mint3000 / imghash

Image perceptual hash calculation for node

imghash

Promise-based image perceptual hash calculation for node.

Installation

npm install imghash

Basic usage

const imghash = require('imghash');

imghash
  .hash('path/to/file')
  .then((hash) => {
    console.log(hash); // 'f884c4d8d1193c07'
  });

// Custom hex length and result in binary
imghash
  .hash('path/to/file', 4, 'binary')
  .then((hash) => {
    console.log(hash); // '1000100010000010'
  });

Finding similar images

To measure similarity between images you can use Hamming distance or Levenshtein Distance. Here's an example of using the first one:

const imghash = require('imghash');
const hamming = require('hamming-distance');

const hash1 = imghash.hash('./img1');
const hash2 = imghash.hash('./img2');

Promise
  .all([hash1, hash2])
  .then((results) => {
    const dist = hamming(results[0], results[1]);
    console.log(`Distance between images is: ${dist}`);
    if (dist <= 20) {
      console.log('Images are similar');
    } else {
      console.log('Images are NOT similar');
    }
  });

API

.hash(filepath[, bits][, format])

Returns: ES6 Promise, resolved returns hash string in specified format and length (eg. f884c4d8d1193c07)

Parameters:

• filepath - path to the image (supported formats are png and jpeg) or Buffer
• bits (optional) - hash length [default: 8]
• format (optional) - output format [default: hex]

===

.hashRaw(data, bits)

Returns: hex hash

Parameters:

• data - image data descriptor in form { width: [width], height: [height], data: [decoded image pixels] }
• bits - hash length

===

.hexToBinary(s)

Returns: hex string, eg. f884c4d8d1193c07.

Parameters:

• s - binary hash string eg. 1000100010000010

===

.binaryToHex(s)

Returns: hex string, eg. 1000100010000010.

Parameters:

• s - hex hash string eg. f884c4d8d1193c07

Further reading

imghash takes advantage of block mean value based hashing method:

• http://stackoverflow.com/questions/14377854/block-mean-value-hashing-method
• http://commonsmachinery.se/2014/09/digital-image-matching-part-1-hashing/