Prior Art That Invalidated Forgent's JPEG Patent

In the mid-2000s, Forgent Networks aggressively asserted U.S. Patent No. 4,698,672 against hundreds of technology companies, claiming its compression methodology covered the baseline Joint Photographic Experts Group (JPEG) standard. The assertion campaign collapsed after public interest groups and defendants challenged the patent's validity at the United States Patent and Trademark Office (USPTO). By presenting extensive prior art documenting the combination of Discrete Cosine Transform, quantization, and run-length variable-length coding prior to the patent's 1986 priority date, challengers proved that the core claims asserted against JPEG were invalid.

The '672 Patent and the JPEG Controversy

U.S. Patent No. 4,698,672, titled "Coding system for reducing redundancy," was filed on October 27, 1986, by inventors associated with Compression Labs, Inc. (CLI), a company later acquired by Forgent Networks. The patent described a technique for compressing digitized image data, particularly video signals.

Forgent began asserting the patent in 2002, targeting digital cameras, scanners, web browsers, and any hardware or software producing baseline JPEG files. Forgent argued that Claim 19—a broad claim directed at processing blocks of pixel data through transformation, thresholding, and variable-length encoding—covered the standard JPEG compression pipeline. The company extracted over $100 million in royalties before facing organized legal challenges.

Key Prior Art Cited Against the Patent

In 2004, the Electronic Frontier Foundation (EFF), along with Public Knowledge and defendant tech companies, filed requests for patent reexamination at the USPTO. The invalidation of Claim 19 and related claims rested primarily on existing technical literature from the late 1970s and early 1980s that predated the patent's October 1986 filing date:

• Early DCT Transform Coding Papers: The core mathematical framework of the JPEG standard relies on the Discrete Cosine Transform (DCT), developed by Nasir Ahmed in 1974. Prior art cited in the reexamination demonstrated that using two-dimensional DCT to transform image pixel blocks into frequency components was already established practice. Studies published by researchers like Wen-Hsiung Chen and William K. Pratt from 1979 to 1984 comprehensively described using DCT combined with thresholding techniques to reduce redundancy in digital images.
• The 1983 CCITT Standardization Documents: International standards bodies had already explored transform coding for video conferencing. Documents submitted in 1983 to the International Telegraph and Telephone Consultative Committee (CCITT, now ITU-T) Study Group XV detailed systems that split frames into blocks, applied 2D-DCT, and quantized the coefficients. These working papers outlined the exact functional sequence claimed by the '672 patent.
• Prior Publications on Run-Length and Huffman Coding of DCT Coefficients: Claim 19 claimed novelty in ordering quantized coefficients and coding both zero-runs and nonzero values using variable-length codes. The EFF presented prior publications—including technical papers on adaptive transform coding from 1981 through 1984—showing that scanning DCT coefficients in order of importance, tracking sequences of zero-value high frequencies, and applying Huffman coding was known technology in image transmission.
• Earlier Video Compression Patents and Systems: Systems developed by competing telecommunications researchers in the early 1980s, including prior patents assigned to institutions like Bell Labs and other compression researchers, disclosed thresholding and scanning algorithms identical in scope to the methods Forgent claimed as exclusive inventions.

USPTO Rejection and Aftermath

In 2006, the USPTO issued an initial rejection and later affirmed a final rejection of the broadest claims of the '672 patent, specifically rejecting Claim 19 as either anticipated or rendered obvious by the cited prior art. The examiner determined that combining block-based DCT, quantization, zero-run coding, and variable-length coding was fully disclosed in public literature before CLI filed its patent application.

Following the USPTO's determinations and unfavorable rulings in multidistrict litigation in the U.S. District Court for the Northern District of California, Forgent Networks settled its remaining lawsuits. The invalidation of the asserted claims established that the baseline JPEG compression process was free of Forgent's patent claims.