by Arecont Vision, LLC
The H.264 compression of megapixel CCTV cameras is perhaps the most important advance in IP camera technology since they were first introduced. Essentially, the H.264 allows non-parallel megapixel and high-definition cameras to use the same bandwidth and store the same as standard-definition VGA cameras. From the shift from black and white cameras to color cameras, the industry had not experienced such a major breakthrough in technology.
Although the migration to Ethernet video surveillance networks continues to grow and although new technologies and products offer more efficient means of achieving high-performance video without any compromises, security professionals should also familiarize themselves with the nuances of IP video to take the greatest possible advantage of the benefits offered by IP-based systems.
There are two issues related to IP video (and more specifically megapixel video) that need to be addressed: bandwidth limitations and the fact that high-definition digital video data requires greater amounts of storage and logging space. The essential thing to achieving the right balance is a single factor: compression. For megapixel video, this compression has been based on MJPEG for a long time. However, the emergence of the H.264 is significantly changing the demands of megapixel video on bandwidth and therefore storage utilization and, for this reason, the aforementioned limitations are already less than a factor.
Overview
Let's take a look at a typical scenario that one may face when considering a network CCTV solution (less compression means higher image quality and larger file sizes). If you increase the compression, you will have smaller files, but lower image quality. This has been an increasingly critical issue as conventional, high-performance megapixel IP cameras gain more traction for overall video surveillance. The criteria one needs to consider are as follows: image quality (compression), image size (resolution) and cropping, the number of images recorded or displayed per second (frame rate).
For example, let's compare two separate images put with the same image quality specifications that use MJPEG compression (a conventional VGA image with 640 x 480 resolution and a 3 megapixel image, with 2048 x 1536 resolution. Each image is transmitted at a rate of 15 frames per second. The 3-megapixel image stream can use between 18 and 25 Mbps, while the VGA stream can require only between 2 or 3 Mbps. If we start by adding cameras (maybe several hundred cameras) the bandwidth requirements increase rapidly. For systems with multiple megapixel cameras, it can be a challenge to transmit and store such large volumes of data without sacrificing, even with the proliferation of Gigabit Ethernet configurations and the rapid decline in the price of storage.
More efficiency
The commitment that is usually made when faced with this barrier of bandwidth and storage, with respect to the image of megapixels, is to lower the quality to solutions with lower resolution such as the D1, the VGA or even devices with lower resolution. However, the use of H.264 offers a higher compression efficiency between 5 and 12 times above average, compared to implementations of MJPEG. This depends on the amount of light in the scene, which produces more or less noise and the amount of movement in the scene.
Let's assume a conservative H.264 compression efficiency of eight times (a 3-megapixel video stream now results in a bandwidth requirement between 2.25 and slightly above 3 Mbps). This is similar to the bandwidth requirement of IP VGA cameras at the same frame rate using the MJPEG. In addition, the H.264's 3-megapixel flow still maintains the intrinsic benefit of 10 times the pixel density of the VGA. Of course, the bit rate depends on the movement in the scene, which could be caused by a PTZ configuration. However, most surveillance cameras are used in fixed applications with nominal motion. Even if the movement is not nominal there will be some increase in the required bandwidth; but over time H.264 will remain more efficient than MJPEG implementations.
Some H.264 cameras have bitrate control, allowing for constant bitrate tuning, which is achieved with varying quality to accurately predict bandwidth and storage requirements. The need to compromise resolution for bandwidth and storage is significantly reduced with the H.264, giving users one more reason to take advantage of the benefits of high-definition megapixel video.
Other times a reduction in the frame rate is made with the megapixel video in order to reduce the bandwidth. Although no commitments to reduce the speed of images are required with the current implementations of the megapixel cameras of the H.264. Megapixel video currently achieves the same real-time frame rates as VGA with almost the same VGA bandwidth requirements; example: 30 ips @ 1280 x 1024 and 24 ips at 1600 x 1200. Putting all the emphasis on security to emulate "real time", it is generally unacceptable to reduce video image speeds, especially for "alarm" conditions or in some applications such as casinos, where recording parameters are subject to legislation. The H.264 becomes more "efficient" with higher frame rates than with lower image rates, but this does not happen with MJPEG.
