Geek asks and answers

How to determine the window size of a Gaussian filter

Gaussian smoothing is a common image processing function, and for the introduction of Gaussian filtering, please refer to here . As you can see, one parameter: standard output will determine the shape of the Gaussian function. However, when we perform convolution with Gaussian filtering, there is one more parameter: the window size of the Gaussian filter must also be determined at the same time. For example, when we use the function fspecialprovided by MATLAB, not only standard output should be provided, but also the size of the window. Intuitively, the larger the Gaussian standard output, the larger the Gaussian kernel window. However, there is no general rule on how to set the correct window size. Any ideas? Thank!

+5
source share
2 answers

The mask sizemasks the number of filters. A larger size, corresponding to a larger convolution mass, as a rule, leads to a greater degree of filtration. As a kind of compromise for greater noise reduction, large filters also affect the quality of image details.

This is like a milestone. Now, coming to Gaussian filter, the main parameter is standard deviation. If you use a 2D filter at the edge of the mask, you probably want the scale to approach 0 .

, , , standard deviation. , , .

, .

+11

.

  • 3 . . this

  • , 6sigma, 3sigma

  • , , 6sigma 6sigma - 1 2.

, .

, (, , ) , . , ​​ , .

+1

More articles:


All Articles