3:08-cv-02912
Nuance Communications Inc v. Abbyy Software House
I. Executive Summary and Procedural Information
- Parties & Counsel:- Plaintiff: Nuance Communications, Inc. (Delaware)
- Defendant: ABBYY Software House et al. (Russia, California, Cyprus); Lexmark International, Inc. (Delaware)
- Plaintiff’s Counsel: Wilson Sonsini Goodrich & Rosati
 
- Case Identification: 3:08-cv-02912, N.D. Cal., 04/03/2009
- Venue Allegations: Venue is asserted based on Defendants transacting business, selling, and offering for sale accused products within the Northern District of California through retail stores, online stores, and third-party retailers.
- Core Dispute: Plaintiff alleges that Defendants’ optical character recognition (OCR) software, printers, and document management platforms infringe eight patents related to OCR, context analysis, and document processing technologies.
- Technical Context: The dispute centers on optical character recognition, a technology that converts scanned images of text into machine-encoded text, which is fundamental for digitizing documents and enabling search and editing capabilities.
- Key Procedural History: This filing is an Amended Complaint, indicating it supersedes a prior version. The complaint asserts a broad portfolio of patents with priority dates spanning from the late 1980s to the early 2000s against products from two distinct sets of defendants, ABBYY (a software developer) and Lexmark (a hardware and software provider). The complaint also includes non-patent claims for trade dress infringement against the ABBYY defendants.
Case Timeline
| Date | Event | 
|---|---|
| 1988-08-10 | Priority Date for ’053, ’489, ’983, and ’342 Patents | 
| 1992-07-14 | U.S. Patent No. 5,131,053 Issues | 
| 1993-11-09 | U.S. Patent No. 5,261,009 Issues | 
| 1995-01-10 | U.S. Patent No. 5,381,489 Issues | 
| 1995-07-25 | U.S. Patent No. 5,436,983 Issues | 
| 2000-03-14 | U.S. Patent No. 6,038,342 Issues | 
| 2001-06-01 | Nuance alleges first use of its packaging trade dress | 
| 2002-06-01 | ABBYY allegedly begins using infringing trade dress | 
| 2004-05-25 | U.S. Patent No. 6,742,161 Issues | 
| 2004-10-26 | U.S. Patent No. 6,810,404 Issues | 
| 2004-11-16 | U.S. Patent No. 6,820,094 Issues | 
| 2009-04-03 | Amended Complaint Filed | 
II. Technology and Patent(s)-in-Suit Analysis
U.S. Patent No. 5,131,053, “Optical Character Recognition Method and Apparatus,” Issued July 14, 1992 (’053 Patent)
The Invention Explained
- Problem Addressed: The patent describes prior art OCR systems as requiring significant manual intervention to process documents, such as marking off columns of text or distinguishing text from graphics, which limited their efficiency and utility for complex page layouts (’053 Patent, col. 1:24-51).
- The Patented Solution: The invention proposes a fully automated system that scans a document to create a bit-mapped image and then "parses" the page to logically distinguish between text and non-text areas without human input (’053 Patent, col. 2:7-24). This parsing is achieved by analyzing the "texture" of small areas, such as the local density of black-to-white transitions, to characterize them as text, graphics, or rulings, which are then grouped into blocks for recognition in the correct reading order (’053 Patent, col. 5:5-25).
- Technical Importance: This automated page layout analysis represented a step toward enabling high-throughput, user-friendly OCR for documents containing complex layouts with mixed text and images, a common feature in magazines, newspapers, and business reports (’053 Patent, col. 1:45-51).
Key Claims at a Glance
- The complaint does not identify any specific claims it asserts, stating only that Defendants infringe "one or more claims" (Compl. ¶50). For the purpose of this analysis, independent claim 1 is examined.
- Claim 1 (Method) Elements:- Optically scanning a medium to produce a bit-mapped image.
- Parsing the page into a plurality of blocks to determine a logical reading order.
- Passing the blocks to a character recognition process in that logical order.
- Recognizing characters in the blocks.
- Providing as output data representing the recognized characters.
 
