By Steve Ma, Director of R&D, Vivotek


Networking technologies bring to IP surveillance the benefits of remote access anytime, anywhere as long as an Internet connection is available. IP surveillance systems can easily be scaled up by connecting small systems together through IP networks, and can incorporate new technologies such as Power-over-Ethernet and wireless LAN. Because of their support for open standard Internet protocols, IP surveillance systems are compatible with a company's existing network infrastructure.

Among these benefits, superior video quality, especially for megapixel technology, is the most important driving force that will encourage continued migration to IP.

Compared with other types of network cameras, megapixel cameras capture more video details. It is thus expected that megapixel network cameras will become the main trend in IP surveillance for much more advanced applications in the future.

Wide Coverage
The factors that contribute to the explosive growth of megapixel network cameras are their ability to provide wider coverage and exceptional detail. A 2-megapixel network camera can cover an area 6 times larger than a VGA network camera. With a 2-megapixel camera compensating for 6 VGA cameras, the installation costs can be significantly reduced.

Exceptional Detail
When monitoring an area, a megapixel camera provides superior image quality compared to a standard resolution camera. As illustrated in the above illustration, which compares a D1, 720p HD (equal to 1MP resolution), and 1080p HD (equal to 2MP resolution) image, the license plate in the 1080p HD image can be easily identified, but not in the case of D1 resolution. Thus, in applications where accurate identification is required, a megapixel camera image can provide detailed information that is obscure when using a VGA camera. The improved pixel count also allows for the application of ePTZ technology, which will be discussed later.

Challenges
Although megapixel cameras possess amazing growth momentum, the challenges of bandwidth, storage, and CPU loading must be addressed before they can truly

During transmission, a megapixel image takes up much more bandwidth than a standard VGA image due to its large file size. It also requires more storage space, and as a result, customers have to expand their bandwidth and storage space, increasing their installation costs.

Before transmission, a megapixel image must be encoded, and when it arrives at the back-end PC or server, the CPU has to decode and resize the images for live viewing. Due to the high pixel count and large image size, encoding and decoding can significantly increase CPU loading, leading to the possibility of system breakdown or reduced system performance.

To thoroughly counteract these problems, a total solution must be designed that involves both bandwidth and CPU loading management. Some IP surveillance vendors address these issues with the promotion of H.264, a compression technology featuring high bandwidth efficiency. However, this only solves part of the problem. With H.264, the file size of a megapixel image can be reduced dramatically by 90%, resulting in significant bandwidth and storage savings. Unfortunately, the problem of CPU loading still exists.

Advances in megapixel surveillance
An IP surveillance setup includes network cameras on the front end and central management software on the back end, with an IP network connecting both parts. The network cameras capture and encode images and transmit them over the Internet in the form of video streams. When the central management software receives the video streams, they will be displayed on a device for live viewing and stored in a recording device. Video streams for live viewing must be decoded and scaled to the proper size before they can be displayed, thus increasing CPU loading.

For efficient bandwidth use and CPU loading management, new technologies has arrived including cropping and ePTZ for simplified image capture, H.264 compression for encoding, and activity adaptive streaming and multiple streams for video streaming.

In many cases, camera images end up containing unnecessary details such as still backgrounds. Therefore, transmitting the full view in megapixel resolution with the redundant data can be a waste of bandwidth and storage space. Cropping functionality allows users to crop unnecessary information and simply transmit video of the target region for viewing or storage. As a result, bandwidth and storage requirements as well as CPU loading can be drastically reduced.

Further enhanching the advantage of megapixel cameras, ePTZ, also known as electronic pan/tilt/zoom, enables users to select a target region for close-up shots by simply clicking on the camera video feed on their screen. By encoding and transmitting simply the image of the target region rather than the entire picture in megapixel resolution, ePTZ allows for more efficient bandwidth usage and CPU management. The electronic pan/tilt/zoom functionality also prevents a megapixel camera from mechanical wear and tear since it contains no moving parts.

H.264 Compression
Another way to utilize bandwidth more efficiently is to use compression technology with a higher compression ratio. MJPEG and MPEG-4 are currently the main compression standards for IP surveillance, but the newly developed H.264 standard will soon overtake MJPEG and MPEG-4 because of its superior compression efficiency.

