Image

class Image<type T = Float32> : linalg::Matrix<T> {
    ...
}

Basically a two-dimensional array with dynamic size.

Pixel access with operator[x, y]:

var centerValue = image[width/2, height/2]
image[10, 20] = 0.0

ImageView as “read-only reference” of sub-areas (with stride).
ImageBasicView as “read-only reference” with continuous pixels (i.e. no stride, just cut off at top or bottom).

Common image-processing member functions:

extension Image<type T> {
- func gaussianBlur(Float sigma) -> Image<T>&
- func medianFilter(Int kernelSize) -> Image<T>&
- func sobelEdges() -> Image<T>&
- func resize(Int width, Int height, Interpolation interpolation) -> Image<T>&
- func threshold(T lower, T upper) -> Image<T>&
}

Same operations as free functions:

func gaussianBlur<type T>(Image<T> image, Float sigma) -> Image<T>
func medianFilter<type T>(Image<T> image, Int kernelSize) -> Image<T>
func sobelEdges<type T>(Image<T> image) -> Image<T>
func resize<type T>(Image<T> image, Int width, Int height, Interpolation interpolation) -> Image<T>
func threshold<type T>(Image<T> image, T lower, T upper) -> Image<T>

Row-Major

Images (unlike matrices) are stored row-major, like:

1 2
4 5
7 8

Datatype & Precision

The default datatype is Float32 (32-bit precision), well suited for image processing and machine vision.

For handling JPEG or PNG images you may use these derivations:

using RGB24Image = cilia::Image<RGB<UInt8>>
using Gray8Image = cilia::Image<UInt8>