Camera DNR (Digital Noise Reduction) GuideBy Ethan Ace, Published Nov 20, 2014, 12:00am EST
Bandwidth spikes are a significant video problem.
An IPVM study found 250 - 500% increase in bandwidth from day to night (see: Testing Bandwidth vs Low Light).
Digital noise reduction is key for reducing night time bandwidth and preventing dramatic spikes which can kill your surveillance storage. In this guide we answer three key questions:
- How much of an issue are bandwidth spikes?
- How much does DNR reduce bitrate?
- What impact does DNR have on image quality?
In order to answer these questions we tested 11 cameras from eight manufacturers in an approximately 1 lux scene, adjusting DNR to various levels from off through maximum, while measuring bandwidth.
- Arecont Vision AV3116DNv1
- Avigilon 1.0-H3-B1
- Axis Q1615
- Bosch NBN-733V
- Bosch NBN-932V
- Dahua IPC-HF3100N
- Hikvision DS-2CD864FWD
- Samsung SNB-5004
- Samsung SNB-6004
- Sony SNC-VB600B
- Sony SNC-VB630
What is DNR?
Digital noise reduction, as it name states, aims to reduce digital noise present in low light surveillance images.
This noise comes from automatic gain control. Gain is an essential tool for delivering any image in low light surveillance, as shown in this comparison of varying gain:
However, there are 2 important side effects:
- Visible noise on the image that obscures fine details (which is still better than the alternative without gain - a completely dark image)
- Increase bandwidth consumption because the encoder sees the visible noise as moving objects that are more difficult to encode.
Digital noise reduction are image processing techniques that aim to eliminate the visible noise to improve the image quality and to lower the bandwidth consumed in encoding.
The two commonly cited types of DNR are 2 dimensional and 3 dimensional DNR. 2 dimensional processes each individual frame / image individually, while 3 dimensional processes across a series of frames over time.
The major cost is having the computing power on-board the camera to do this processing.
The major risk is blurring objects in the image when trying to reduce noise (shown in depth at the conclusion of this guide).
We tested 11 cameras from seven manufacturers. Of these, only three (Avigilon, Axis, Dahua) did not allow any control of digital noise reduction. We tested all cameras in the same scene, our interior conference room, to see what effects different DNR settings had on bitrate.
DNR On vs. Off
In most cameras tested, bandwidth dropped by an average of ~70% when turning DNR on at default settings. Only a single camera, the Samsung SNB-5004, saw reductions below these levels.
DNR Defaults vs. Max
The effects of moving from default settings to the maximum offered by the camera varied widely. In some cameras, such as the Arecont AV3116DNv1 and the Samsung SNB-6004, this increase in DNR had almost no effect. However, others saw much greater reductions, with the highest being the Hikvision 864 (~90%) and the Bosch NBN-733V (~75%). Note that increasing DNR settings to high levels may result in blur, shown below.
Applying high levels of digital noise reduction is not without drawbacks, as it may introduce motion blur which much resembles the blur created by slow shutter. This blur is mainly caused by temporal noise reduction, which compares changes in the scene between frames in order to determine what is noise and what is not.
For example, the video below shows a Samsung SNB-6004 at maximum DNR settings, with pronounced blur and ghosting as our subject walks through the scene.
Compare this to lower settings such as the defaults in the video below. Blur is still present, though drastically reduced.
Finally, with DNR off, visible noise increases dramatically, but blur is eliminated:
Test your knowledge
Take this 6 question quiz now
19 reports cite this report:
Back to Top