PHOTOGRAPHY: On Noise and Digital Noise Reduction

Ever wondered what those dark and light spots are in your image when you take pictures in the dark? Well, let us help you put a term to them. They are known as “noise” and are prevalent in surveillance systems, camcorders, digital cameras, or any other video monitoring tool recordings.


From a technical point of view, ‘noise’ refers to the interference that occurs when audio and video signals are transmitted or reproduced. The most common factor leading to noise is low lighting levels. The chips of the cameras constantly pick up signals and byproducts from other electronic circuits within it, which also leads to noise and there can be more such reasons. The videos having noise issues translate into fuzzy ones.

Noise can be a serious problem and pose a bit of headache, specifically in video monitoring applications. One very telling example of this situation is when noise in the image dominates so much that the user cannot positively identify critical details like license plate numbers or faces of suspected people; nor he(she) can monitor dimly lit areas at retail stores in a nighttime video.

Although, the image sensors of security camera systems are designed to amplify signals at night to compensate the low light levels, but the problem arises when the noise captured by sensor elements also gets amplified along with these audios and video signals.

Types of Digital Noise

When we say “noise”, it could refer to any of the three different types as explained below.

Salt and Pepper noise
When the image sensor of the camera is overheated or there are dust particles inside the camera housing, this type of noise is produced. The artifacts in the final image are of different hues as compared to other pixels around them. These artifacts occur randomly and look like white and dark dots, justifying its name.

Gaussian noise
The pixels in the image of Gaussian noise appear slightly different from their original color, with the final image providing a softer tone of the actual objects. The main cause of Gaussian noise is the interference caused by electric components in the camera circuitry. This type of image disruption is also known as ‘Amplified Noise’.

Film Grain
This is the third type of noise and depends on the intensity level of the video signals. It occurs mainly because of low lighting levels. Videos taken at night times (under poor lighting conditions) display some uniform blending & texturing and the final image appears to be blurred and hazy.

We know that with every problem, sooner or later, there comes a solution too. Fortunately, we have already found a solution for the problem under discussion called ‘Digital Noise Reduction’ (DNR). Digital noise reduction is a vital capability of surveillance cameras to take images and videos under low lighting conditions. These days, most of the security monitors and recorders are featuring DNR. Let’s get a better understanding of this solution:

Digital Noise Reduction

Improving the illumination and providing optimum lighting levels at the scene might be one solution to capture noise-free images. However, it may not always be possible for all video monitoring applications. In such scenarios, preserving the image and video quality becomes the top priority of surveillance camera manufacturers. DNR effectively reduces noise and preserves salient details in security footage. It also helps to save up space on the storage devices like hard disk drive.

How Does DNR Function?

DNR makes use of an algorithm which analyses two consecutive image frames and eliminates the grains that do not appear in the previous frame. This image is then transferred to the DVR or NVR with a size reduction of 70%, which undoubtedly is clear and easy on the eye.

2D and 3D Digital Noise Reduction

2D-DNR stands for 2 Dimensional Digital Noise Reduction and 3D-DNR means 3 Dimensional Digital Noise Reduction. These digital noise reduction techniques are divided into two categories:

Spatial noise reduction: This noise reduction technique analyzes each frame and cancels the pixels that might produce noise in the final image. The greatest advantage of this method is that it is very effective at reducing blur motion.

Temporal noise reduction: Similar to above one, this technique uses temporal algorithms to analyze sequential frames to differentiate pixels that are likely to produce noise.

Though both these methods are beneficial, but have with their limitations as well. Spatial DNR may produce blurred edges of objects in the image and temporal DNR is not so much effective for producing images of moving objects. It’s recommended to use the blend of two. Some surveillance cameras might not use them simultaneously, but switch between the two techniques depending on the requirement.

So, the words like noise, DNR, spatial and temporal noise reduction won’t sound unfamiliar to you anymore. On the contrary, you would be able to explain them to others in dilemma!