x-img-diff-js

compare 2 images considering into translation

Downloads in past

Stats

StarsIssuesVersionUpdatedCreatedSize
x-img-diff-js
0.3.55 years ago5 years agoMinified + gzip package size for x-img-diff-js in KB

Readme

x-img-diff-js
CircleCI npm version
JavaScript(Web Assembly) porting project for Quramy/x-img-diff, which extracts structual information of a bit different 2 images.

Demonstration

See https://reg-viz.github.io/x-img-diff-js/

Usage

Node.js

You need Node.js >= v8.0.0
npm install x-img-diff-js pngjs

const fs = require('fs');
const PNG = require('pngjs').PNG;

const detectDiff = require('x-img-diff-js');

function decodePng(filename) {
  return new Promise((resolve, reject) => {
    fs.readFile(filename, (err, buffer) => {
      if (err) return reject(err);
      resolve(PNG.sync.read(buffer));
    });
  });
}

async function main() {
  const [img1, img2] = await Promise.all([
    decodePng('demo/img/actual.png')),
    decodePng('demo/img/expected.png')),
  ]);
  const diffResult = await detectDiff(img1, img2);
  console.log("diff result:", diffResult);
  console.log("the number of matching area:", diffResult.matches.length);
  console.log("img1's macthing area bounding rect:", diffResult.matches[0][0].bounding);
  console.log("img2's matching area bounding rect:", diffResult.matches[0][1].bounding);
  console.log("diff marker rectangulars in img1's matching area", diffResult.matches[0][0].diffMarkers.length);
}

main();

Browser

See demo derectory in this repository.

API

function detectDiff

detectDiff(img1: Image, img2: Image, opt?: DetectDiffOptions): Promise<DetectDiffResult>

  • img1, img2 - Required - Input images.
  • opt - Optional - An object to configure detection.

type Image

type Image = {
  width: number;
  height: number;
  data: Uint8Array;
}

type DetectDiffOptions

A option object. See https://github.com/Quramy/x-img-diff#usage .

type DetectDiffResult

type DetectDiffResult = {
  matces: MatchingRegions[];
  strayingRects: Rect[][];
}

  • matces - An array of each matching region.
  • strayingRects - An array of keypoints recatangle. strayingRects[0] corresponds to img1, strayingRects[1] does to img2.

type MatchingRegions

type MatchingRegions = {
  bounding: Rect;
  center: Rect;
  diffMarkers: Rect[];
}[];

  • bounding - Bounding rectangle of this region.
  • center - Center rectangle of this region.
  • diffMarkers - An array of different parts.

A MatchingRegions is a couple of objects. The 1st corresponds to img1, and 2nd does to img2. And you can get how far the region moved using center property.
// m is an item of DetectDiffResult#mathes
const translationVector = {
  x: m[1].center.x - m[0].center.x,
  y: m[1].center.y - m[0].center.y,
};

type Rect

type Rect = {
  x: number;
  y: number;
  width: number;
  height: number;
}

Represents a rectangle.

function detectDiff.getBrowserJsPath

detectDiff.getBrowserJsPath(): string 

Returns the absolute path of the JavaScript file which should be loaded in Browser env.

function detectDiff.getBrowserWasmPath

detectDiff.getBrowserWasmPath(): string 

Returns the absolute path of the Web Assembly(.wasm) file which should be loaded in Browser env.

How to build module

  1. Clone this repo and change the current directory to it.

  1. Get OpenCV source code

``` git clone https://github.com/opencv/opencv.git cd opencv git checkout 3.1.0 cd .. ```
  1. Get x-img-diff source code

``` git clone https://github.com/quramy/x-img-diff.git ```
  1. Execute docker

$ docker-compose build
$ docker-compose run emcc

Run module in your local machine

python -mhttp.server
open http://localhost:8000/index.html

License

MIT.