CCITT Compression: Group 4 Fax Encoding in PDF

What Is CCITT Compression?

The name comes from the Comité Consultatif International Téléphonique et Télégraphique — the international committee that originally standardized it for fax machine transmission in the 1980s. Today, the standard is maintained by the ITU (International Telecommunication Union) and referred to as ITU-T T.4 (Group 3) and T.6 (Group 4).

The core insight behind CCITT is simple: most of a scanned black-and-white page is white. Rather than storing every individual white pixel, the algorithm records run lengths — for example, "the next 24,000 pixels are white." This makes the compressed data far smaller than the raw image data, with no information lost at all.

In the PDF specification, CCITT compression is applied through the CCITTFaxDecode filter. When a PDF viewer opens a page with scanned content, it reads this filter, decompresses the bitonal stream, and renders the image. The standard is supported by every PDF reader ever made — from Adobe Acrobat to browser-based viewers.

How CCITT Group 4 Works

Group 4 uses two-dimensional coding, which is what makes it so efficient. Here is the process from scan to compressed PDF:

1. The scanner produces a bitonal bitmap. Every pixel is either 0 (white) or 1 (black). A standard A4 page at 300 DPI produces about 8.5 million pixels — nearly 1 MB of raw data before any compression.
2. Run-length encoding handles each row. Instead of writing every pixel, the encoder writes the length of each alternating run. A row that starts with 800 white pixels then 12 black pixels is stored as two numbers: (800, 12).
3. Two-dimensional coding compares adjacent rows. Group 4 looks at each line relative to the line above it. Because consecutive lines of text are often nearly identical, only the differences between lines need to be stored — dramatically reducing data.
4. The result is stored as a CCITTFaxDecode stream in the PDF. The PDF stores the compressed bitonal data along with a filter dictionary specifying K: -1 (indicating Group 4), image dimensions, and color space.

The PDF Filter Dictionary

Here is what the CCITT filter parameters look like inside a PDF stream:

<<
  /Type          /XObject
  /Subtype       /Image
  /Width         2551          % pixels wide  (A4 @ 300 DPI)
  /Height        3508          % pixels tall
  /ColorSpace    /DeviceGray
  /BitsPerComponent 1         % bitonal: 1 bit per pixel
  /Filter        /CCITTFaxDecode
  /DecodeParms   << /K -1 /Columns 2551 >>
>>
% K = -1  →  Group 4 (T.6)
% K =  0  →  Group 3 1D (T.4)
% K >  0  →  Group 3 2D (T.4)

Group 3 vs. Group 4 at a Glance

Real-World Examples

Benefits of CCITT Group 4

CCITT vs. JBIG2: Which to Use?

CCITT Group 4 has dominated bitonal PDF compression for decades, but JBIG2 — introduced in the PDF 1.4 specification — can sometimes produce even smaller files. Here is how they compare:

Common Mistakes to Avoid

• Applying CCITT to grayscale or color scans. CCITT only works on bitonal images. Trying to use it on a grayscale photo will produce errors or incorrect output. Use Flate or JPEG for those.
• Scanning at too low a resolution before compression. CCITT is not a cure for a blurry scan. For readable text, scan at a minimum of 200 DPI — 300 DPI is the standard. CCITT will faithfully preserve whatever resolution you capture.
• Confusing the K parameter values. K=0 is Group 3 one-dimensional (not Group 4). For maximum compression, you need K=-1. This is a common source of confusion in custom PDF generators and libraries.
• Using CCITT for documents with mixed content. If a page contains both a black- and-white scan and a color photo or logo, CCITT can only handle the bitonal regions. Mixed-content PDFs typically combine multiple compression filters — one per image object.
• Assuming all scanners output true bitonal images. Many consumer scanners dither grayscale into pseudo-bitonal images. This reduces CCITT efficiency dramatically. Use a dedicated document scanner with a hardware bitonal (1-bit) output mode.

Frequently Asked Questions

• What is CCITT compression?

  CCITT compression is a lossless encoding method designed for bitonal (pure black-and-white) images. Originally developed for fax machines, it is now the standard for compressing scanned documents in PDF files via the CCITTFaxDecode filter. Group 4 is the most efficient variant and produces no quality loss whatsoever.

• What does CCITTFaxDecode mean in a PDF?

  CCITTFaxDecode is the PDF filter name for CCITT compression. When a PDF reader encounters this filter, it decompresses the bitonal image data using the CCITT algorithm. The K parameter in the filter dictionary determines which variant is used: K=0 is Group 3 1D, K>0 is Group 3 2D, and K=-1 is Group 4.

• What is the difference between CCITT Group 3 and Group 4?

  Group 3 uses one-dimensional run-length encoding and was designed for fax transmission over noisy phone lines — it includes error correction that adds overhead. Group 4 uses two-dimensional encoding, comparing each scan line to the previous one, and drops error correction since it assumes a reliable channel. Group 4 produces significantly smaller files and is the standard for PDF documents.

• Is CCITT compression lossless?

  Yes, completely. CCITT Group 4 is always lossless — every black and white pixel is reproduced identically after decompression. No information is discarded. This is critical for legal and medical documents where text sharpness and pixel-perfect accuracy are required by law or regulation.

• When should I use CCITT instead of JBIG2?

  Use CCITT Group 4 when maximum compatibility and guaranteed accuracy matter — legal filings, medical records, government archives, or any context where a single corrupted character is unacceptable. Use JBIG2 when file size is the top priority and you know your audience uses modern PDF software. JBIG2 can achieve better compression but carries compatibility risks with older systems.

• Can CCITT compression be used on color images?

  No. CCITT compression only works on bitonal images — exactly two values per pixel: black or white. It cannot be applied to grayscale or color images. For those, use JPEG (DCTDecode), JPEG 2000 (JPXDecode), or Flate (FlateDecode) compression inside the PDF.

Related Glossary Terms