A blazing-fast and compact JavaScript library dedicated to efficiently decoding JPEG images.

Using npm:

npm install jay-peg

Using yarn:

yarn add jay-peg

Use the decoder providing a JPEG data buffer as input.

import JPEG from 'jay-peg';

const jpegBuffer = /* your JPEG buffer here */;
const imageMarkers = JPEG.decoder(jpegBuffer);

console.log(imageMarkers);

The output consists of a structured array of image markers:

[
  {
    type: 65496,
    name: "SOI",
  },
  {
    type: 65505,
    name: "EXIF",
    length: 16382,
    identifier: "Exif\x00\x00",
    entries: [Object],
  },
  {
    type: 65499,
    name: "DAC",
    length: 132,
    tables: [[Object], [Object]],
  },
  // ... and so forth
  {
    type: 65497,
    name: "EOI",
  },
];

The decoder function accepts a JPEG buffer as its sole argument and returns an array of image markers.

• buffer: A Buffer or Uint8Array containing the JPEG image data.

An array of objects representing various markers found in the JPEG image.

Each ImageMarker object in the output array adheres to the following structure:

• type (Number): The marker type.
• name (String): The marker name.
• length (Number): The length of the marker data.
• Additional properties specific to certain marker types.

Performance is a key focus of jay-peg. 4 sizes of images were benchmarked:

• small: 300 × 150, 8KB image
• medium: 800 × 600, 70KB image
• large: 1920 × 1080, 332KB image
• huge: 2448×3264, 2.2MB image

For each of these, the decoding speed was measured as follows:

Benchmarked: small:  x 13,393 ops/sec ±4.77% (96 runs sampled)
Benchmarked: medium:  x 12,894 ops/sec ±0.10% (99 runs sampled)
Benchmarked: large:  x 9,241 ops/sec ±0.25% (99 runs sampled)
Benchmarked: huge:  x 2,672 ops/sec ±0.12% (100 runs sampled)

Measures were taken in an MacBook Air 2024, Apple M3 w/16GB of RAM.

jay-peg is released under the MIT License