Image processing - Noise notes

The noise in image processing is mainly divided into two categories

Additive noise ： It has nothing to do with the image intensity itself , It can be regarded as a direct superposition .

Multiplicative noise ： It is related to the image signal , It often changes with the change of image signal .

Additive noise conforms to a certain probability distribution , There are mainly several kinds of ：

Gaussian noise （ White noise ）： Normal distribution , White spot . Gaussian filtering （ mean value ）.

Salt and pepper noise ： The random change of some pixels caused by the pulse , Black and white dot . Use nonlinear filtering （ The median ）

Poisson noise ： Poisson distribution .

At present, image denoising is divided into three categories ： Filter based approach （Filtering-Based Methods）、 A model-based approach （Model-Based Methods） And a learning based approach （Learning-Based Methods）.