fingerprint

Documentation

Overview

Package fingerprint provides fingerprinting of images.

Index

• func BMVDelta(h, g BMVHash) float64
• func ColorHistDelta(h, g ColorHist) float64
• type BMVHash
  • func BMVHashFromString(s string) (BMVHash, error)
  • func NewBMVHash(img image.Image) BMVHash
  • func (h BMVHash) HammingDistance(v BMVHash) int
  • func (h BMVHash) Image(width, height int) *image.Gray
  • func (h BMVHash) String() string
• type ColorHist
  • func ColorHistFromString(s string) (ColorHist, error)
  • func NewColorHist(img image.Image) ColorHist
  • func (h ColorHist) Image(width, height int) *image.RGBA
  • func (h ColorHist) String() string

Functions

func BMVDelta

func BMVDelta(h, g BMVHash) float64

BMVDelta returns the difference between the two block mean value hashes h and g. The difference is in the range [0,1] with 0 for identical hashes and 1 for maximal different hashes (i.e. a Hamming distance of 64). The lowest significant difference in BMVHashes is 2 bits corrsponding to a delta of 2/64 = 0.03125.

func ColorHistDelta

func ColorHistDelta(h, g ColorHist) float64

ColorHistDelta returns the difference between the two color histograms h and g. The difference is in the range [0,1] with 0 for identical color histograms and 1 for maximal different histograms.

Types

type BMVHash

type BMVHash uint64

BMVHash is the 64 bit block mean value hash of an image. The following references contain more information: Di Wu, Xuebing Zhou, and Xiamu Niu. 2009. A novel image hash algorithm resistant to print-scan. Signal Process. 89, 12 (December 2009), 2415-2424. As described in Christoph Zauner: "Implementation and Benchmarking of Perceptual Image Hash Functions" DIPLOMARBEIT, FH Hagenberg, Juli 2010.

func BMVHashFromString

func BMVHashFromString(s string) (BMVHash, error)

BMVHashFromString parses the hexadecimal number in s and panics if s cannot be parsed to a uint64.

func NewBMVHash

func NewBMVHash(img image.Image) BMVHash

NewBMVHash computes the block mean value hash of img. The following degenerate case return special values:

• If one or both dimensions of the image are < 8 then BMV hash of 0 (i.e. 64 0s) is returned.
• If one dimension is smaller than 16 a fingerprint of 64 1s is returned.

func (BMVHash) HammingDistance

func (h BMVHash) HammingDistance(v BMVHash) int

HammingDistance returns the Hammig distance between the bit strings of h and g.

func (BMVHash) Image

func (h BMVHash) Image(width, height int) *image.Gray

Image "reconstructs" the original image through gray panels. The generated image has dimensions width x height. Both dimensions must be at least 8.

func (BMVHash) String

func (h BMVHash) String() string

String returns h in hexadecimal form.

type ColorHist

type ColorHist [24]byte

ColorHist is a normalized (i.e. the dominant color is represented as 255) color histogram based on the colors from the Greta Mecbeth Color Picker.

func ColorHistFromString

func ColorHistFromString(s string) (ColorHist, error)

ColorHistFromString converts 24 base36-digits to a ColorHist.

func NewColorHist

func NewColorHist(img image.Image) ColorHist

NewColorHist computest the color histogram of img.

func (ColorHist) Image

func (h ColorHist) Image(width, height int) *image.RGBA

Image reconstructs the original image from the color histogram.

func (ColorHist) String

func (h ColorHist) String() string

String produces a string representation by renormalizing the histogram to 36 so that it can be encoded in 24 base-36 digits.