Filter

Arithmetic

Arithmetic algorithms modify images through mathematical operations on pixel values. They are not used to smooth, but rather to adjust contrast, brightness, or sharpness:

Deviation highlights deviations and makes differences in the image visible.
Emphasis enhances specific image areas or features.
Sharpening emphasizes edges and details for a sharper appearance.
Gamma Correction adjusts brightness and contrast depending on the image display (e.g., monitor adjustment).
Subtract highlights differences between two images or versions.
Pow (power transformation) specifically changes the brightness distribution to emphasize dark or light areas.

Depending on the project, these methods are suitable for image enhancement, analysis and adjustment.

Deviation

Detects deviations from the target value in the image – highlights areas that deviate from the expected pattern.

Flow

To achieve results, the color image must be converted to grayscale. This can be done using the "To Grayscale (Image)" node.

Parameter set

Parameter	Type	Description	Effect Min	Effect Max
Image	Image	A grayscale image is required. You can use the "To Grayscale (Image)" node.
Mask Width (Pixel)	Int32	Set the width of the mask used for the deviation calculation. Larger values cover a wider area, helping to detect larger or more significant deviations.	Width: 5px Height: 5px	Width: 200px Height: 200px
Mask Height (Pixel)	Int32	Set the height of the mask used for deviation calculation. Larger values cover a larger area, which helps detect larger or more significant deviations.	Width: 10px Height: 2px	Width: 2px Height: 10px
Return	Region	A binary black-and-white image that highlights deviations from the expected pattern. This is useful for detecting anomalies or inconsistencies in the image.

Emphaszie

Adjusts the contrast and brightness of an image – improves the appearance and visibility of details.

Flow

Parameter set

Parameter

Type

Description

Value range

Effect Min

Effect Max

Image

Image → Input

Both grayscale and color images can be used. If a grayscale image is to be used, the image may need to be converted beforehand using the "To Grayscale (Image)" node.

Contrast

Single

Adjustment factor to increase or decrease image contrast. A higher value emphasizes the differences between light and dark areas.

> 0.01

Brightness

Int32

Higher values increase the overall brightness and make the image brighter.

>= 0

Sharpening

Sharpens an image by increasing the contrast between adjacent pixels – improving detail and clarity.

Flow

Parameter set

Parameter	Type	Description	Effect - Mono	Effect - Color
Image	Image → Input	Both grayscale and color images can be used. If a grayscale image is to be used, the image may need to be converted beforehand using the "To Grayscale (Image)" node.
Image	Image → Output	Transformed image with improved sharpness.

Gamma Correction

Adjusts the gamma value of the image – controls brightness and contrast for an optimized display.

Flow

Parameter set

Parameter	Type	Description	Effect - Mono	Effect - Color
Image	Image → Input	Both grayscale and color images can be used. If a grayscale image is to be used, the image may need to be converted beforehand using the "To Grayscale (Image)" node.
Image	Image → Output	Higher values increase gamma correction, which emphasizes dark pixels and attenuates bright areas, allowing for better segmentation in darker areas of the image.

Substract

Subtracts pixel values from one image from another – highlights differences between the two images.

Flow

Parameter set

Parameter	Type	Description
Image 1	Image	The first image from which the pixel values are subtracted.
Image 2	Image	The second image subtracted from the first image.
Return	Image	A new image showing the result of pixel-wise subtraction and highlighting the differences between the two input images.

Pow

Raises pixel values to a power – increases or decreases brightness to enhance contrast or emphasize detail.

Flow

Parameter set

Parameter	Type	Description
Image	Image → Input	Both grayscale and color images can be used. If a grayscale image is to be used, the image may need to be converted beforehand using the "To Grayscale (Image)" node.
Power	Single Beispiel 1	The exponent by which each pixel value is increased - Example with Power = 1.1
Power	Single Beispiel 2	Example with Power = 1.5
Return	Image	See pictures of Power Example 1 and 2

Smoothing

Smoothing algorithms are used to reduce noise in images and make surfaces appear more uniform. Depending on the method, they vary in the degree to which details or edges are preserved:

The Average filter calculates the mean of the neighborhood and creates a simple blur.
The Bilateral filter combines spatial proximity and color similarity, smoothing surfaces while preserving edges.
The Gaussian filter gives more weight to nearby pixels, providing a smooth, even smoothing effect.
The Median filter replaces each pixel with the median of its surroundings and is particularly effective at removing point-like noise.

Average

Applies an averaging filter – smooths the image and reduces noise or grainy structures.

Please use only positive and odd parameters : 1,3,5,7 …

Flow

Parameter set

Parameter	Type	Description	Effect - Min	Effect - Max
Image	Image → Input	Both grayscale and color images can be used. If a grayscale image is to be used, the image may need to be converted beforehand using the "To Grayscale (Image)" node.
Kernel Width (Pixel)	Int32	Width of the neighborhood used for averaging. Higher values increase smoothing but may blur details. The value must be positive and odd.	Width: 3px Heigth: 3px	Width: 9px Heigth: 3px
Kernel Height (Pixel)	Int32	Height of the neighborhood used for averaging. Higher values increase smoothing but may blur details. The value must be positive and odd.	Width: 3px Heigth: 9px	Width: 9px Heigth: 9px

Bilateral

Bilateral smoothing is a noise reduction filter that smooths surfaces while preserving edges. Unlike simple blur filters, it takes into account both the distance between pixels and their color similarity.

Flow

Parameterset

Parameter	Type	Description	Value range	Effect Min	Effect Max
Image	Image	Both grayscale and color images can be used. If a grayscale image is to be used, the image may need to be converted beforehand using the "To Grayscale (Image)" node.
Diameter	Int32	Diameter of each pixel neighborhood used in filtering. Higher values increase smoothing. Diameter determines the size of the area considered: A small value only has a local effect. A large value smooths over a larger area and is more computationally intensive.	>0	Diameter:5 SigmaColor 75 SigmaSpace 125 t=95ms	Diameter:25 SigmaColor 75 SigmaSpace 125 t=485Minms
Sigma Color	Single	Standard deviation of the color space, which affects how much the filter takes into account intensity differences between pixels. Higher values increase overall smoothness. SigmaColor controls the sensitivity to color differences: With small values, only very similar colors are averaged (edges remain sharp). With large values, different colors also become visible (stronger blurring).	0 - 254	Diameter:5 SigmaColor: 25 SigmaSpace: 125	Diameter:5 SigmaColor: 125 SigmaSpace: 125
Sigma Space	Single	Standard deviation of spatial distance, which affects how much the filter takes into account the proximity of pixels. At higher values, more distant pixels can influence the smoothing. SigmaSpace determines how much spatial distance counts: Small values limit the effect to nearby pixels. Large values also include more distant pixels and increase smoothing.	0 - 254	Diameter:5 SigmaColor: 75 SigmaSpace: 25	Diameter:5 SigmaColor: 75 SigmaSpace: 200

Gauss

Applies a Gaussian filter to an image and smooths it by averaging the pixel values within a specified Gaussian kernel size.

The Gaussian filter is suitable for smooth blurring and reducing fine noise. It is primarily used as a preprocessing step, for example, before edge detection.

Flow

Parameter set

Parameter	Type	Description	Effect - Min	Effect - Max
Image	Image → Input	Both grayscale and color images can be used. If a grayscale image is to be used, the image may need to be converted beforehand using the "To Grayscale (Image)" node.
Kernel Width (Pixel)	Int32	Width of the Gaussian kernel. Higher values increase smoothing but may blur details. The value must be positive and odd.	Width: 3px Heigth: 3px	Width: 9px Heigth: 3px
Kernel Height (Pixel)	Int32	Height of the Gaussian kernel. Higher values increase smoothing but may blur details. The value must be positive and odd.	Width: 3px Heigth: 9px	Width: 9px Heigth: 9px

Median

This node allows you to apply a median filter to an image, which reduces noise by replacing each pixel value with the median value of its neighborhood.

The median filter is particularly useful for removing point-like "salt and pepper" noise. It smooths noise without blurring edges as much as average or Gaussian filters.

Flow

Parameter set

Parameter	Type	Description	Effect Min	Effect Max
Image	Image	A grayscale image is required. You can use the "To Grayscale (Image)" node.
Kernel Size (Pixel)	Int32	The range of neighboring pixels used to calculate the median. Higher values increase noise reduction but may blur fine details. The value must be positive, odd and greater than 1.	Kernel Size: 3	Kernel Size: 11

Transformation

Transformation algorithms change the geometric representation of an image without directly editing its content. They are used to adjust size, orientation, or distortion:

Anisotropic Scaling changes the height and width independently (e.g., stretching or shrinking).
Isotropic Scaling enlarges or reduces the image evenly in all directions.
Resize adjusts the image to the desired size.
Translate moves the image by a specific amount in the x or y direction.
Flip mirrors the image horizontally or vertically.
Rotate Center rotates the image around its center point.
Undistort corrects distortions (e.g., caused by camera optics).

These methods are used to adjust, align or geometrically correct images.

Anisotropic Scaling

Scales the width and height of an image using different factors, enabling non-proportional resizing. In image processing, this is primarily used for correction and normalization, allowing objects to be analyzed consistently regardless of the recording conditions.

Flow

Parameter set

Parameter

Type

Description

Value range

Effect Min

Effect Max

Image

Both grayscale and color images can be used. If a grayscale image is to be used, the image may need to be converted beforehand using the "To Grayscale (Image)" node.

Original = Scale 1
Scale Width: 1
Scale Height: 1
1200*1200px

Scale Width

Single

Factor by which the image width is scaled.

Example values:
0.5: 50%
1.0: 100%
1.5: 150%

Scale Width: 1
Scale Height: 1.25
1200*1500px

Scale Width: 2
Scale Height: 1
1500*750px

Scale Height

Single

Factor by which the image height is scaled.

Example values:
0.5: 50%
1.0: 100%
1.5: 150%

Scale Width: 0.1
Scale Height: 0.1
120*120px

Scale Width: 1.7
Scale Height: 1.7
2040*2040px

Isotropic Scaling

Scales an image evenly in both directions while maintaining the aspect ratio. Isotropic scaling ensures consistent scaling and is ideal when proportions and shapes need to be preserved exactly, for example, in AI or classification systems.

Flow

Parameter set

Parameter	Type	Description	Value range	Effect Min	Effect Max
Image	Image	Both grayscale and color images can be used. If a grayscale image is to be used, the image may need to be converted beforehand using the "To Grayscale (Image)" node.		Original = Scale 1 1200*1200px
Scale	Single	Uniform scaling factor applied to both width and height.	Example values: 0.5: 50% 1.0: 100% 1.5: 150%	Scale: 0.5 600*600px	Scale: 3 3600*3600px

Resize

Changes image size to defined dimensions – width and height are adjusted specifically to the desired number of pixels.

Flow

Parameter set

Parameter	Type	Description	Value range	Effect Min	Effect Max
Image	Image	Both grayscale and color images can be used. If a grayscale image is to be used, the image may need to be converted beforehand using the "To Grayscale (Image)" node.
Width	Int32	Desired width in pixels.	>1	Width: 100px Height: 100px	Width: 500px Height: 1000px
Height	Int32	Desired height in pixels.	>1	Width: 100px Height: 100px	Width: 500px Height: 1000px

Translation

Moves an image horizontally and/or vertically – free areas are filled with black pixels.

Flow

Parameter set

Parameter	Type	Description	Value range	Effect Min	Effect Max
Image	Image	Both grayscale and color images can be used. If a grayscale image is to be used, the image may need to be converted beforehand using the "To Grayscale (Image)" node.
X (Pixel)	Int32	Horizontal offset. Positive values shift the image content to the right, while negative values shift it to the left.	>0 to the right <0 to the left	X: 100px Y: 100px	X: -250px Y: 100px
Y (Pixel)	Int32	Vertical offset distance. Positive values shift the image content downward, while negative values shift it upward.	>0 downward <0 upward	X: 150px Y: -250px	X: -250px Y: -150px

Flip

Flips an image horizontally or vertically – creates a mirror image along the selected axis.

Flow

Parameter set

Parameter	Type	Description	Value range
Image	Image	Both grayscale and color images can be used. If a grayscale image is to be used, the image may need to be converted beforehand using the "To Grayscale (Image)" node.
Mode	Enum 1	Direction in which the image should be mirrored: Flip horizontally	Horizontal
	Enum 2	Direction in which the image should be mirrored: Horizontally and vertically mirrored	HorizontalVertical
	Enum 3	Direction in which the image should be mirrored: Flip vertically	Vertical

Rotate Center

Rotates an image by a specified angle – rotation occurs around the center.

Angle > 0: Rotation to the left

Angle < 0: Rotation to the right

Flow

Parameter set

Parameter	Type	Description		Effect
Image	Image	Both grayscale and color images can be used. If a grayscale image is to be used, the image may need to be converted beforehand using the "To Grayscale (Image)" node.
Angle [deg]	Int32	Angle by which the image is rotated. Angle > 0: Rotation to the left Angle < 0: Rotation to the right	Example: 15°	Example: -45°
Resize To Fit	Boolean		Example: -45° Resize: True 1469*1469px	Example: -45° Resize: True 1697*1697px

Undistort

Removes the so-called fisheye effect from the image – corrects extreme wide-angle distortion.

We will provide the data upon request depending on the sensor and lens.

Flow

Parameter set

Parameter	Type	Description	Value range
Image	Image	Both grayscale and color images can be used. If a grayscale image is to be used, the image may need to be converted beforehand using the "To Grayscale (Image)" node.
Camera Params	Camera Params	Please use the “Make Camera Parameters (Undistort)” node for this. Fx: Focal length of the system (sensor - optics) in the x-direction Fy: Focal length of the system (sensor - optics) in the y-direction Cx: Optical center of the system in the x-direction, specified in pixel coordinates (usually the center of the sensor) Cy: Optical center of the system in the x-direction, specified in pixel coordinates (usually the center of the sensor) We will provide the data upon request depending on the sensor and lens.	Example VIU13 - 2.6mm lens: fx: 6454.156558 fy: 6454.156558 cx: 2104.000000 cy: 1560.000000
Distortion Params	Distortion Params	Please use the “Make Distortion Parameters (Undistort)” node for this. K1: Distortion parameter K1 K2: Distortion parameter K2 P1: Distortion parameter P1 P2: Distortion parameter P2 K3: Distortion parameter K3 We will provide the data upon request depending on the sensor and lens.	Example VIU13 - 2.6mm lens: k1: -0.667539 k2: -0.608235 p1: 0.015444 p2: 0.003268 k3: 0