These properties are read-only.

Bounding boxes, stored as either a 4-element [*x*,*y*,*width*, *height*]
vector or an *M*-by-4 matrix of individual bounding
boxes.

Weights, stored as a vector containing a weight for each bounding
box. The weights are used to define the coefficients of the filter.

Filter coefficients, stored as a numeric value.

Filter center, stored as [*x*,*y*]
coordinates. The filter center represents the center of the bounding
rectangle. It is calculated by halving the dimensions of the rectangle.
For even dimensional rectangles, the center is placed at subpixel
locations. Hence, it is rounded up to the next integer.

For example, for this filter, the center is at [3,3].

These coordinates are
in the kernel space, where the top-left corner is (1,1). To place
the center in a different location, provide the appropriate bounding
box specification. For this filter, the best workflow would be to
construct the upright kernel and then call the `rot45`

method
to provide the rotated version.

Filter size, stored as a 2-element vector. The size of the kernel
is computed to be the dimensions of the rectangle that bounds the
kernel. For a single bounding box vector [*x*,*y*,*width*, *height*],
the kernel is bounded within a rectangle of dimensions [(*width*+*height*)
(*width*+*height*)-1].

For cascaded rectangles, the lowest corner of the bottom-most
rectangle defines the size. For example, a filter with a bounding
box specification of [3 1 3 3], with weights set to 1, produces a
6-by-5 filter with this kernel:

Filter orientation, stored as the string `'upright'`

or `'rotated'`

.