Classifications

• • G—PHYSICS
  • G06—COMPUTING; CALCULATING OR COUNTING
  • G06F—ELECTRIC DIGITAL DATA PROCESSING
  • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
  • G06F16/90—Details of database functions independent of the retrieved data types
  • G06F16/95—Retrieval from the web
  • G06F16/951—Indexing; Web crawling techniques
• • G—PHYSICS
  • G06—COMPUTING; CALCULATING OR COUNTING
  • G06F—ELECTRIC DIGITAL DATA PROCESSING
  • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
  • G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
  • G06F16/33—Querying
  • G06F16/332—Query formulation
• • G—PHYSICS
  • G06—COMPUTING; CALCULATING OR COUNTING
  • G06F—ELECTRIC DIGITAL DATA PROCESSING
  • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
  • G06F16/90—Details of database functions independent of the retrieved data types
  • G06F16/95—Retrieval from the web
  • G06F16/955—Retrieval from the web using information identifiers, e.g. uniform resource locators [URL]
  • G06F16/9554—Retrieval from the web using information identifiers, e.g. uniform resource locators [URL] by using bar codes
• • G—PHYSICS
  • G06—COMPUTING; CALCULATING OR COUNTING
  • G06F—ELECTRIC DIGITAL DATA PROCESSING
  • G06F7/00—Methods or arrangements for processing data by operating upon the order or content of the data handled

