Selects endmember candidates by identifying extreme points along coordinate axes. This function implements a deterministic approach to endmember initialization by finding the minimum and maximum values along each coordinate axis. If insufficient unique extreme points are found, it supplements with random projections.
extreme_coordinates(data, p, random = FALSE)A numeric matrix where each row represents a data point and each column represents a coordinate/dimension.
Number of endmembers to select.
A logical value indicating the selection strategy. If TRUE, selects
randomly p points from all dimensions simultaneously. If FALSE (default),
iteratively selects extreme points dimension by dimension.
A list containing:
An integer vector of length p containing the row indices of
the selected endmember candidates from the input data matrix.
# Create sample data
set.seed(123)
data("demo_data")
# Select 3 endmembers using deterministic approach
result1 <- extreme_coordinates(demo_data, p=3, random=FALSE)
print(result1$indices)
#> [1] 3 1 5
# Select 3 endmembers using random approach
result2 <- extreme_coordinates(demo_data, p=3, random=TRUE)
print(result2$indices)
#> [1] 5 2 4