Atsushi Nogami, Team Leader of the Business Planning Team, Security System Division at Panasonic, presented some existing hurdles when implementing IP video surveillance solutions at the Global Digital Surveillance Forum, part of SecuTech Expo 2007 in Taipei, Taiwan.
The digital era allows for enhanced surveillance options, which are expanding. IP video offers many benefits, such as less materials and labor, ease of adding cameras and use of existing Ethernet cable installations. This convenience allows for flexible camera placement and relocation throughout an installation. This is called "Reducing TCO." The flexibility of the cabling is really the advantage of the IP camera. This makes IP solutions practical, cost-effective and convenient. One of the main benefits of digital surveillance is remote monitoring, even in long-distance situations. By overcoming distance, efficiency for operation cost is increased, with centralized monitoring and system management via WAN, wireless or VPN.
IP Video: Recent Improvements
Improved video compression has helped make digital video surveillance a reality. The developments of MPEG-4 and H.264 compression formats can provide smaller data sizes compared to JPEG, allowing more cameras in the same network. It also allows for easy installation of additional cameras on the network, reducing the costs for new cables and labor. The basic performance of the IP was lower than analog cameras used to be. This inhibited the penetration of IP cameras. However, this problem has dramatically improved, thanks to newly developed CCDs and DSPs. Recent IP camera sensitivity is becoming almost equal to analog camera sensitivity. In addition, the dynamic range of IP cameras has improved a lot, after incorporating SD III into IP cameras.
IP Video: Exceeding Analog
Now, some of the functionalities of IP cameras are getting better than analog. One main advantage of IP video surveillance over analog surveillance is Power over Ethernet, or PoE. By using this method, it can reduce cabling and material costs. This allows Ethernet cables to not just carry data (video) but power as well, allowing cameras to be added or moved easily. Progressive scan provides clear images and higher dynamic resolution, which are necessary for video analytics to work properly. The last example and the most critical example which exceeds analog video is megapixel technology. Analog video is limited by TV standard NTSC, with 525 lines, and PAL with 625 lines.
Therefore, increasing the number of pixels in the CCD beyond a certain level was not meaningful in analog. But, IP video does not need to stay in the TV standard. Megapixel cameras offer a larger frame with higher picture resolution to effectively monitor a larger area. This image was taken with the WV-NP244 camera, which is well-known in the industry for good picture quality. Using megapixel cameras, users have a wider area, with higher definition. Why use surveillance systems? They are for identification, which is a fundamental use, to identify people or a license plate. Megapixel cameras can do it. Analog cannot. This is a big advantage of IP.
The True Impacts on IP Video Expansion
As IP has had so much progress and improvement, intelligent video has become a recent topic. There is "left object," "crossing wire," "facial recognition" or "face detection," and "people counting. Some of them have real uses now, in certain applications. However, there are still a couple of issues to be addressed for ¨intelligence〃 to become widely used in the market.
Intelligence Needs to be on the Camera Side
In an IP system, if we try to generate ¨intelligent data〃 or what is normally called "metadata〃 on the management PC side, real-time simultaneous decoding and signal detection from multiple camera data streams by one PC are required. This is a large CPU load. Even if a Zeon double CPU with a 3 GHz processor machine was used, the maximum number of cameras that could be processed would be four to five units. For demonstration purposes, this would be OK, but not for real applications. With the normal network supporting 50 to 100 cameras, this approach is not practical. So, an intelligent engine needs to be distributed in the camera or encoder side, called an edge device.
Intelligence is in the Camera DSP
Installing ¨intelligent software〃 onto standard DSPs such as the Texas Instruments DM 642 or other DaVinci type DSPs is becoming used practically today. We are seeing this type of approach from several manufacturers. But, additional DSPs becomes an additional cost to the camera and it would limit application. From a historical perspective, real technical breakthroughs are providing additional value without cost increase, or even a decrease, being widely used and making that value a standard. In this scenario, intelligence needs to be embedded into camera DSP, or I should say the new camera DSP, which can handle not only video signal processing and encoding but generate intelligent data (metadata).
This type of new DSP needs to be developed. When this type of DSP becomes available, intelligence can be provided without additional cost and be widely used. Besides these requirements for intelligent video to be widely used, the most critical issue is picture quality. We are now seeing a certain progress of image analysis algorithm. It is probably becoming a usable level.
But, the most critical issue is picture because intelligence is a result of picture analysis. By having intelligent video, people expect improved productivity, more efficiency and better identification. But if the picture is bad, the operation would become a disaster, with too many false alarms and missing critical incidents. So this is called, "Garbage in, garbage out!" Continuous improvement of the picture quality is the key for the success of intelligent video. Good picture good intelligence. Bad picture worse than no intelligence, I would say.
Key Points for Intelligence
IP video is expanding based on common IP advantages, with recent improvements exceeding analog performance. Does intelligent video impact IP video expansion? There are several conditions. One, an intelligent engine needs to be an edge device. Two, an intelligent engine need to be in the camera DSP to eliminate additional cost. Lastly, the most critical issue for intelligent video is continuous improvement of the picture quality.