If it is unacceptable to reduce the image speed of the video many times the decrease in image quality (increasing compression) is used to compensate. However, increased compression increases the amount of unwanted noise and image effects. To some extent it will make the image unusable. It is a delicate balance. With the H.264 the desired balance is achieved, resulting in high compression, full image rates and high definition resolution, allowing forensic resolution to benefit from megapixel video with minimal compromise.
An additional consideration is the components of the system itself. Servers or computers have their own limitations, including the ability to process data quickly. This can be considered as its "internal bandwidth".
It is not correct to say that the CPU load has to be higher in order to decode H.264 versus MJPEG. The mathematical complexity lies in the coding of H.264, which is done in the camera. The complexity of decoding MJPEG and H.264 is similar. In fact, real applications show a reduced CPU load, due to smaller flow sizes under similar quality settings.
Software platforms for handling and recording video using H.264 decoding solutions must perform similar or even better CPU treatment with H.264 megapixel cameras, compared to MJPEG versions. Under this premise, the number of H.264 megapixel cameras handled by each server should be the same or better than with MJPEG devices.
Solution H.264
The latest introduction of H.264 compression for video surveillance is eliminating the degree of compromise between image size and resolution versus bandwidth and capacities required for storage (especially when considering the use of megapixel IP cameras).
Arecont Vision's implementation of the H.264 (MPEG-4 – Part 10), offers up to twenty-five times greater compression than conventional MJPEG megapixel devices, along with higher resolution. The introduction of the new HD megapixel cameras with H.264 video compression really improves the cost effectiveness in relation to storage and bandwidth.
Although the implementation of H.264 is complex, it is recently being used in commercial encoders for radio broadcasts and consumer-based products. The challenge for the company was to offer H.264 technology at a low cost to finally decrease the price of megapixel IP cameras to the level of standard definition VGA IP cameras that are strongly penetrating the current security industry. Fortunately, some cost reductions are inherent in the H.264 standard itself (its overall complexity is distributed over a large number of operations that can be combined to your satisfaction and adjusted to a given set of application requirements and budgets).
Arecont Vision's advantage in achieving this goal is in the world-class expertise in the field of FPGA-based hardware image processing. More specifically, the H.264 encoder implemented in a single FPGA offers 80 trillion operations per second (a feat that would require 25 Pentium processors). This results in better compression (reduced flow size) with high video quality in a typical road surveillance scene, with vehicles and people in motion, which is up to twenty-five times higher than the MJPEG. This level of compression would be impossible in a low-cost implementation without a nonlinear temporal pre-encoder of its own. As a result, a high-quality, 2-megapixel video stream at 24 ips or a 5-megapixel video stream at 9 ips can be as low as between 1.5 and 2.5 Mbps.
The new cameras of 1.3, 2, 3 and 5 megapixels H.264, can be taken to other proportions, such as resolutions of 1920 x 1080 or 1280 x 720 for a high definition format of 16 x 9. The line supports multi-stream capabilities for streaming video streams in different video formats, image rates, bit rates, and resolutions. The resolution is fully controllable, which makes it possible to achieve full resolution or less to conserve more bandwidth and storage, and achieve a total display field with a lower resolution, preserving the regions of interest with a full resolution of megapixels.
Arecont's H.264 megapixel cameras offer progressive full motion recording HD resolution, along with a variety of additional features, including multizonal motion detection. The new cameras also support the RTSP protocol for direct streaming of H.264 video in third-party software operators, such as Apple's QuickTime, Windows Media Player or VLC Player, as well as custom SDKs for integration into third-party management platforms.
The trend is changing as we see more professional video surveillance systems shifting to an IP-based platform. This is largely due to advances such as the H.264 HD megapixel cameras. In addition to a marked advance in overall performance, the new H.264 megapixel cameras, and related devices, are a major relief from previous sacrifices between bandwidth utilization, flow capabilities, file size, and image quality for large-scale recording and transmission.
