em_thresh: Compute likely pixel category based on EM algorithm...

In srvanderplas/ShoeScrubR: Functions for cleaning data from CSAFE's Longitudinal Shoe Study

Description

EM algorithm is for normally distributed groups; this assumption is likely not accurate, but it is fast and effective. If both N and scale_factor are NULL, the full image will be used, which will be slow.

Usage

em_thresh(img, scale_factor = 10, N = ifelse(is.null(scale_factor),
  pmin(length(img), 30000), NULL), ngroups = 3, quiet = T, ...)

Arguments

img	Image
scale_factor	Alternative to N - scales image by a factor of scale_factor
N	Number of points to sample from the image (speeds up computational time)
ngroups	Number of clusters
quiet	suppress output from normalmixEM using sink()?
...	additional arguments to mixtools::normalmixEM

srvanderplas/ShoeScrubR documentation built on Nov. 27, 2019, 2:09 p.m.