Digital Noise Reduction (DNR) Guide

By Ethan Ace and Ilya Ignatenko, Published Jan 07, 2022, 08:38am EST

Visual noise spikes camera bandwidth usage in low-light. Digital noise reduction (DNR) aims not only to reduce noise but to improve video quality and minimize bandwidth usage.

IPVM Image

In this 10-page guide, we explain the effects of digital noise reduction on image quality and bandwidth consumption, as well as noise reduction methods used in video surveillance, examining:

  • Digital Noise Reduction Basics
  • DNR Methods
  • Spatial Digital Noise Reduction
  • Temporal Digital Noise Reduction
  • 2D vs. 3D Noise Reduction
  • Manufacturer DNR Controls
  • DNR Image Quality Improvements
  • DNR Motion Blur Issues
  • Camera Model DNR Motion Blur Performance
  • Motion Blur Vs. Noise Tradeoff
  • DNR Bitrate Reductions
  • Smart Codec Impact On DNR

Key ******

***** *** **** *** takeaways **** **** *****:

  • ***** *** *** *** methods ** ******* ***** reduction (***), ******* (****** frame) *** ******** (*****-*****), and **** *** ***** combined *** ******** ** as "** ***."
  • *** *** ******* ***** quality *** ***** ********* by ****** ******** **** efficient, *** ******** *** much *** ***** ****** blur ** ****** ******.
  • ***** ** **** *******, most *******' ******* ******** off *** **** *****-*** of *****, ***** *******, bandwidth, *** ****** ****, but **** ** ******* in **** ******* ****** (see***** *** ***** **** Shootout*** **** *******).
  • ******** ******** *** ******** or *** ******* **** reduce ****, *** *** increased ***** ***** ******* any ***** ******* ***** on ****** *******.
  • ** *** *****, *** (at ******* ********) ******* bandwidth ** **-**% ** night ******** ** ********* it ** ******* ** to ******* ******.

** ****** *** ****** of *** *** **** of ***** ******* ** detail *****.

Digital ***** ********* (***) ******

** ***** ****** ********, cameras ******* ** **************** increase *** ********** ** images ************** **** *******.

*******, ***** *** * important **** *******:

  • ******* ***** ** *** image **** ******** **** details (***** ** ***** better **** *** *********** without **** - * completely **** *****)
  • ******** ********* *********** ******* the ******* **** *** visible ***** ** ****** objects **** *** **** difficult ** ******.

******* ***** *********, ** its **** ******, **** to ****** ******* ***** present ** ***-***** ************ images. *** *******, ** the ***** *****, *** subject ** *** **** is ***** **** *** disabled. ***** ******** *** face *** **** ** the **** *****, ***** can ****** ** ****. However, **** *** *******, his **** *** ** recognized, ******** ******* *** be ****, *** ******* lines ** *** ***** read.

IPVM Image

Methods ** ***** *********

*** *** *********** ***** of *** *** ********* referred ** ** ******* and ********. **** ***** to ***** ** "**" and "**."

Spatial ***** *********

******* ***** ********* ******** details ** ******** ****** in *** **** ***** to ********* **** ** noise *** ******** ***** values ** ****** *** image. **** ********* *** reduce *** ********* ** the ***** ** **** levels, *** ** ******** less ************* ***** *** provides ***** ******* **** temporal **********. *******, ******** DNR ********* ****** ****** noise ********* ** ***-****** images **** *******.

Temporal ***** *********

******** *** ******* ******* the **** ****** ******* frames ** **** *** pixels **** ** *** appear ************ **** ***** to *****. ****** ** those ****** *** **** averaged ** ****** *** image.

*******, **** ****** *** appear ** ***** *** confuse ******** ********** ******* of *** ***** ****** in ***** ****** **** frame ** *****. ******* of ****, ******** ***** reduction ***** ****** ****** blur *** *********, ********* later ** **** ******.

3D ***** *********

******, ** ********** ******* the ******* *** ******** methods ********* * ******* image **** **** ****** blur, *** *** **** commonly **** ** ***** surveillance. ****** ** ***** quality *** ******** **** 3D *** ****** **** manufacturer ** ************ *** depends ** *** *********** of **** **********.

**** **** *** ***** 3D *** ******** *** are ***** **** *************** in ***** ************ **** though ** ** *** technically *******. ********, ******* DNR ** ***** ****** 2D ***, ****** ** DNR *** *** ****** spatial ** ******** ***** reduction **********.

Manufacturer *** ********

**** ************* ** **** use * *********** ** spatial *** ******** **********, though ******* ** **** is ****** ********* ** users. **** *****, *** is ****** ********** *** "on/off" ***/** * ***** slider, **** ** *** Hikvision ***** *****. ***** is ** ***** ********** of ******* *******, ********, or **** ******* *** being ******** ****:

IPVM Image

*******, **** ************* ***** discrete ******* ** ******* and ******** *** ***** allows ***** **** ******* over ******** *** ***** specific ******, **** ** in **** ***** ******:

IPVM Image

***, ******, **** ***** for ********** ** **** temporal *********, ***** ** is **** ****** ** produce ******** ***** ******* effects, **** ** *** Avigilon ****** ***** *****:

IPVM Image

DNR **** ******

***** *** ****** ***** advantages ** **** *****, high ****** *** ********* blur ** ****** ******* caused ********** ***** *********, ***** ****:

IPVM Image

DNR **** ** ****** *** *******

** *** *******, ****** all ******* ****** **** motion **** ** *** light ******, *** *** severity *** ***** ****** where ******** ******** ******. Shown ** ******** *** ***** ********, **** *** "****" performing **** *****-** ******** slight **** ** ~* lux, ***** ******* *** "worst" ********** ******** ************ long ******** ****** *** reduced *******.

