phash

README

phash

A simple perceptual hashing library in Go.

Usage :

f, err := os.Open("image.jpg")
if err != nil {
    panic(err)
}
defer f.Close()
img, _, err := image.Decode(f)
if err != nil {
    panic(err)
}
hash1 := phash.DTC(img)
hash2 := phash.DTC(img)

if phash.Distance(hash1, hash2) == 0 {
    fmt.Println("these images sure do look alike.")
}

Documentation

Index

• func DTC(img image.Image) (phash uint64)
• func Distance(a, b uint64) int
• func FindKeypoints(src image.Image) geometry.Points
• func FromTriangles(src image.Image, triangles []triangle.Triangle) []uint64

Functions

func DTC

func DTC(img image.Image) (phash uint64)

DTC computes the perceptual hash for img using phash dtc image technique.

1. Reduce size to 32x32
2. Reduce color to greyscale
3. Compute the DCT.
4. Reduce the DCT to 8x8 in order to keep high frequencies.
5. Compute the median value of 8x8 dtc.
6. Further reduce the DCT into an uint64.

func Distance

func Distance(a, b uint64) int

Distance returns the distance between two hashes by calculating the number of different bits in the hash

func FindKeypoints

func FindKeypoints(src image.Image) geometry.Points

FindKeypoints returns a list of points that are interesting/key points It does that by detecting corners using cornerdectect.Fast9

func FromTriangles

func FromTriangles(src image.Image, triangles []triangle.Triangle) []uint64

FromTriangles calculates 3 perceptual hash of src per triangles.

A triangle is transformed into it's equilateral version then we run a DTC on the 3 different angles of the triangle.

When two triangle have the same perceptual hash, it means that the features in the triangle are perceptually similar.

Triangles could come from FindKeypoints or your own library.

This function will start a goroutine per CPU.