Topcon UA-10 Series, Instruction Manual

Page: 532 / 534

Example of “Size: 5”
The target pixel weight is 36/256. The weight values of peripheral pixels are arranged as
follows from the pixel closest to the target pixel;
24/256, 16/256, 6/256, 4/256, 1/256.
As one peripheral pixel is far away from the target pixel, the weight is smaller.
The weight is determined by using Gaussian distribution function according to the [Size],
[STDEV (X)] and [STDEV (Y)] parameters.
As the [Size] parameter is larger, the peripheral pixels have larger influence. As [STDEV (X)]
and [STDEV (Y)] parameters are larger, the weight of the peripheral pixel is larger and the
image is blurred.
Bilateral filter:
This is the processing method to reduce the noise included in an image.
When this method is used, the peripheral pixel values of the target pixel are obtained and the
Gaussian filter, which is determined by taking the distance and difference in values from each
pixel into consideration, is applied.
This method is effective to remove noise without blurring profile (edge). Sometimes it takes
long time to perform this processing method depending on the combination of parameters.
The size is determined according to [Diameter of each pixel neighborhood]. According to
[SigmaColor] and [SigmaSpace], the target pixel value is calculated by using the following
two types of Gaussian filters;
- Gaussian filter which is determined by taking the distance between pixels into
consideration;
- Gaussian filter which is determined by taking the difference in values between pixels into
consideration.
As the [Diameter of each pixel neighborhood] parameter is larger, the peripheral pixels have
larger influence. As the [SigmaColor] and [SigmaSpace] parameters are larger, the difference
in values between pixels has larger influence. Consequently the image is blurred.