@CONFERENCE\{IMM2010-05896,
    author       = "S. Darkner and A. L. Dahl and R. Larsen and A. Skimminge and E. Garde and G. Waldemar",
    title        = "Segmentation by Large Scale Hypothesis Testing - Segmentation as Outlier Detection",
    year         = "2010",
    month        = "may",
    keywords     = "Segmentation, outlier detection, large scale hypothesis testing, locally adjusted threshold",
    booktitle    = "Proceedings of the International Conference on Computer Vision Theory and Applications (VISAPP), 2010",
    volume       = "",
    series       = "",
    editor       = "",
    publisher    = "",
    organization = "",
    address      = "",
    url          = "http://www2.compute.dtu.dk/pubdb/pubs/5896-full.html",
    abstract     = "We propose a novel and efficient way of performing local image segmentation. For many applications a threshold
of pixel intensities is sufficient but determine the appropriate threshold value can be difficult. In cases with
large global intensity variation the threshold value has to be adapted locally. We propose a method based on
large scale hypothesis testing with a consistent method for selecting an appropriate threshold for the given
data. By estimating the background distribution we characterize the segment of interest as a set of outliers
with a certain probability based on the estimated densities thus with what certainty the segmented object is
not a part of the background. Because the method relies on local information it is very robust to changes in
lighting conditions and shadowing effects. The method is applied to endoscopic images of small particles submerged
in fluid captured through a microscope and we show how the method can handle transparent particles
with significant glare point. The method generalizes to other problems. THis is illustrated by applying the
method to camera calibration images and {MRI} of the midsagittal plane for gray and white matter separation
and segmentation of the corpus callosum. Comparing the methods corpus callosum segmentation to manual
segmentation an average dice score of 0.86 is obtained over 40 images."
}