In the previous article, we introduced commonly used grayscale morphology tools, including dilation, erosion, opening operation, closing operation, etc.; today we will discuss image filtering, which is a commonly used image preprocessing tool in vision; filtering tools include: BoxFilter (Box filter), mean filter (Blur), Gaussian filter (GaussianBlur), median filter (MedianBlur), bilateral filter (BilateralFilter), etc.; if there are deficiencies, you can leave a message in the comment section, and everyone will learn together and make progress together!
1. Box filter (BoxFilter)
Box filtering is a very useful linear filtering, and the simplest mean filtering is the case of box filtering normalization. It can be said that box filtering is useful for all occasions where the sum of pixels in a certain neighborhood is required, such as mean filtering, guided filtering, calculation of Haar features, and so on. Its characteristics are just one word: fast! It can reduce the complexity of O(MN) summation, variance and other operations to O(1) or approximately O(1) complexity, that is to say, it has nothing to do with the size of the neighborhood. It is a bit like an integral graph. , but seems to be faster than the integral graph (depending on how it is implemented).
principle:
First look at a set of test renderings: