Max Homogeneity

The Maximum Homogeneity Neighborhood Node lays 8 different areas around one Pixel and applies a measurement of homogeneity which is then processed in some sort of weight. Then, it calculates each areas mean-value and combines both mean-value and weight to the new pixel value, so that the most homogeneous areas take the most part in the new value of the pixel and areas that lay over edges are not taken into account. Via a parameter you can regulate the influence of the weights.

Options

Lambda

The Parameter to determine the influence of the weights via weights^lambda. For lambda = 0 the filter becomes a simple mean filter which filters the most noise but preserves no edges. For lambda -> infinity only the most homogeneous area is taken to get the new Pixel-Value thus Edges are most preserved but the denoising is not as strong.

Window Span

The Window Span parameter determines the span of the window in one direction. The resulting window size is given by: span*2+1 in each dimension. So the bigger it gets, the more noise is filtered out but the less details are preserved.

Dimension Selection

This component allows the selection of dimensions of interest. If an algorithm cannot, as it only supports fewer dimensions than the number of dimensions of the image, or shouldnot, as one wants to run the algorithm only on subsets of the image, be applied on the complete image, the dimension selection can be used to define the plane/cube/hypercube on which the algorithm is applied. Example 1 with three dimensional Image (X,Y,Time): An algorithm cannot be applied to the complete image, as it only supports two dimensional images. If you select e.g. X,Y then the algorithm will be applied to all X,Y planes individually. Example 2 with three dimensional Image (X,Y,Time): The algorithm can be applied in two, three or even more dimensions. Select the dimensions to define your plane/cube/hypercube on which the algorithm will be applied.

Out of Bounds Selection

The 'OutOfBounds Strategy' is used when an algorithm needs access to pixels which lie outside of an image (for example convolutions). The strategy determines how an image is extended, for examples see here

Column Selection

Column Creation Mode

Mode how to handle the selected column. The processed column can be added to a new table, appended to the end of the table, or the old column can be replaced by the new result

Column Suffix

A suffix appended to the column name. If "Append" is not selected, it can be left empty.

Column Selection

Selection of the columns to be processed.

Views

Image Viewer

Another, possibly interactive, view on table cells. Displays the selected cells with their associated viewer if it exists. Available views are:
- Histogram Viewer
-- This viewer shows the histogram of the currently selected image.
- Missing Value Viewer
-- An empty viewer that is shown when the input cell has no value to display.
- Image Viewer
-- This viewer renders the selected image-cell.
- Combined View
-- A viewer shown when the user selects an interval of rows and columns in the viewer. This viewer combines all images and labelings in the selected interval to one image by rendering them next to each other. Alternatively, the images and labelings can be layed over each other.
- BigDataViewer
-- A viewer shown when the user selects an interval of rows and columns in the viewer. This viewer combines all images and labelings in the selected interval to one image by rendering them next to each other. Alternatively, the images and labelings can be layed over each other.
- XML
-- XML tree
- Labeling View
-- View on a labeling/segmentation

Workflows

• 01_Spot_DetectionKNIME Hub
• 01_Spot_Detection modif KNIME Forum

Links

• No links available

Developers