U.S. Patent No. 5,381,489, “Optical Character Recognition Method and Apparatus,” Issued January 10, 1995 (’489 Patent)
The Invention Explained
- Problem Addressed: As with the parent ’053 Patent, this invention addresses the need for an automated method to distinguish text from graphics on a scanned page to enable efficient OCR processing (’489 Patent, col. 1:42-51).
- The Patented Solution: This patent focuses on a specific method for distinguishing text from non-text areas by creating a "text map array." This is accomplished by analyzing statistical properties of a bit-mapped image, specifically by determining the "number of on-bits" (pixel density) and the "number of phase changes" (transitions from black-to-white) within discrete segments of the image (’489 Patent, col. 2:63-68; col. 7:15-34). If these values fall within predetermined ranges characteristic of text, the corresponding area in the text map array is identified as text.
- Technical Importance: This statistical, texture-based analysis provides a computational method for automatically segmenting a document page, forming a foundational step for the automated layout analysis described in the ’053 Patent (’489 Patent, col. 8:1-18).
Key Claims at a Glance
- The complaint does not identify any specific claims it asserts (Compl. ¶55). For the purpose of this analysis, independent claim 1 is examined.
- Claim 1 (Method) Elements:- Identifying an array element as text by:- Determining if the number of "on-bits" in a first predetermined number of bytes is within a first predetermined range.
- Determining if the number of "phase changes" from on-bits to off-bits in those bytes is within a second predetermined range.
 
 
- Identifying an array element as text by:
Multi-Patent Capsules
- U.S. Patent No. 5,436,983, “Optical character recognition method and apparatus,” Issued July 25, 1995 (’983 Patent) - Technology Synopsis: This patent, from the same family as the ’053 and ’489 patents, also addresses the automated parsing of scanned documents. It claims a character recognition process that combines template matching with feature analysis, allowing the system to recognize a wide variety of fonts while maintaining processing speed (’983 Patent, col. 2:25-36).
- Asserted Claims: "one or more claims" (Compl. ¶60).
- Accused Features: The ABBYY FineReader products and Lexmark devices performing OCR are accused of infringement (Compl. ¶¶60-61).
 
- U.S. Patent No. 6,038,342, “Optical character recognition method and apparatus,” Issued March 14, 2000 (’342 Patent) - Technology Synopsis: Continuing from the same family, this patent discloses a "context analysis" process that occurs after initial character recognition. This process resolves ambiguities in character shapes (e.g., distinguishing an "S" from an "s") by examining the relative sizes and positions of surrounding characters to determine context, such as identifying words or discerning between upper and lower case letters (’342 Patent, col. 3:26-32; col. 25:5-13).
- Asserted Claims: "one or more claims" (Compl. ¶65).
- Accused Features: The ABBYY FineReader products and Lexmark devices performing OCR are accused of infringement (Compl. ¶¶65-66).
 
- U.S. Patent No. 5,261,009, “Means for resolving ambiguities in text passed upon character context,” Issued November 9, 1993 (’009 Patent) - Technology Synopsis: This patent focuses on resolving ambiguities in OCR output by applying a probabilistic model. It describes a method of assigning a probability value to different possible character strings (e.g., "Onions" vs. "Onions") based on the statistical likelihood of character sequences (n-grams) in the English language, and then selecting the string with the highest probability (’009 Patent, col. 2:9-18).
- Asserted Claims: "one or more claims" (Compl. ¶70).
- Accused Features: The ABBYY FineReader products and Lexmark devices performing OCR are accused of infringement (Compl. ¶¶70-71).
 
- U.S. Patent No. 6,810,404, “Computer-based document management system,” Issued October 26, 2004 (’404 Patent) - Technology Synopsis: This patent shifts focus from pure OCR to document management. It describes a system for automatically importing, indexing, categorizing, and archiving electronic documents. A key feature is the creation of a standardized "STG file" for each document, which stores metadata and extracted text, enabling efficient searching and organization regardless of the original document's format (’404 Patent, col. 3:59-4:24).
- Asserted Claims: "one or more claims" (Compl. ¶75).
- Accused Features: ABBYY's FlexiCapture and FormReader, and Lexmark's DocMP and Document Distributor products are accused of infringement (Compl. ¶¶75-76).
 
