IEC 15444-2-05:2010 – JPEG 2000 Part 2 Extensions: Technical Overview of the Compound Image File Format Amendment

Introduction and Scope

The IEC 15444 series is identical to the ISO/IEC 15444 standard that defines the JPEG 2000 image coding system. Part 2, formally titled JPEG 2000 Image Coding System — Part 2: Extensions, extends the core coding technology specified in Part 1. The version referred to as IEC 15444-2-05:2010 corresponds to Amendment 5 to ISO/IEC 15444-2:2004, which was published in 2010. This amendment focuses on the updated encapsulation for the compound image file format, commonly known as the JPX file format.

The scope of IEC 15444-2-05:2010 is to provide a file format that can represent compound images – documents that combine continuous-tone imagery (photographs) with discrete-tone elements such as text, graphics, and line art – in a single, highly compressed, and interoperable file. The amendment also updates and clarifies the box structure of the JPX file format to improve consistency with the broader JPEG 2000 family and to enable more efficient handling of multiple codestreams, layers, and composited scenes.

Beyond the file format extensions, the underlying Part 2 standard includes several advanced coding features:

• Arbitrary wavelet decomposition: permits different wavelet transforms (e.g., reversible 5/3, irreversible 9/7, and custom kernels) per tile or per component.
• Multi-component transforms (MCT): arbitrary linear transforms applied across multiple image components to decorrelate data beyond the standard RGB-to-YCbCr conversion used in Part 1.
• Enhanced region-of-interest (ROI) coding: improved scaling-based ROI enabling selectively higher quality in arbitrary shapes.
• Multiple codestreams: ability to embed several independent JPEG 2000 codestreams within one file to support layering, multi‑view, or variable quality representations.

This article focuses on the key technical aspects introduced by Amendment 5 and the broader Part 2 framework, with an emphasis on compliance considerations for implementers.

Technical Requirements

File Format Extensions (JPX Box Structure)

The amendment updates the JPX file format, which is based on the same box/logical structure as the baseline JP2 file format. A JPX file may contain the following key boxes:

• File Type Box (ftyp): identifies the file as jpx or jpxb (for fragmented codestreams). The amendment adds new brands to facilitate compound image handling.
• Header Box (jp2h): includes image metadata such as resolution, colour space (e.g., sRGB, sYCC, ICC profiles).
• Contiguous Codestream Box (jp2c): carries a single JPEG 2000 codestream. Multiple such boxes may appear for multiple codestreams.
• Compositing Layer Box (comp): defines how multiple codestreams or images are composited into a final rendered scene. This is critical for compound images where a background photograph is overlaid with high‑contrast text.
• Animation Box (anim): optional box supporting simple image sequences (for example, a slideshow of compound images).
• XML Box (xml): permits embedding of metadata in XML for interoperability with other document formats.

The amendment specifically clarifies the semantics of the Compositing Layer Box and the Animation Box to ensure that layers can refer to codestreams via relative offsets and that the rendering order is well defined.

Multi‑Codestream and Compound Image Requirements

When a JPX file contains multiple codestreams, each one must conform to the JPEG 2000 Part 1 or Part 2 syntax, as appropriate. The amendment does not change the codestream syntax but rather the rules for combining codestreams in a single file:

• Each codestream must be carried in a separate jp2c box (or a jp2f fragment box for fragmented codestreams).
• If the file uses compositing, the Compositing Layer Box must provide a complete description of how the decoded image surfaces are combined (blending modes, transparency, and composition order).
• All codestreams must use a consistent colour space for the composite, or the colour transforms defined in the Colour Specification Box must be sufficient to map each component to the common space.

Advanced Transform Parameters

Part 2 allows implementers to define custom wavelet filters and multi‑component transforms within the codestream itself. The amendment does not modify these but reinforces that the associated marker segments (such as MCT marker segments) must be correctly signalled in each codestream.

Implementation Highlights

For developers building encoders or decoders that target IEC 15444-2-05:2010 compliance, the following highlights are critical:

• Compositing Layer Box: The comp box must contain a list of layer descriptors, each referencing a codestream by index, its spatial position (in the reference grid), and a blending rule (standard over/under, additive, or subtractive). The amendment provides a normative C‑pseudo code algorithm for reconstructing the final composite image.
• Animation support: The anim box can specify a sequence of images (from multiple codestreams or the same codestream with a progression) and a frame delay. This is useful for slide‑show presentations of compound images but should not be used for full‑motion video (which is not within the original scope of the standard).
• Arbitrary wavelet decomposition: The decoder must be prepared to load custom wavelet filter coefficients from the codestream’s marker segments. The Part 2 specification defines the exact syntax for such user‑defined filters; the amendment does not alter this. However, conformance testing can be more challenging when non‑standard filters are used.
• Colour consistency: If the JPX file uses multiple codestreams that each specify different colour spaces (e.g., one sRGB, another with an ICC profile), the composition should take place after colorimetric mapping to a common profile. The file format may include a Colour Specification Box with a 3×3 matrix or a look‑up table to perform the transformation at decode time.

Compliance and Testing Notes

To achieve claimed compliance with IEC 15444-2-05:2010, an encoder or decoder must satisfy the conformance requirements specified in the amendment and its parent document, ISO/IEC 15444-2:2004 (which itself incorporates several earlier amendments). The main compliance areas are:

• Codestream conformance: All codestreams within a JPX file must conform to ISO/IEC 15444-1 (core codestream) or ISO/IEC 15444-2 (extended codestream) as appropriate. In practice, if any codestream uses Part‑2‑only features (custom wavelet, MCT beyond 3×3), the entire file should be labelled with the correct brand.
• File format conformance: The JPX box structure must follow the normative syntax defined in Annex A. The amendment introduced strict rules for ordering: for example, the Compositing Layer Box must appear before any codestream that it references. A compliant decoder shall reject files where a reference occurs before a necessary box.
• Test vectors: The ISO/IEC JTC 1/SC 29/WG 1 committee provides a set of conformance test streams. For Amendment 5, the test suite includes several JPX files with two or three codestreams, non‑trivial compositing, and one animation file. Conformance can be verified by comparing the decoder output (decoded pixel data per layer and final composite) against the reference images.
• Interoperability: Since the compound image file format is designed for archiving, legal documents, and scanned manuscripts, the compliance statement should specify the supported features (e.g., “this product supports JPX up to 4 codestreams and full compositing as per IEC 15444-2-05”).

Frequently Asked Questions