Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
ISO/IEC 16485-04, adopted in Canada as CAN/CSA ISO/IEC 16485-04, defines the Mixed Raster Content (MRC) model for representing and compressing documents that contain a combination of text, line art, continuous-tone images, and graphics. This standard aligns with ITU-T Recommendation T.44 and provides the foundational architecture used in modern document imaging workflows, including PDF, network scanners, digital copiers, and archival systems.
The MRC model enables significant compression gains — often 5–10 times over single-method compression — by decomposing a page into several layers, each optimized for a specific content type. The standard covers both the encoding syntax and the decoding process, ensuring interoperability across different implementations.
Under the MRC model, a page is divided into a sequence of stripes, each stripe containing up to three layers:
| Layer | Type | Compression Method | Typical Content |
|---|---|---|---|
| Background (Layer 0) | Continuous-tone | JPEG, JPEG 2000 | Photographs, gradients |
| Foreground (Layer 1) | Continuous-tone | JPEG, JPEG 2000 | Colored text, logos, line art |
| Mask (Layer 2) | Binary | JBIG2, G4 (MMR) | Text shapes, thin lines |
The mask layer controls the pixel-by-pixel selection between the background and foreground layers, enabling sharp edges for text while maintaining continuous-tone quality in image regions.
Each page is partitioned into horizontal stripes (typically 256 lines high). Each stripe is independently encoded, which allows parallel processing and partial decoding for viewport navigation. Stripes are further divided into blocks that can hold different types of data, including instructions for layer merging and color space conversions.
The standard requires that background and foreground layers be coded with a lossy or lossless continuous-tone algorithm (e.g., JPEG, JPEG 2000, or deflate), while the mask layer must use a binary compression method (JBIG2 preferred for text, G4 for legacy compatibility). Key requirements include:
A conforming encoder must perform page segmentation to assign each pixel to one of the three layers. This segmentation is the most computationally intensive part of the encoding. The standard does not mandate a specific segmentation algorithm, leaving room for innovation, but the output bitstream must conform to the MRC syntax as described in the normative clauses.
Decoding follows a strict pipeline: decode each layer independently, then composite using the mask. The compositing rule is: output = mask ? foreground : background. Decoders must support all allowed compression algorithms and color spaces; fallback modes are allowed for unsupported methods only if the encoder sets a compatibility flag.
For a product to claim compliance with ISO/IEC 16485-04 (or CAN/CSA ISO/IEC 16485-04), it must pass conformance testing as defined in the standard’s Annex A. Key compliance checks include:
Although self-declaration is permitted, certification bodies such as CSA Group offer formal verification programs. Many enterprise contracts and government procurement requirements mandate CSA-certified implementations for document archiving and exchange.
ISO/IEC 16485-04 and CAN/CSA ISO/IEC 16485-04 are copyright protected by their respective standards bodies. This article provides general technical guidance and does not substitute the official normative text. © 2026