- U.S. Patent No. 6,820,094, “Computer-based document management system,” Issued November 16, 2004 (’094 Patent) - Technology Synopsis: This patent is related to the ’404 Patent and further details a document management system. It discloses a method for organizing documents into a hierarchical collection of electronic folders and "clipped documents" (compound documents), where the relationships are managed by a central "DCO file." This system uses a "virtual document storage" concept where a single document file can be linked in multiple folders without duplication (’094 Patent, col. 5:39-6:1).
- Asserted Claims: "one or more claims" (Compl. ¶80).
- Accused Features: ABBYY's FlexiCapture and FormReader, and Lexmark's DocMP and Document Distributor products are accused of infringement (Compl. ¶¶80-81).
 
- U.S. Patent No. 6,742,161, “Distributed computing document recognition and processing,” Issued May 25, 2004 (’161 Patent) - Technology Synopsis: This patent describes a network-based document processing system. It discloses a method where a user can submit a document from a local terminal (e.g., via a scanner or e-mail) to a remote server, which then performs processing functions like OCR and returns the processed document to the user. This architecture allows users to access powerful processing capabilities without needing to install the requisite software locally (’161 Patent, col. 1:5-13; Abstract).
- Asserted Claims: "one or more claims" (Compl. ¶85).
- Accused Features: ABBYY's Recognition Server and Lexmark's Document Distributor products are accused of infringement (Compl. ¶¶85-86).
 
III. The Accused Instrumentality
- Product Identification: The complaint accuses two categories of products: - ABBYY Products: A suite of OCR and data capture software, including FineReader, FlexiCapture, FormReader, Recognition Server, and others (Compl. ¶21).
- Lexmark Products: A wide range of Lexmark printers and multifunction devices, as well as document management software platforms such as Lexmark Document Management Platform ("DocMP") and Lexmark Document Distributor (Compl. ¶¶42, 51, 61).
 
- Functionality and Market Context: - The accused ABBYY products are alleged to provide functionality for performing OCR, which involves converting images of text into editable and searchable data (Compl. ¶50). Products like FlexiCapture and Recognition Server are aimed at enterprise-level data capture from documents like forms and invoices (Compl. ¶¶75, 80, 85).
- The accused Lexmark products are hardware devices (printers, all-in-ones) that allegedly incorporate OCR functionality, as well as standalone software platforms for managing and distributing documents within an organization (Compl. ¶¶51, 76, 81, 86).
- The complaint alleges that these products are sold and marketed throughout the United States, including in the Northern District of California, through various channels including retail stores like Fry's, online stores, and third-party retailers like Amazon.com (Compl. ¶¶35, 38, 43, 46).
 
IV. Analysis of Infringement Allegations
The complaint does not provide sufficient detail for a claim chart analysis. For each of the eight patents-in-suit, the infringement allegations are pleaded in a generalized narrative format. The complaint alleges that the ABBYY and Lexmark product lines infringe "one or more claims" of each patent but fails to identify any specific claims or map any particular product feature to any specific claim element (e.g., Compl. ¶¶50, 55). The allegations broadly state that the accused products perform OCR or document management functions that fall within the scope of the patents. For instance, regarding the '053 Patent, the complaint alleges that the "FineReader line of products" infringes by "performing OCR" and that various Lexmark devices infringe by being "devices for performing OCR" (Compl. ¶¶50-51). This level of generality is repeated for all asserted patents.
No probative visual evidence provided in complaint.
- Identified Points of Contention:- Scope Questions: Due to the lack of specificity, the primary point of contention will be establishing any connection between the accused products and the patent claims. A foundational question for the court will be whether the complaint's allegations meet the pleading standards for patent infringement under Federal Rules of Civil Procedure as interpreted by the Supreme Court in Iqbal and Twombly.
- Technical Questions: A central technical question will be whether the specific methods of page parsing, text/graphic discrimination, context analysis, and document management recited in the asserted patents are actually practiced by the accused products. For example, regarding the ’489 Patent, a key question will be whether the accused products create a "text map array" by analyzing "on-bits" and "phase changes" in the manner claimed, or if they use alternative, non-infringing methods for page segmentation developed since the patent's 1988 priority date.
 
