kernel

README

kernel

small convolution package for images using go

Documentation

Index

• func Convolution(img image.Image, kernel [][]float64, stride, dilation, padding []int, ...) image.Image
• func CrossCorelation(img image.Image, kernel [][]float64, stride, dilation, padding []int, ...) image.Image
• func InverseCCorelation(img image.Image, kernel [][]float64, stride, dilation, padding []int, ...) image.Image
• func InverseConvolution(img image.Image, kernel [][]float64, stride, dilation, padding []int, ...) image.Image

Functions

func Convolution

func Convolution(img image.Image, kernel [][]float64, stride, dilation, padding []int, zeronegatives bool, threads bool) image.Image

Convolution does the convolution operation per channel on an image. threads will parallelize the convolution

func CrossCorelation

func CrossCorelation(img image.Image, kernel [][]float64, stride, dilation, padding []int, zeronegatives bool, threads bool) image.Image

CrossCorelation does the convolution operation per channel on an image. If the kernel is small and picture is small then don't bother setting the threads to big.

func InverseCCorelation

func InverseCCorelation(img image.Image, kernel [][]float64, stride, dilation, padding []int, zeronegatives bool) image.Image

InverseCCorelation does the reverse cross corelation operation per channel on an image. This is kind of like the back propagation of an CNN.

func InverseConvolution

func InverseConvolution(img image.Image, kernel [][]float64, stride, dilation, padding []int, zeronegatives bool) image.Image

InverseConvolution does reverse convolution operation per channel on an image. This is kind of like the back propagation of an CNN.