Image noise

From Wikipedia, the free encyclopedia

Noise clearly visible in this image (Panasonic Lumix LX1)
Noise clearly visible in this image (Panasonic Lumix LX1)

Image noise is a random, usually unwanted, fluctuation of pixel values in an image. Image noise can originate in film grain, or in electronic noise in the input device (scanner or digital camera) sensor and circuitry, or in the unavoidable shot noise of an ideal photon detector.

Image noise is most apparent in image regions with low signal level, such as shadow regions or underexposed images.


Contents

[edit] Useful noise

High levels of noise are almost always undesirable, but there are cases when lower levels of noise may be useful, for example to prevent discretization artifacts (color banding or posterization). Noise purposely added for such purposes is called dither.

[edit] Noise problems with digital cameras

Image on the left has exposure time of >10 seconds in low light. The image on the right has adequate lighting and 0.1 second exposure.
Image on the left has exposure time of >10 seconds in low light. The image on the right has adequate lighting and 0.1 second exposure.

In low light, or at high exposure index (ISO speed) settings, digital cameras tend to produce images with more apparent image noise. The two examples show a typical difference (best seen at full-size) between a well-lit subject and one in low light.







[edit] Image Noise Types

Image with salt and pepper noise.
Image with salt and pepper noise.

[edit] Salt And Pepper Noise

An image containing salt and pepper noise will have dark pixels in bright regions and bright pixels in dark regions. This type of noise might be the result of classification errors of the surface material. Corruption in conversion from analog to digital will also cause salt and pepper noise. [1]

[edit] Gaussian Noise

An image containing Gaussian noise has a normally distributed addition of intensity at each point in the image region.[1]





[edit] Image noise removal

Noise cannot be removed without the loss of some information in the form of image detail. Nevertheless, noise-reduction algorithms have been developed to analyze an image and determine whether unduly dark or light pixels are, in fact, details or are more likely to be due to noise.


[edit] Linear filters

One method to remove noise is by convolving the original image with a smoothing kernel. The sharpness of the image is reduced as the kernel grows in width. This is because the kernel's width determines the size of the neighborhood of pixels that will affect the final intensity of the pixel being smoothed. Different linear filters determine different weights to be used for neighboring pixels.[2]

Linear filtering reduces noise, but the blurring also reduces sharpness which can be a major drawback to image quality.[1]

Mean/Box Filtering

In mean filtering each of the neighborhood pixels are given equal weight in determining the intensity of the pixel being smoothed. In effect, any lack in conformance of a pixel to its neighborhood is averaged out. The values of the kernel are the same and and add up to one.[2]

Gaussian Filtering

In Gaussian filtering each of the neighborhood pixels are given weights in approximation of a Gaussian function in determining the intensity of the pixel being smoothed. In effect, pixels closer to the pixel being smoothed are more important than pixels farther from it. For this very same reason, Gaussian filtered images do not experience ringing unlike mean filtered images.[3]


[edit] Non-Linear filters

A median filter is an example of a non-linear filter that is fairly good at preserving image detail[4]. To run a median filter:

  1. consider each pixel in the image
  2. sort the neighbouring pixels into order based upon their intensities
  3. replace the original value of the pixel with the median value from the list

This type of filter is very good at removing salt and pepper noise from an image, and causes relatively little blurring of edges, and hence is often used in computer vision applications.

[edit] Low- and high-ISO noise examples

[edit] Video noise

In video and television, noise refers to the random dot pattern that is superimposed on the picture as a result of electronic noise, the 'snow' that is seen with poor (analog) television reception or on VHS tapes. Interference and static are other forms of noise, in the sense that they are unwanted, though not random, which can affect radio and television signals.

[edit] See also

[edit] References

  1. ^ a b c Shapiro, Linda G., and George C. Stockman. Computer Vision. Upper Saddle River, New Jersey: Prentice-Hall, Inc., 2001.
  2. ^ a b Fisher, Bob. "Mean Filter." Spatial Filters. 8 Jun 2008 <http://homepages.inf.ed.ac.uk/rbf/HIPR2/mean.htm>.
  3. ^ Fisher, Bob. "Gaussian Smoothing." Spatial Filters. 8 Jun 2008 <http://homepages.inf.ed.ac.uk/rbf/HIPR2/gsmooth.htm>.
  4. ^ Fisher, Bob. "Median Filter." Spatial Filtering. 8 Jun 2008 <http://homepages.inf.ed.ac.uk/rbf/HIPR2/median.htm>.

[edit] External links