Fully Featured, Fast and Accurate OCR

Scan a printed or typed document, indicate the zones of nterest -- or have the system detect the text and graphic blocks for you -- and execute the character recognition. A few mouse clicks beat long hours of work as READIRIS converts your paper documents into editable computer files: 10 to 25 times faster than manual retyping!

READIRIS also recognizes tables of figures and recreates the table structure by assigning your figures to separate cells; your numeric data are immediately ready for further processing with your spreadsheet.

Based on the proven Connectionist technology from I.R.I.S., READIRIS represents the best OCR has to offer. READIRIS uses every means available to make computers read. Font-independent feature extraction is complemented by self-learning techniques derived from a neural network. The system can learn new characters through context analysis: linguistic knowledge about syllables and words for up to 26 languages improves the OCR performance.

User verification in pop-up style not only flags doubtful characters but substantially increases the system's precision. All solutions confirmed by the user are memorized, increasing speed and confidence as you go along. Using READIRIS means rendering it more intelligent each time! This powerful learning tool not only allows you to train READIRIS on special characters such as mathematic or Greek symbols but also to cope with distorted fonts as you will find in real documents.

• RECOGNITION
• Connectionist technology
• Omnifont feature extraction
• Self-learning routines derived from a proprietary neural network technique
• Contextual and linguistics analysis resolves joined characters (ligatures) and recomposes fragmented characters
• Powerful learning module
• Lindskew tolerance 7 degrees
• LINGUISTIC SUPPORT
• Support of 26 languages, including Eastern-European languages
• Cyrillic alphabet (included)
• Linguistic modules using a 100,000 word lexicon for each language
• CHARACTER SETS
• American and European character sets
• Cyrillic alphabet (option)
• Trainable on any special symbol: Greek, mathematic, etc.
• Numeric mode for tables of figures
• PRINTING QUALITY
• Typeset, typed, laserprinted and draft dot matrix printed documents
• FONTS
• Virtually any proportional and monospaced typeface
• Normal, bold, italic and underlined typestyles
• USER INTERFACES
• Button bar based operation allows "one-click" OCR
• Adjustable pop-up user verificaation
• Stand-alone operation or call from within any Windows applicataion through Connect capability
• PREVIEW
• Image preview with 4 zoom levels
• Adjustable manual windowing of relevant zones
• Automatic page analysis with discrimination of text and graphics
• Storage of zoning templates for similar documents
• OPERATION
• Page-by-page OCR and fully automatic recognition of OCR batch
• Integratable through comprehensive DDE module
• Support of multitasking, even on 386-based PC
• TEXT OUTPUT
• Windows clipboard and all leading wordprocessors and text ormats, including WordPerfect and Word
• Detection and encoding of paragraphs, hyphenation
• TABLE OUTPUT
• Windows clipboard and generic table format, including Excel and Lotus 1-2-3
• GRAPHIC OUTPUT
• Bi-Level images in (compressed) TIFF and Paintbrush (PCX) format Detection and encoding of paragraphs, hyphenation
• SCANNER SUPPORT
• Large range of flatbed, sheefed, hand-held scanners and imaging boards (Xionics and Kofax)
• Image-enhancing, noise reducingHP Accupage technology
• Compliance with Twain and SPI scanner standards
• Fax input in TIFF and Paintbrush (PCX) format

• 386 based PC, PS/2 or compatible computer
• single parallel port
• 4 MB RAM, 8 MB RAM preferred
• MS-Windows 3.1 or higher