RegisterUtilities.jl

Utility types and functions for image-registration workflows in the HolyLab ecosystem.

Installation

This package depends on RegisterCore.jl, which lives in the HolyLab registry. Add that registry once before installing:

using Pkg
pkg"registry add General https://github.com/HolyLab/HolyLabRegistry.git"
Pkg.add("RegisterUtilities")

Usage

Counter — column-major multi-dimensional index iterator

Counter yields every Vector{Int} index from [1,1,…,1] to max in column-major (first-index-fastest) order. Unlike CartesianIndices, the yielded values are plain vectors, which is convenient for arithmetic.

using RegisterUtilities

c = Counter((2, 3))
collect(c)
# [[1,1],[2,1],[1,2],[2,2],[1,3],[2,3]]

length(Counter((4, 5, 6)))  # 120

block_center — center coordinate of a block

Returns the 1-based center of a block with the given dimensions, using the same convention as MismatchArray blocks: center of dimension i is (sz[i] >> 1) + 1.

block_center(5)       # (3,)
block_center(4, 6)    # (3, 4)
block_center(5, 5, 5) # (3, 3, 3)

quadratic — quadratic-form array

Fills a 2-D array where each element (i,j) equals uᵀ Q u, with u = [i,j] - center and center = block_center(m,n) .+ shift. The second method wraps the result in a MismatchArray with a given denominator.

using RegisterUtilities, LinearAlgebra

Q = [1.0 0.0; 0.0 1.0]   # identity — distance² from center
A = quadratic(5, 5, [0, 0], Q)
# 5×5 matrix of squared distances from the block center (3, 3)

using RegisterCore
denom = ones(5, 5)
ma = quadratic(denom, [0, 0], Q)   # returns a MismatchArray

tighten — narrow element type

Converts a heterogeneous or overly-wide array (e.g. Array{Real}) to the narrowest concrete type that holds all its values.

A = Real[1, 2.0, 3f0]
B = tighten(A)
eltype(B)   # Float64