Skip to content

img_hash.md

Latest commit

 

History

History
42 lines (35 loc) · 1.57 KB

img_hash.md

File metadata and controls

42 lines (35 loc) · 1.57 KB

img_hash Module (contrib)

Perceptual image hashing for similarity comparison.

ImgHashBase

const ImgHashBase = struct {
    fn createAverageHash() !ImgHashBase;
    fn createPHash() !ImgHashBase;
    fn createBlockMeanHash(mode: i32) !ImgHashBase;
    fn createMarrHildrethHash(alpha: f32, scale: f32) !ImgHashBase;
    fn createRadialVarianceHash(sigma: f64, numOfAngleLine: i32) !ImgHashBase;
    fn createColorMomentHash() !ImgHashBase;
    fn deinit(self: *ImgHashBase) void;
    fn compute(self: ImgHashBase, input: Mat, output: *Mat) void;
    fn compare(self: ImgHashBase, hashOne: Mat, hashTwo: Mat) f64;
};

Algorithms

Algorithm Speed Robustness Best For
AverageHash ⚡⚡⚡ Basic Quick duplicate detection
PHash ⚡⚡⚡ Good General similarity matching
BlockMeanHash ⚡⚡ Good Rotation-tolerant matching
MarrHildrethHash ⚡⚡ Very Good Edge-based similarity
RadialVarianceHash ⚡ Excellent Rotation-invariant
ColorMomentHash ⚡⚡ Good Color-based similarity

Example

var hasher = try cv.contrib.img_hash.ImgHashBase.createPHash();
defer hasher.deinit();
var hash1 = try cv.Mat.init(); defer hash1.deinit();
var hash2 = try cv.Mat.init(); defer hash2.deinit();
hasher.compute(img1, &hash1);
hasher.compute(img2, &hash2);
const distance = hasher.compare(hash1, hash2); // 0 = identical