Introduction
A link in this month’s Tech. Brief – Why the market has already tipped to IP – discusses the shift in the video surveillance user market from CCTV technology to IP video surveillance camera technology, in the international market, and the reasons for the shift. While such a trend may not be as pronounced in India, it is undeniable that many new requirements for video surveillance infrastructure specify IP and not CCTV.
The advantages of IP video surveillance are well known and well documented. However, the drawbacks of IP video surveillance too are well known: the increase in storage space and the increase in transmission bandwidth. These drawbacks are further amplified, in end-user consideration, with the increasing introduction and greater commoditisation of megapixel IP video surveillance cameras in the market.
This brief will look at how to address the storage and bandwidth issues of IP video surveillance cameras through the management of the a camera’s frame-rate.
What is an acceptable Frame-rate?
The standard frame-rate for video systems is 25 fps (PAL) or 30 fps (NTSC). The human brain processes images perceived by the eyes at 24 fps as fluid motion. As the frame-rate keeps dropping the brain registers stutters to the fluid motion, with the stuttering becoming more pronounced at 10 fps and below.
The current user trend, with respect to IP video surveillance infrastructure, is to try to get as much megapixel as possible, for a camera. The minimum resolution required by users is 4CIF (704 x 576). However, HD (High-Definition) resolutions are also gaining traction among users: 1280 x 720 (720p) and 1920 x 1080 (1080p).
A typical 1080p resolution video surveillance camera – approximately 2 MP (megapixel) – requires a bandwidth of 7 – 8 Mbps to output a stream of video at 30 fps, using the H.264 video compression codec (Base Profile); and a bandwidth of 4 – 5 Mbps for the same stream, using the H.264 Advanced Profile. For more information on how to calculate the bandwidth requirements for a given camera specification, please refer the July 2009 Tech. Brief: Video Surveillance Bandwidth Requirements – Calculation of Utilization.
A user is usually loath to compromise on video quality, beyond a certain level, as would be required if the bit-rate of the camera was throttled using the camera settings. This leaves the frame-rate as the only camera parameter that can be adjusted, without significantly impacting the quality of the video stream.
The bandwidth savings that adjusting the frame-rate can deliver is significant. As mentioned earlier, a 1080p video stream at 30 fps, compressed using H.264 (Base Profile), requires a bandwidth of 7 – 8 Mbps. Adjusting the frame-rate down to 15 fps brings down the bandwidth requirement to below 4 Mbps. Further reducing the frame-rate to 5 fps sees the bandwidth requirement come down to just over1 Mbps.
Given these savings in bandwidth requirement, and concomitantly storage space, how does the user decide on the optimal frame-rate for a surveillance installation? There are four considerations to keep in mind when deciding on the frame-rate to set for a video surveillance installation:
Scene under surveillance
A scene involving fast-moving objects will benefit significantly from a higher frame-rate; with, furthermore, an inverse relationship between the fps parameter and the size of the moving object. A scene involving a slow-moving object (say, a pedestrian) will look fine even at 5 fps, while a scene involving a car travelling at a reasonable speed will look jerky at 5 fps. There are certain industry guidelines for deciding on frame-rates for specific settings:
 |
Casino tables – 30 fps |
|
Cash counting settings (bank tellers) – 12 to 15 fps |
|
Vehicular traffic (not at traffic stops and parking lots) – 15 fps |
|
Pedestrian traffic – 5 fps |
|
Parking lots, traffic cameras, overview scenes – 1 to 3 fps |
|
Low-activity settings – 1 fps |
Movement in relation to camera frame
As a further qualification to the point about the “type of scene under surveillance”, the movement of the object within the camera frame also has an impact on the optimal frame-rate. Movement along the vertical plane of the camera frame (i.e. the object moves towards or away from the camera) will allow for a lower frame-rate setting than movement along the horizontal plane of the camera frame (i.e. objects moving across the camera’s line of vision).
Camera Sensor features
The quality of the camera sensor will also play a part in determining how low a frame-rate a user can drop to, for a given surveillance scene. The key camera sensor features that affect the quality of images, at low frame rates, are: scanning (progressive vs. interlaced) and shutter-type (global vs. rolling). A progressive-scan image has less blurring than an interlaced-scan image, while a global-shutter image has less blurring than a rolling-shutter image.
Available Light
Finally, the available light during a surveillance recording can also impact the optimal frame-rate setting. This is because some camera vendors configure their cameras for low shutter-speed as ambient illumination drops. A lower shutter-speed results in blurring of moving objects, and dropping the frame-rate in such a scenario could easily result in unacceptable image quality. Under the circumstance, it is advisable that users implementing low frame-rate plans ensure that the ambient illumination, in the surveillance area, is always constant.
|
||||
| Figure 1: Frame rates for a video surveillance setting and the impact on perceived video quality – 5 fps, 10 fps, 15 fps, and 30 fps |
Other Factors to consider
In addition to the above considerations, some camera vendors offer ways to circumvent the frame-rate/storage-space conundrum, if not the frame-rate/bandwidth conundrum, through the use of features available on the camera software.
Setting different frame-rates for different situations
The VMS will allow the user to set different frame-rates depending on whether activity has been detected in the area under surveillance. The user may opt to carry out no recording when there is no movement, and to record at the full frame-rate whenever there is any movement. Although such a setting will have no bearing on the transmission bandwidth that will need to be planned for the surveillance infrastructure, it will have a major impact on the secondary-storage capacity planning.
Local viewing at a higher frame-rate and recording at a lower frame-rate
Since local viewing is on the local (wired) network, where bandwidth is not rationed, cameras that support multi-streaming will allow a user to view (locally) at the full frame-rate while recording at a lower frame-rate. Multi-streaming is the feature by which a video source generates multiple streams of the same video stream at different image quality (resolution, frame-rate, compression codec).
Conclusion
Multi megapixel IP video surveillance cameras are rapidly gaining favour with users, on account of the perceived benefits and falling cost. In such a situation, optimizing network bandwidth and storage space becomes important. This brief discusses some of the methods that can optimize bandwidth and storage-space, without sacrificing significantly on image quality.
Mistral offers a selection of IP video surveillance cameras, with resolutions ranging from D1 (4CIF) to multi megapixel. These cameras support advanced features that will allow a user to optimize the use of valuable network bandwidth and storage-space while delivering superior image quality.
