Many within the surveillance industry are deploying IP video surveillance cameras and networked recorders using the same design and engineering strategies used for building analog CCTV camera and DVR-based systems. On the surface this makes sense: surveillance is surveillance; the fundamental optics and geometry remains the same regardless of the medium. What isn’t the same is how the IP systems operate under day and night conditions.
As H.264 adoption rates increase, many security executives are finding that during the night recording activity and, consequently, storage consumption increase significantly. The reason isn’t due to activity in the field of view; it is because of the way H.264 works and the way surveillance recorders use video motion detection to determine when to record.
All compression – H.264, MPEG-4 or MJPEG – faces exactly the same challenge: when there is less light, the image will get noisier and cause the compression to be less efficient. Analog systems do the same inside the DVR. Fortunately for IP users, new technologies offer the flexibility to cope with the situation.
H.264 vs. MJPEG
In the past, MJPEG was the predominant recording format, each frame was a full picture of the scene, and bandwidth was consistent between frames regardless of motion. H.264 video streams consist of one full picture every second called an “I” frame, separated by a series of “P” frames, representing changes against the last “I” frame. This is how H.264 streams use up to 80 percent less bandwidth and storage compared to MJPEG. However, in low-light scenarios, H.264 camera bandwidth and storage consumption can double, reducing the savings compared to MJPEG. As a camera switches to night mode, the camera’s imager increases the gain to compensate for the low lighting which increases the image “noise,” often seen as static in the video.
Further compounding this H.264 issue: many network video recorders use video motion detection as the single means of determining when to record. A typically designed system will record less than 15 percent of the daylight hours due to motion; however, at night, this can skyrocket to more than 90 percent. The reason is again due to an increase in the gain. This creates a lot of noise in the image, which collectively is seen as substantial motion by the video motion detection system. This tells the NVR to record the H.264 video, and now the increased bandwidth in the camera stream becomes two times the storage of the daytime video as well.
Mega Pixel Multiplier
A key enabler of H.264 adoption has been the increased demand for megapixel and HD cameras. Previously, with MJPEG the storage and bandwidth consumption was cost-prohibitive for high-resolution imaging technologies. Take, for example, a 1-megapixel camera using MJPEG at 10 FPS with zero compression. This generates 16.4 megabits per second in bandwidth, compared to 2.5 megabits per second with H.264 at the exact same settings. This is an 85-percent savings in bandwidth and storage. So, many more megapixel cameras are being used. In addition, megapixel cameras have less light sensitivity than standard resolution analog cameras, which increases the amount of time the cameras are operating in a “noisy” mode. This situation contains factors which lead to unexpected storage and bandwidth consumption: Increased numbers and resolution of megapixel cameras, increased hours of “noisy” operation and recording during low-light periods make the gap between day- and night-recording requirements even bigger.
More Cameras and Longer Retention Times
Finally, two more factors are making an impact on storage strategies. IP surveillance products are reducing total system cabling cost, which allows customers to shift budget resources to invest in more camera locations than was possible with coaxial and fiber optic connected cameras. Second, security departments are expected to retain video surveillance recordings for longer periods of time, exacerbating the need for a new strategy.
Same Problem, New Strategy
Balancing day- and night-surveillance requirements isn’t a new problem; however, the impact of not acknowledging the difference is now seen: systems are consuming significantly more storage by recording non-events at night. The compounding effects of H.264, the percentage of motion recording, and megapixel cameras deployed in the CCTV analog way (without adjusting the default camera parameters) results in a “noisy” surveillance operation.
Small changes can go a long way in preventing this. Adding devices like PIR or IR illuminators can dramatically reduce the amount of storage consumed at night by increasing the accuracy of motion detection during low light hours. Developments in today’s latest IP cameras and in-camera chip technology enable much better noise reduction capabilities inside the camera so as not to negatively affect storage. Efficiently designed video analytics can also filter out “noise” in the scene by triggering recording only on events that are object-based.
The bottom line: Test and tune the system during both modes of operation so you’re not surprised by the amount of nighttime and low light storage.