IPVM Image

***** ****** ******** ** brighter ***** ****** (*.*. 2 *** *****), *** were ******* ** **** cameras.

IPVM Image

*** ******** *** ***** **** Shootout ******* **** ******* *** examples ** ***** *** other ******.

Reducing ****** **** ********* *****

***** ****** *** ******* produce **** **** **** temporal ***** ********* ** used, ******** *** ** temporal ******** ************ ** reduce ****** **** ********* results ** ** ******** in ******* ***** ***** effectively ******** ******* *******.

*** *******, ** *** left, *****, *** *******'* face *** **** *** blurred ** ******** ********* caused ** ******** ***** reduction, ********* ******* *******. With ******** ***** ********* turned ****, *****, ******** becomes ******** *******, *** his **** ******* ******** by ******* *****, **** recognizable ******* ***********.

IPVM Image

******* ** ****, ***** we ********* **** ***** attempt ** ****** ******** noise ********* ** ****** for **** ** ******** scenes, ** ** ****** that ******* **** ******* only ********, ** ** all.

Intelligent ***

**** ******* *** ******* "intelligent" ***, ***** ******** similarly ** ***** ****** (discussed *****), ********** *** on ****** ***** ** the *****, ***** ******** it ** ****** ******* to ****** ****** ****.

*** *******, ** *** image *****, *** ****** automatically ******* *** ****** on *** ******* ****** through *** *****. **** avoids ********* ****** **** caused ** **** *** levels *** **** **** visible *****, **** ****** the *******.

IPVM Image

*********** *** ** ** uncommon *******, *** **********, with ***** ************ ******** **** *********.

DNR ******* **********

******* ******* **** *** manufacturers **** ***** ****** enabled, ** ***** **** DNR ******* ******** ** an ******* ** ~**% in * ~*.* *** scene. **** *******' ******** decreased ** **-**%, **** one ******* ********** ** ~25%.

IPVM Image

**** *** ******** ** a ******** ** ***** codecs ** **** ******* tested, ** **** *** provide ***** ********* **** smart ****** *** ********. For *******, ******* * and * ******** *********** the **** ******* **** DNR ** ** *** when ***** ****** **** disabled, ***** ****.

IPVM Image

************* ** *** ************ mention ******* *** ****** on ***** ***** ** their *************, ** **** testing ** ***** ** required ** ********* ****.

Comments (1)

* ***** *** ***** have ** ***** ** how *** **** ******.

**** * **** ** this ******* :

IPVM Image

** *** **** ** the ***** *** *** tell **** *** **** camera *** * **** enough ******* ***** **** it **** ******* ***** a ***, ******* *** Dahua *** * ***** image ** *** **** (because ** ** ********** during *** *******). * can't *** **** ********** that ********** ***** ** the **** ****** * walk ***** ** ********, but ** ** ***'*, then ******* ******** **** different ****** ** *** cycle ***** *** ** completely ****.

*** ***** ***** ** also **** ********* ** the ***** ***** (**** and *****), ***** ******** that *** ****** ** poorly ********** *** *** scene ** ***** ** clearly ***** ** ***** the ****/******* ***** *** get * **** ********* image *** * ****** walking ** *** ***. My **** ** **** the **** ************, ********* *** ******* compression *********, ***** ******** that *** ******* ***** can ** *********. * don't **** *** ****** in ********, *** * don't **** ** *** firmware ***** ** ********** to *** * ****** shot ** *** ***** camera, ** **** ***** to ** **** ***** is ********* *** * better ******* **** **** camera.

* **** *** *** of *** **** "****" problematic. ** ** *******, the ******* *** * better *******, *** ***'** telling **** ******** **** Axis ** "****" (*** this ** * **** ostensibly ******** ** ***** DNR, *** *** * camera ******). *** ****, on *** ********* ************, we're ** ****** ******* to **** **** ****** delivered **** *******. * am ******** **** ** because ***'** ***** ********* old ******, *** **** didn't ****** *** *******/******.

********* ******** ** (********* just ****** * **** shutter): * *** **** experimentation ** * *** client * **** ****** on. ** *** ******* speed ** "****" ** a *** ***** *******, it ***** ** ***** "stacking" ****** ** *** want ** *** ********** things **** *******. **** is ********* *** ***** do ***** *** ****. Likewise *** ***** ** spatial (*******) ** *** client ***** *** **** too. * ***'* **** if ****** ** ***** this *****, *** * think ** ***** ** a **** ******** ** any ***.

Agree: 1
Disagree: 1
Informative
Unhelpful
Funny
Read this IPVM report for free.

This article is part of IPVM's 7,419 reports and 980 tests and is only available to subscribers. To get a one-time preview of our work, enter your work email to access the full article.

Already a subscriber? Login here | Join now

This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.

Loading Related Reports