Feature-Finding > Pattern Mapping
Processes and colors pixels based on their surrounding image pattern. Available tools for pattern analysis are local two-point correlation properties, mapping of selected FFT intensities, and matching of local FFTs to selected references. This function can take some time to complete depending on the size of the image and Window Size parameter.
After first clicking “Calculate”, a parallel-pool of CPU resources will be created which may take several seconds. This parallel-pool is good for 30 minutes and will not have to be re-created within that time. It is recommended that this function not be run on images larger than 1200×900, and that the window size be kept to 40 pixels or less. Try to resize the image as small as possible for best performance. Pattern Mapping still performs surprisingly well on images which appear pixelated.
- Two-Point Correlation: Calculates the two-point correlation function at each pixel, using a surround window of a specified size. The orientation, size, and shape of the “hotspot” of a pixel-window’s two-point correlation tell you information about the window’s local directionality, the extent of that directionality, and the size of features within the window.
- FFT Mapping: Allows you to define a mask to place around the FFT taken about each pixel, and either sum the intensities or take the peak intensity within the mask. This measurement is then assigned to each pixel. In essence, you are mapping out traits about each pixel’s local FFT, which can reveal local pattern differences in the image. The size of the window taken about the click to define the mask is determined by Window Size. It is often advised to set a large Window Size prior to clicking to define the mask, and then reducing the Window Size to something much smaller to perform the mapping.
- FFT Matching: Allows to define a fixed number of reference FFTs (i.e., “fingerprints”) by clicking points within the image. The size of the window taken about each click to define the “fingerprints” is determined by Window Size . Each pixel’s local FFT is then compared to each “fingerprint”, and labeled according to which “fingerprint” it matched the best.
b. Window Size
Window size to consider about each pixel for the pattern map
c. Section Grid
Grid size square length to split the image into sections to avoid out-of-memory issues. A section grid of 3 splits the image in a 3×3 grid and processes each square in series
Calculate pattern map
Mapped parameter to display
f. Colormap and Limits
Colormap to use for display, along with upper and lower limits for the colormap. Click “Auto” to automatically set the default limits.
a. Number of Bins
Use a set number of bins.
b. Bin Size
Use a fixed bin size. Enter the bin size in the same units that were mapped across the image.
Use watershed binning
- Window: Window size for gradient map calculation prior to watershed
- Weakness: Weakness setting for watershed binning
d. -90° Tolerance
Range of angles away from -90 to switch from negative positive (only active for Two-Point Correlation method and and Degrees display)
e. Noise Removal
Requires some kind of binning be done first. Reduces noise within the selected bins by performing Smart Dilation on each bin. Each time you run this tool, it is performed on the current state of the binned map, so be sure to “Undo” the last Noise Removal, if you want the next Noise Removal to be performed on the original binning. Only the last Noise Removal you perform will get saved as part of the Pattern Mapping Recipe step.
Consider feature size (area) when cleaning. Enter critical feature size and Logic for size cleaning.
b. Convex Area/Area
Consider feature convex area/area (shape) when cleaning enter critical feature convex area/area and Logic for convex area/area cleaning.
Consider feature misorientation (value difference from surroundings) when cleaning. Enter critical feature misorientation and logic for convex area/area cleaning.
Iterations for cleaning to loop over. Check or uncheck “Cumulative” to choose whether features get updated with cleaned value as soon as they get removed during an iteration.
a. Segment Bins
Draw white lines in between bins. Enter thickness for lines to draw between bins. At least a binned pattern map is required to use this tool. If there is a cleaned patterned map, the segment bins will operate on that.
Use thresholding to segment pattern map. Enter minimum and maximum values to select in segmentation. If a binned patterned map is present, the segment bins will operate on that, unless there is a cleaned map is present, in which case the segment bins will operate on that.
Need more help with this?
Chat with an expert now ››