Storage) to achieve a high performance system. The main advantage of this approach is that it can be easily extended by additional functionalities (e.g. increased storage, device upgrade/expansion, automatic video processing, or security improvements). Hence, the system becomes fully scal-able and exible. Network video recorders have built-in management functionalities and extension is usually not allowed. The main advantage of an NVR-based management system is the easy installation process. A typical video management system has the following functionalities:

• Viewing: To provide a user interface for viewing the recorded and live video of multiple cameras. Some systems have multi-screen viewing functionalities.

• Recording: To record and store videos manually, either continuously (scheduled) or triggered by events.

• Query/retrieve: To query (e.g. by date or by event) the recorded videos and to display the query results.

• Event management: To create and identify events triggered by external entities (e.g. barcode reader, camera tampering or motion detection software). For the dierent events the sys-tem can produce dierent responses (e.g. video recording, audio/siren alarm, SMS or e-mail notication).

• Camera management: To provide a user interface for conguring the camera settings (e.g.

frame rate, resolution, compression format, or PTZ preset positions).