H.264 is a high performance video compression standard that boasts a much higher compression ratio than MJPEG or MPEG-4, drastically reducing file sizes and conserving valuable network bandwidth. With a 90 % reduction in file size, a 2MB image can be drastically reduced to 20KB with H.264, a 50 % reduction in bandwidth or storage requirements compared with MPEG-4. As such, uncompromised image quality and less required bandwidth and storage space make H.264 ideal for megapixel cameras.
At a fixed bandwidth, an H.264 stream can be transmitted in higher resolution compared with MJPEG and MPEG-4 streams. For example, the same bandwidth can transmit a H.264 video stream in full-HD 1080p (1920x1080) resolution, but only an MJPEG stream in CIF (352x288) and MPEG-4 stream in HD 720p (1280x720). As a result, H.264 can improve the viewing experience without increased bandwidth requirements.

On-board storage
Back-end recording devices such as PCs or NVRs are widely used in daily surveillance. However, transmitting video streams containing no changes to back-end storage devices for continuous recording will consume a great deal of bandwidth. To reduce bandwidth requirements, a more efficient method is to store video images in the camera, such as on a SD/SDHC card, and have them transmitted only when an event occurs or when the operator needs to access the recorded data. The network is then only used by streams for live viewing, event-triggered recording, or backup. On-board storage allows for continuous recording while ensuring more efficient usage of bandwidth resources and storage space. It also guarantees constant recording, even when the network is disconnected.

With rapid advancement in memory card technologies and capacity, storage on the front-end on the camera itself will become a major trend. An SD card with a maximum capacity of 4GB announced in 2000 can only save snapshots, but the 32GB SDHC card launched in 2005 can continuously record 1Mbps video images for three days. 2009 saw the development of the 2TB SDXC card, which can store 1MB images continuously for up to 6 months and is large enough to fulfill the majority of consumers' demands.

Activity adaptive streaming
During normal monitoring, there is no need to receive high-definition images, only recognizable ones. As a result, transmitting a megapixel image at a high frame rate can prove to consume unnecessary bandwidth.. However, during event-triggered recording, high quality images and smooth video streams are necessary; in this case, a high frame rate is needed.

Activity adaptive streaming is a technology designed to use bandwidth in a smart way. It allocates bandwidth usage dynamically with a configurable frame rate according to the importance of the content. For example, during normal monitoring, the frame rate can be set low, e.g. 1 fps, to prevent video streams from taking up bandwidth. In an event-triggered situation, the frame rate will increase to a higher level, e.g. 30 fps, to allow for smooth and high quality video feeds. Activity adaptive streaming can optimize bandwidth usage during monitoring while ensuring superior image quality during recording.

Multiple streams
Imagine how much workload the CPU would be burdened with if video streams could only be transmitted at full frame rate or in megapixel resolution. Besides the stream for storage, the central management software needs to process the megapixel video stream for living viewing at the same time, including decoding and resizing images to fit the display setting. This can be a big drain on CPU power. If a stream is encoded by H.264, it will consume even more CPU power because of the complexity of H.264.
Usually, users can make do with CIF images for live viewing; only when an event occurs will they need to receive higher resolution images or video streams at full frame rate. Images for recording, on the contrary, must be of high quality at all times. In order to address the two different demands, a camera must support multiple streams.

Multiple streams allow each video stream to be delivered in a different resolution, frame rate, and image quality for individual quality or bandwidth demands. As a result, the camera can simultaneously transmit a small image in CIF format for real-time monitoring and a large megapixel image for storage. The CIF image can be directly displayed on the screen without much decoding or further scaling, thereby drastically reducing CPU loading. In addition, because different devices such as PCs and mobile phones have different requirements for image sizes, resolutions, and frame rates, multiple streams give users a higher level of flexibility for dealing with camera images on different platforms.

Due to the difficulty in transmission, recording, and playback of megapixel video streams, network cameras must be flexible enough to deliver optimized streams for specific applications so as to avoid system overload. This is where multiple streams come in. It is expected that multiple streams will ultimately be a standard specification for megapixel surveillance.

Author background
With a B.S, M.S, and Ph.D degree in Electrical Engineering from National Taiwan University, Steve Ma is the Director of R&D at VIVOTEK. He joined VIVOTEK in 2000 and has accumulated solid sales management, business development, product definition and design experiences in the IP surveillance industry as well as broad product knowledge from semiconductors to complete solutions.