V. Key Claim Terms for Construction
Because the complaint fails to assert specific claims, this analysis is based on the first independent claim of the lead patents.
- For the ’053 Patent: - The Term: "parsing the page into a plurality of blocks"
- Context and Importance: This term appears to be the central inventive concept of Claim 1. Its construction will be critical to determining infringement, as it defines the core action of automatically analyzing a page's layout. Practitioners may focus on this term because the defendants may argue their modern layout analysis techniques do not constitute "parsing...into a plurality of blocks" as described in the 1988-priority-date specification.
- Intrinsic Evidence for Interpretation:- Evidence for a Broader Interpretation: The specification describes the goal as constructing "a reasonable interpretation of the page layout" and outputting an "ordered set of block descriptors that trace the normal reading sequence," which could suggest the term covers any automated process that logically segments a page for OCR (’053 Patent, col. 5:20-25).
- Evidence for a Narrower Interpretation: The specification describes a specific method for achieving this parsing, based on analyzing "texture" by measuring the "local density of black-to-white transitions" to characterize areas as graphics, text, or rulings, and then finding "paths of white space" to separate them (’053 Patent, col. 5:5-15). A defendant may argue the claim should be limited to this specific statistical approach.
 
 
- For the ’489 Patent: - The Term: "determining whether the number of phase changes from on-bits to off-bits...is greater than or equal to a fourth predetermined number and less than or equal to a fifth predetermined number"
- Context and Importance: This term defines the specific mathematical rule for identifying text based on its "texture." The viability of the infringement claim will depend on whether this specific rule, with its predetermined numerical ranges, reads on the algorithms used in the accused products.
- Intrinsic Evidence for Interpretation:- Evidence for a Broader Interpretation: The general concept is using the frequency of black-to-white pixel transitions to identify text. Plaintiff may argue that any algorithm that relies on this principle infringes, regardless of the precise numerical thresholds used.
- Evidence for a Narrower Interpretation: The claim language explicitly requires comparing the number of "phase changes" to fourth and fifth "predetermined number[s]." The specification provides exemplary ranges, stating a phase change count between 8 and 22 indicates text (’489 Patent, col. 7:15-18). Defendants may argue this limitation confines the claim to systems that use these or similar specific, hard-coded numerical thresholds, potentially distinguishing it from modern, more dynamic or machine-learning-based approaches.
 
 
VI. Other Allegations
- Indirect Infringement: The complaint alleges both induced and contributory infringement for all patents. The allegations are made "upon information and belief" and do not plead specific facts, such as identifying user manuals or instructions that would encourage infringing use (e.g., Compl. ¶50).
- Willful Infringement: Willfulness is alleged for all asserted patents against all defendants. The complaint asserts these allegations "upon information and belief" without providing a factual basis for pre-suit knowledge of the patents, such as prior correspondence or litigation (e.g., Compl. ¶52).
VII. Analyst’s Conclusion: Key Questions for the Case
- A primary procedural issue will be one of pleading sufficiency: does the complaint's assertion of infringement of "one or more claims" of eight patents, without identifying a single asserted claim or providing any detail on how any accused product meets any claim limitation, state a plausible claim for relief or will it be subject to dismissal?
- A core issue of claim scope will be whether the methods described and claimed in the early-priority patents, which are based on specific statistical and rule-based heuristics from the 1980s (e.g., counting "phase changes"), can be construed to cover the potentially more advanced and dynamic OCR and layout analysis algorithms used in modern software.
- A central evidentiary question will be one of technical proof: assuming the case proceeds, Plaintiff will bear the burden of demonstrating, through technical evidence and expert testimony, that the accused products—many of which are complex, multi-component systems—actually practice the specific steps recited in the patent claims, moving far beyond the generalized allegations in the complaint.