Calculate rectilinearity - a measure of how well a shape fits a rectangle
after optimal rotation.
Usage
get_rectilinearity(x, ..., .cols = NULL)
Arguments
- x
A matrix (nx2), list of matrices, or tibble with coo columns.
- ...
Additional arguments (reserved for future use).
- .cols
Column name(s) to process when x is a tibble. If NULL,
automatically detects columns containing coo objects.
Value
If x is a single matrix: returns a numeric scalar
If x is a list: returns a numeric vector
If x is a tibble: returns a numeric vector extracted from coo column
Details
Rectilinearity measures how rectangular a shape is by testing multiple
rotations and finding the orientation that best fits a rectangle.
Algorithm from Zunic & Rosin (2003). Higher values indicate more rectangular shapes.
Computationally intensive - tests 4n rotations where n is the number of points.
References
Zunic, J., & Rosin, P. L. (2003). Rectilinearity measurements for polygons.
IEEE Transactions on Pattern Analysis and Machine Intelligence.
Examples
get_rectilinearity(shapes$cat)
#> [1] 0.1343944
# Use in measure() on mini bot for the sake of speed
bot[1:2, ] %>% measure("rectilinearity")
#> # A tibble: 2 × 4
#> coo type dummy coo_rectilinearity
#> <out> <fct> <fct> <dbl>
#> 1 (138 x 2) whisky a 0.430
#> 2 (168 x 2) whisky a 0.459
