Traditional Noise Reduction Algorithm in AKVIS Noise Buster AI

Traditional Noise Reduction Algorithm

AKVIS Noise Buster AI allows you to remove both luminance and color noise from your digital photographs. Before version 12.0, the software used the traditional noise reduction algorithm. Since version 12.0, the program uses trained neural networks. You can switch to the old algorithm by selecting the corresponding option in the Preferences.

Source Image Noise Removal
Noise Removal

Noise on any color picture can be divided into luminance noise and color noise. Luminance noise is perceived as dark dots or small blots, color noise represents color spots on areas having a different color. Another name for color noise is chroma noise.

The Histogram below the Settings is a graphical depiction of the noise level and noise components of the original picture. The gray area shows the luminance noise; the red area shows the color noise. The more the histogram is shifted to the right, the higher the noise level is. And vice versa, if the histogram is shifted to the left, the image has a low noise level. The height of the histogram displays the number of pixels having this noise level.

When adjusting the parameters, pay attention to the right part of the histogram, as it shows the amount of highly noised pixels. The amount of pixels in the left part shows low noised pixels that are a part of the picture, its natural background.

Image Noise Histogram
Image Noise Histogram


On the Settings Panel, adjust the effect parameters:

Fade (0-100%). This parameter sets the mixing ratio of the filtered image and the original. At 100% all elements of the photo defined as noise are smoothed in accordance with the value of the parameter Smooth Level. As you move the slider to the left (i.e. reduce the value of the parameter), the original image noise mixes into the filtered image; at 0% there is no filtration at all. In most cases a certain amount of original noise adds to the natural look of an image and allows restoring small details.

Weak Fading
Fade = 10%
Intensive Fading
Fade = 90%

Quality (1-20). The parameter reduces the number of color spots but significantly increases the processing time.

Base Smoothing Quality
Quality = 3
High Smoothing Quality
Quality = 10

Noise Level (0-100). The parameter defines which elements of the image are to be considered noise (luminance or color) and which are to be considered important details. At high value of the parameter small details can be defined as noise and, therefore, be smoothed.

Details Are Not Affected by Smoothing
Noise Level = 0
Details Smoothed Out
Noise Level = 100

Smooth Level (0-100). The parameter sets the extent to which the elements defined as noise should be smoothed. The higher the parameter, the smoother the filtration result. However, at high values of the parameter, an image can lose detail and look flat. High values of the parameter Smooth Level for color noise can produce an alteration of colors and loss of small color details.

Soft Smoothing
Smooth Level = 20
Intensive Smoothing
Smooth Level = 100

Improve Detail Parameter Group:

Remove Moiré Check-Box. Use this mode to reduce unwanted moiré patterns and remove halftone from scanned newspaper photos. The degree of smoothing is regulated by the Blur parameter.

Coarse-Grain Image
Halftone Image

Restore Coarse-Grain Image
Blur = 1.0
Restore Coarse-Grain Image
Blur = 4.0


Noise Buster AI v. 12.1 - Free 10-day Trial    Download