web-digital-watermarking

Embed text in an image's frequency domain and reveal it through the image's Fourier spectrum. The browser implementation is self-contained and does not load or bundle OpenCV.js.

Install

npm install web-digital-watermarking

This package is intended for browsers. For Node.js, use node-digital-watermarking.

Usage

const dw = require('web-digital-watermarking');

const srcImageUrl = 'http://localhost:8080/srcImg.png';
const watermarkText = 'github.com/zy445566';
const fontSize = 1.1;

async function run() {
    const encodedImageUrl = await dw.transformImageUrlWithText(
        srcImageUrl,
        watermarkText,
        fontSize
    );
    const encodedImage = document.createElement('img');
    encodedImage.src = encodedImageUrl;
    document.body.appendChild(encodedImage);

    const spectrumImageUrl = await dw.getTextFormImageUrl(encodedImageUrl);
    const spectrumImage = document.createElement('img');
    spectrumImage.src = spectrumImageUrl;
    document.body.appendChild(spectrumImage);
}

run();

Images loaded from another origin must be served with an appropriate CORS header because the implementation reads their pixels through Canvas.

Implementation

• Radix-2 FFT handles power-of-two dimensions.
• Bluestein's algorithm handles arbitrary image dimensions.
• A two-dimensional FFT embeds the text mask into conjugate frequency bins.
• The inverse FFT reconstructs the encoded image without native or WebAssembly dependencies.