Speckle vs Gaussian Noise?
Isotropic and anisotropic noise-removal techniques! What is Mean, Median, and Bilateral Filtering?
In the last article, we covered the contrast enhancement of medical images and the enhancement operation was severely affected by noise components in the input data. Therefore, in this article, we will cover the understanding and the removal of noise components from medical image data.
Noise components are inevitable during the data acquisition process in the computer graphics applications. Furthermore, the types of noise depend on the acquisition device. The following two types of noise are commonly corrupting the image data:
- Additive White Gaussian Noise (AWGN)
- Multiplicative/Speckle Noise
Additive White Gaussian Noise (AWGN)
AWGN is the one of the most common type of noise and it is responsible for the image quality degradation. Noise is generated through several natural sources, for example electronic sensors. AWGN follows normal distribution (Gaussian distribution) with zero mean and it is additive in nature. Let’s consider an original noise-less image, which is corrupted during the acquisition and transmission process by AWGN and the corrupted image is represented by:
where η~N(0, σ²) is the normally distributed AWGN with zero mean and σ² variance.