Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
IEC 62665 addresses a compelling challenge: enabling blind and visually impaired people to access printed documents without requiring specialised Braille literacy. The standard defines a texture map system — a two-dimensional cell pattern printed alongside conventional text on ordinary paper — that can be scanned by a pen-shaped device and converted to speech via text-to-speech technology. This International Standard, developed by IEC Technical Committee 100 (Audio, video and multimedia systems and equipment), bridges the gap between traditional print publishing and digital accessibility, creating a universal format where the same printed page serves both sighted and visually impaired readers simultaneously.
The texture map is a two-dimensional array of cells printed on paper alongside the regular text. Each texture map consists of the following elements:
| Element | Description | Function |
|---|---|---|
| Cell | Minimum square element (smallest printed dot) | Basic binary information unit (black = 1, white = 0) |
| Unit | 11 × 11 cell square area | Structured data block containing encoded character information |
| Data matrix | Two-dimensional cell patterns from encoded text data | Contains the compressed and error-corrected text content |
| Alignment line | Solid border line with tick marks surrounding the texture map | Guides the scanning pen to recognise the texture map boundary and orientation |
Text is first converted from its character representation into a compressed binary stream. The compression algorithm combines Pack processing and LZSS (Lempel-Ziv-Storer-Szymanski) processing to reduce data volume. The compressed data then undergoes Reed-Solomon error correction encoding using Galois field GF(2048), which allows the system to recover the original text even if the printed texture map is partially smudged, scratched, or degraded.
IEC 62665 defines two distinct processing workflows — encoding (text to texture map) and decoding (texture map to speech):
The encoding process transforms source text into a printed texture map through the following stages:
The decoding reverses this process. The user passes a pen-shaped scanning device over the texture map. The device captures the cell pattern image, applies image processing to extract the data matrix, runs error correction, decompresses the data, and feeds the recovered text to a text-to-speech engine for audio output.
| Stage | Input | Output | Key Algorithm |
|---|---|---|---|
| Image capture | Printed texture map | Raw pixel image | Optical scanning at ≥ 600 dpi |
| Symbol extraction | Raw pixel image | Binary matrix | Alignment line detection and perspective correction |
| Error correction | Binary matrix with errors | Corrected binary data | Reed-Solomon GF(2048) |
| Decompression | Compressed binary | Encoded text stream | LZSS + Pack decompression |
| Speech synthesis | Text with pronunciation hints | Audio output | Text-to-speech engine |
The standard defines precise requirements for the printed texture map to ensure reliable scanning:
| Parameter | Specification | Rationale |
|---|---|---|
| Printing resolution | ≥ 600 dpi | Enough to resolve individual cells while compatible with standard office printers |
| Texture map size | Varies by data volume (Table 1 in standard) | Proportional to text length; approximately 1 cm² per 100 characters |
| Margin from edge | ≥ 10 mm | Prevents clipping during scanning and binding |
| Alignment line width | 1 cell width | Minimum detectable line for pen sensor |
| Notch position | Top-left corner of alignment frame | Identifies orientation and start of data read |
The standard is designed to be language-independent at the core encoding level, but the annexes specifically address Japanese and English text pre-processing. The SpeechioEncode and SpeechioSymbol functions provide the framework for encoding character sets with different phonetic requirements. For languages not explicitly covered, the encoding scheme can be adapted by defining appropriate character-to-speech mapping tables following the patterns established in the provided annexes.
While both are 2D matrix codes, they serve fundamentally different purposes. A QR code typically encodes a URL or short text (up to a few thousand characters) and redirects the user to digital content via a smartphone. The IEC 62665 texture map is designed to encode the full text content of a printed document (potentially tens of thousands of characters) directly on the page, with embedded speech control codes for direct text-to-speech conversion. The texture map is also optimised for pen-based scanning rather than camera-based scanning.
IEC 62665 was published in 2012 (first edition) and revised in 2015 (second edition). Given the rapid evolution of mobile accessibility technology (smartphone OCR, AI-powered text-to-speech), the texture map approach represents a specific niche: low-tech, low-cost accessibility for printed materials where smartphones may not be available or appropriate. The technology is patented by Original Design Inc. and Kosaido Co., Ltd. in Japan, which has influenced its adoption pattern.
Technically, yes — a sticker or tip-in page with the texture map could be added to existing books. However, the practical value depends on the book type. For books with static content (legislation, technical manuals, academic texts), retrofitting is feasible. For frequently updated content, printing the texture map as part of the original production process is far more cost-effective. The standard includes provisions for the notch marking system that can identify the texture map position on any page, enabling partial retrofitting of selected pages.