Image Filtering

Description

This step could perform filtering process on the input images with different kinds of filters.

Parameters of Input and Output

Image/Color

[Input] original images

Image/Color

[Output]filtered images

Parameters

BilateralFilter

Instruction:There are three parameters in this filter. The diameter determines the range of filtering area. When diameter increases, more original image information would be retained, so that the image is still clear. Otherwise, smaller diameter would blur the image. The paramter ‘sigmaColor’ represents the differnce of pixel value between pixels and ‘sigmaSpace’ represents the difference of spatial distance between pixels. These two values affect the filter result jointly.
Default Value:diameter=3; sigmaColor=3;sigmaSpace=3.
Suggested Value:To set according to the real scenarios.

Boxfilter

Instruction:The only parameter is the kernel size during the filtering process and its unit is pixel. When the parameter is increased, the image is blurred.
Default Value:kernelSize=100.
Suggested Value:To set according to the real scenarios.

DfFilter

Instruction: The full name of the filter is Discrete Fourier Filter and it mainly used to filter out the signals with high frequencies, noise and edges for example. This filter is capable of noise reduction and edge enhancement.
Default Value:None
Suggested Value:None

GaussianFilter

Instruction: This kind of filter processes the images with the Gauss functions. There are four parameters here: sizeX and sizeY, referring to the size of the Gauss kernel and when they increase, the image would be blurred; sigmaX and sigmaY, referring to the variances in the Gauss function and the increase of them would also blur the image.
Default Value: sizeX=3,SizeY=3;sigmaX=3,sigmaY=3.
Suggested Value: To set according to the real scenarios.

HaarFilter

Instruction:Smooth noise reduction processing, which can filter out part of noise.
Default Value:None
Suggested Value:None