JBIG2

JBIG2 is a lossless and optionally lossy compression standard for bi-level images. It was developed by the Joint Bi-level Image Experts Group and published in 2000. Unlike earlier standards like CCITT Group 4, JBIG2 leverages similarities between characters and symbols within a document to achieve higher compression ratios. It works by identifying repeating patterns of pixels and representing them efficiently.

The standard offers two primary compression methods: lossless and lossy. The lossless mode ensures perfect reconstruction of the original image, making it suitable for applications where data integrity is paramount. The lossy mode, on the other hand, can achieve significantly higher compression by discarding some image information that is perceptually insignificant. This is particularly effective for scanned documents containing text and line drawings, where minor variations in identical characters are unlikely to be noticed by the reader.

JBIG2's lossy compression relies on a technique called "symbol coding." It first segments the image into regions and identifies unique symbols or patterns. These unique symbols are then stored once in a dictionary, and subsequent occurrences of the same symbol are replaced by references to its entry in the dictionary. For lossy compression, a degree of similarity is allowed when matching symbols, leading to the creation of a more compact dictionary and further compression.

The standard has been incorporated into the PDF file format, allowing for efficient storage and transmission of scanned documents. While it offers substantial improvements in compression efficiency over older methods, the complexity of its implementation and potential for data loss in lossy mode are considerations for its application.

---