IP cameras have made enormous gains in image quality, resolution and frame rate in a very short time. Today, it is not uncommon to see video resolutions of 20 megapixels or greater at frame rates of 15 to 30 fps.
IP surveillance architectures rely on distributed processing to encode and compress video from raw formats captured by the camera imager into streams that pass across common Ethernet networks.
Every camera is an encoder, and as you add more cameras, you naturally need more encoding capacity. The question is, how do you scale up decoding capacity to view the rapidly growing population of cameras?
The Problem: Lack of IP Decoding for Video Management Servers
Video Management Servers (VMS) are generally designed to receive the encoded video from cameras and write the video data as a file directly to disk, without the need to decode the video. This is ideal for storage, as the files remain compressed and the CPU/GPU of the receiving VMS is not taxed with much processing.
VMS design has traditionally been premised on forensic, post event, evidence storage and review in order to prosecute criminals or avoid trip and fall liability lawsuits. Generally VMS systems assume a very limited number of live cameras being simultaneously viewed. For this reason, scalable IP decoding has not been a focus for VMS manufacturers.
However, as IP cameras proliferate, their utility for purposes such as safety monitoring, industrial operations monitoring, traffic monitoring and general situational awareness increases. Security operation centers, traffic management centers and operation management centers increasingly demand more cameras to be displayed. But traditional VMS architectures are not up to the task.
The Solution: Ultra High Definition Video Wall Processors
A modern x86 architecture PC with a contemporary GPU is capable of decoding approximately 150 frames per second of 1080p (2 megapixel) H.264 compressed video. This represents only five video streams at 30fps, and only one stream at 4K (8 megapixels).
A common trick of the trade is to encode two streams at each camera, one for recording at full frame rate and resolution, and one at lower resolution and frame rate for live viewing. But this defeats the purpose of using a megapixel camera. Why compromise the quality of the video that the operator can see?
Related: Orlando Magic Bring Hybrid IPTV’/tag/RGB_Spectrum’>RGB Spectrum‘s MediaWall 4200 processor as each facility’s nerve center, feeding two video walls. One wall is L-shaped, comprising a 12×3 array of narrow bezel Samsung 46” LCD monitors, each with a resolution of 1366×768. The second wall is of similar size, but U-shaped.
Each system receives, manages and monitors video streams routed from 12 specialized NVRs (Network Video Recorders). Every NVR, in turn, processes images from more than 1000 IP-based security cameras located around the city. In terms of system flexibility, each video wall can display up to 300 IP cameras simultaneously. Any camera image can be scaled to any size—up to filling the entire video wall. For simplicity, one operator can drive the video wall and control all 12 NVRs.