segmentation

Documentation

Overview

*

• This is probably the most important package in the repository. It contains
• all the segmentation algorithms and the utilities to transform from an
• image to a graph and from a DisjointSet forest to an image.

Index

• func IntensityDifference(p1 graph.Pixel, p2 graph.Pixel) float64
• func NNWeight(p1 graph.Pixel, p2 graph.Pixel) float64
• type ImageColor
  • func (color *ImageColor) RGBA() (uint32, uint32, uint32, uint32)
• type Segmenter
  • func New(img image.Image, graphType graph.GraphType, weightfn graph.WeightFn) *Segmenter
  • func (s *Segmenter) GetResultImage() image.Image
  • func (s *Segmenter) SegmentGBS(sigma, k float64, minSize int)
  • func (s *Segmenter) SegmentHMSF(sigmaSmooth, minWeight float64)
  • func (s *Segmenter) SetRandomColors(val bool)

Functions

func IntensityDifference

func IntensityDifference(p1 graph.Pixel, p2 graph.Pixel) float64

*

• Computes the absolute difference between the two pixels intensities.

func NNWeight

func NNWeight(p1 graph.Pixel, p2 graph.Pixel) float64

*

• Computes the Euclidean distance between two pixels.
• For p1 = (r1, g1, b1) and p2 = (r2, g2, b2):
• d = sqrt((r2 - r1)^2 + (g2 - g1)^2 + (b2 - b1)^2)

Types

type ImageColor

type ImageColor struct {
	// contains filtered or unexported fields
}

*

• Color type that implements color.RGBA

func (*ImageColor) RGBA

func (color *ImageColor) RGBA() (uint32, uint32, uint32, uint32)

*

• Returns the RGBA values of the given color

type Segmenter

type Segmenter struct {
	// contains filtered or unexported fields
}

*

• Type used to run all the segmentation algorithms.
• It stores the graph, the resultset, the original image, the graph
• obtained from the image and if it will generate a result image with
• random colors.

func New

func New(img image.Image, graphType graph.GraphType,
	weightfn graph.WeightFn) *Segmenter

*

• Returns a new Segmenter, generates a graph of the given graph type from
• the given image using the given weight function to compute the edge
• weights.

func (*Segmenter) GetResultImage

func (s *Segmenter) GetResultImage() image.Image

*

• Returns the result image. Returns nil if no segmentation algorithm
• has been executed before.

func (*Segmenter) SegmentGBS

func (s *Segmenter) SegmentGBS(sigma, k float64, minSize int)

*

• Performs the image segmentation using the "Graph Based Segmentation"
• algorithm. It uses sigma to apply a gaussian filter with it to the image
• to smooth it before running the algorithm.
• k and minSize are the algorithm parameters. For more information on this
• algorithm refer to either my report which link is on the repo's README or
• to: http://cs.brown.edu/~pff/papers/seg-ijcv.pdf

func (*Segmenter) SegmentHMSF

func (s *Segmenter) SegmentHMSF(sigmaSmooth, minWeight float64)

*

• Performs the image segmentation using the "Heuristic for Minimum Spanning
• Forests" algorithm. It uses the weightfn to compute the weight of the
• graph edges. minWeight is the only parameter.
• For more information on this algorithm refer to either my report which link
• is on the repo's README or to:
• http://algo2.iti.kit.edu/wassenberg/wassenberg09parallelSegmentation.pdf

func (*Segmenter) SetRandomColors

func (s *Segmenter) SetRandomColors(val bool)

*

• Sets the random color attribute to true or false according to val