Filters for Dummies

Preface: This guide is a basic guide to the types of filters available for the LMI Gocator family of products. This is designed for a basic guide only. Each of these filters are standard imaging processing filters and more information can be found on them online. Many of these filters are more applicable to 2D images rather than 3D – such as an equalize filter.

Median Filer
• Center pixel of an M x M window is replaced by the median value of that window
• Uses: Smoothing filter that preserves edges while reducing noise

Gaussian
• Also known as Gaussian Linear Filter, Gaussian Blur and Gaussian Smoothing
• Applies a Gaussian function to each pixel
• Smoothing filter that reduces noise and detail by “blurring” the image

Open Filter

• Erosion filter followed by a Dilation filter
• Can “open” up a gap between objects

Close Filter

• Dilation filter followed by Erosion filter
• Can “close” holes in the data

Erode

• Reduces boundaries of regions of pixels. Holes will appear larger
• Removes small-scale details and reduces the size of regions of interest
• Uses: edge detection

Dilate

• Enlarges boundaries of regions of pixels. Dilations will remove holes in data
• Uses: edge detection

Morph Gradient

• Difference between dilation and erosion of a given image. Each pixel value indicates the contrast intensity in the close neighborhood of that pixel.
• Uses: Edge Detection and Segmentation

Sobel Magnitude

• Creates an image with emphasized edges

Laplacian

• Creates image with 2nd derivative of each pixel. It measures the rate of change from surrounding pixels
• Uses: Edge detection

Negative

• Darkest areas appear lightest (2D, grayscale image) or highest Z values appear as lowest Z values and vice versa

Equalize

• “Sharpens an image”. Light areas become while and dark areas become black

Binarize

• Pixels are given either a 1 or 0 value.
• Changes 3D image into 2D image
• Use: Segmentation

Percentile

• Removes specified range of data on low and high end of percentile
• Uses: can remove “floaties” and other noise

Relative Threshold

• Removes data outside of specified threshold of region of interest

Crop Only

• Crops the scan data to the user-defined region

Mask with Input

• Uses the surface input into the tool as a mask on the data. Any points in the filtered data will be set to null if the input surface is null at the same location. For example, the Gaussian filter can extend data along the edges, adding data in areas that contain null values. This filter would remove data that the Gaussian filter introduces, preserving the null values. This filter should follow any filter that introduces this kind of unwanted data.