Definitions

• 11/098,0308 entitled CONTENT ACCESS WITH HANDHELD DOCUMENT DATA CAPTURE DEVICES
• U.S. Patent Application No. 11/098,014 entitled SEARCH ENGINES AND SYSTEMS WITH HANDHELD DOCUMENT DATA CAPTURE DEVICES
• U.S. Patent Application No. 11/097,103 entitled TRIGGERING ACTIONS IN RESPONSE TO OPTICALLY OR ACOUSTICALLY CAPTURING KEYWORDS FROM A RENDERED DOCUMENT
• U.S. Patent Application No. 11/098,043 entitled SEARCHING AND ACCESSING DOCUMENTS ON PRIVATE NETWORKS FOR USE WITH CAPTURES FROM RENDERED DOCUMENTS, U.S.
• Patent Application No. 11/097,981 entitled INFORMATION GATHERING SYSTEM AND METHOD
• U.S. Patent Application No. 11/097,089 entitled DOCUMENT ENHANCEMENT SYSTEM AND METHOD
• U.S. Patent Application No. 11/097,835 entitled PUBLISHING TECHNIQUES FOR ADDING VALUE TO A RENDERED DOCUMENT
• U.S. Patent Application No. 11/098,016 entitled ARCHIVE OF TEXT CAPTURES FROM RENDERED DOCUMENTS
• Paper documents have an enduring appeal, as can be seen by the proliferation of paper documents in the computer age. It has never been easier to print and publish paper documents than it is today. Paper documents prevail even though electronic documents are easier to duplicate, transmit, search and edit.
• Figure 1 is a data flow diagram that illustrates the flow of information in one embodiment of the core system.
• Figure 2 is a component diagram of components included in a typical implementation of the system in the context of a typical operating environment.
• Figure 3 is a block diagram of an embodiment of a scanner.
• Figure 4 is a diagram showing the typical environment in which embodiments of the system operate.
• FIG. 5 is a flow diagram showing steps typically performed by the system to implement bookmarks.
• Figure 6 is a diagram showing a scanning device that has an integrated highlighter and pen.
• Figure 7 is a flow diagram showing steps typically performed by the system to process text capture actions.
• Figures 8A-8D illustrate how two characters of an alphabet often have approximately the same Relative * shape in multiple fonts.
• Figure 9 is a diagram that illustrates the approach used by the facility in some embodiments to learn an entirely new set of symbols using self recognition.
• Figure 10 is a diagram that shows a subset of the symbols shown in Figure 9 that self-define their vertical and horizontal extent.
• Figure 11 is a diagram that shows relationships between some of the symbols shown in Figure 9. DETAILED DESCRIPTION OVERVIEW
• a system for interpreting and interacting with rendered documents e.g., printed or displayed documents
• the system is not directly concerned with recognizing and interpreting characters per se; it does not concern itself with recognizing and understanding printed characters, though it may in some cases perform this function. Rather, the system assumes that a version of the document is already known, that machine- readable (e.g., ASCII or some other machine-readable text) source or reference version document is at hand, or is machine-accessible, or will be made available in the future.
• the system uses various features (including text) in rendered documents for navigation (i.e., determining location within a document). The location is in turn used to enable a rich set of user functions and interactions, some of which will be described below.
• the system is based in part on the process of interpreting and deciphering the patterns of marks (e.g., the text and any rendered supplemental informational marks) in documents to determine location information.
• this location information is with reference to the document itself - e.g., location within the document, often down to a single paragraph, sentence, word and even single character.
• the location information can be converted to location on a display screen, a printed page, etc.
• Print is used in its generic sense to documents rendering in any form that is readable to a human (e.g., on paper, on a display screen, in Braille format, etc.). It should be understood that in many cases various of the features and applications of system apply quite well to non-alphanumeric rendered content - such as punctuation, graphics and images, special marks, etc. Embodiments of the system include these additional uses. PART I - INTRODUCTION
• the system uses a sample of text captured from a paper document, for example using a handheld scanner, to identify and locate an electronic counterpart of the document.
• the amount of text needed by the facility is very small in that a few words of text from a document can often function as an identifier for the paper document and as a link to its electronic counterpart.
• the system may use those few words to identify not only the document, but also a location within the document.
• the system implements these and many other examples of "paper/digital integration" without requiring changes to the current processes of writing, printing and publishing documents, giving such conventional rendered documents a whole new layer of digital functionality.
• Terminology A typical use of the system begins with using an optical scanner to scan text from a paper document, but it is important to note that other methods of capture from other types of document are equally applicable. The system is therefore sometimes described as scanning or capturing text from a rendered document, where those terms are defined as follows:
• a rendered document is a printed document or a document shown on a display or monitor. It is a document that is perceptible to a human, whether in permanent form or on a transitory display.
• Scanning or capturing is the process of systematic examination to obtain information from a rendered document.
• the process may involve optical capture using a scanner or camera (for example a camera in a cellphone), or it may involve reading aloud from the document into an audio capture device or typing it on a keypad or keyboard.
• a scanner or camera for example a camera in a cellphone
• This section describes some of the devices, processes and systems that constitute a system for paper/digital integration.
• the system builds a wide variety of services and applications on this underlying core that provides the basic functionality.
• Figure 1 is a data flow diagram that illustrates the flow of information in one embodiment of the core system. Other embodiments may not use all of the stages or elements illustrated here, while some will use many more.
• Text from a rendered document is captured 100, typically in optical form by an optical scanner or audio form by a voice recorder, and this image or sound data is then processed 102, for example to remove artifacts of the capture process or to improve the signal-to-noise ratio.
• a recognition process 104 such as OCR, speech recognition, or autocorrelation then converts the data into a signature, comprised in some embodiments of text, text offsets, or other symbols.
• the system performs an alternate form of extracting document signature from the rendered document.
• the signature represents a set of possible text transcriptions in some embodiments. This process may be influenced by feedback from other stages, for example, if the search process and context analysis 110 have identified some candidate documents from which the capture may originate, thus narrowing the possible interpretations of the original capture.
• a post-processing 106 stage may take the output of the recognition process and filter it or perform such other operations upon it as may be useful. Depending upon the embodiment implemented, it may be possible at this stage to deduce some direct actions 107 to be taken immediately without reference to the later stages, such as where a phrase or symbol has been captured which contains sufficient information in itself to convey the user's intent. In these cases no digital counterpart document need be referenced, or even known to the system.
• next stage will be to construct a query 108 or a set of queries for use in searching.
• Some aspects of the query construction may depend on the search process used and so cannot be performed until the next stage, but there will typically be some operations, such as the removal of obviously misrecognized or irrelevant characters, which can be performed in advance.
• the query or queries are then passed to the search and context analysis stage 110.
• the system optionally attempts to identify the document from which the original data was captured.
• the system typically uses search indices and search engines 112, knowledge about the user 114 and knowledge about the user's context or the context in which the capture occurred 116.
• Search engine 112 may employ and/or index information specifically about rendered documents, about their digital counterpart documents, and about documents that have a web (internet) presence). It may write to, as well as read from, many of these sources and, as has been mentioned, it may feed information into other stages of the process, for example by giving the recognition system 104 information about the language, font, rendering and likely next words based on its knowledge of the candidate documents.
• the next stage will be to retrieve 120 a copy of the document or documents that have been identified.
• the sources of the documents 124 may be directly accessible, for example from a local filing system or database or a web server, or they may need to be contacted via some access service 122 which might enforce authentication, security or payment or may provide other services such as conversion of the document into a desired format.
• marksup may be associated with, and apply to, a rendered document and/or the digital counterpart to a rendered document, or to groups of either or both of these documents.
• actions may be taken 140. These may be default actions such as simply recording the information found, they may be dependent on the data or document, or they may be derived from the markup analysis. Sometimes the action will simply be to pass the data to another system. In some cases the various possible actions appropriate to a capture at a specific point in a rendered document will be presented to the user as a menu on an associated display, for example on a local display 332, on a computer display 212 or a mobile phone or PDA display 216. If the user doesn't respond to the menu, the default actions can be taken.
• Figure 2 is a component diagram of components included in a typical implementation of the system in the context of a typical operating environment.
• the operating environment includes one or more optical scanning capture devices 202 or voice capture devices 204.
• Each capture device is able to communicate with other parts of the system such as a computer 212 and a mobile station 216 (e.g., a mobile phone or PDA) using either a direct wired or wireless connection, or through the network 220, with which it can communicate using a wired or wireless connection, the latter typically involving a wireless base station 214.
• the capture device is integrated in the mobile station, and optionally shares some of the audio and/or optical components used in the device for voice communications and picture-taking.
• Computer 212 may include a memory containing computer executable instructions for processing an order from scanning devices 202 and 204.
• an order can include an identifier (such as a serial number of the scanning device 202/204 or an identifier that partially or uniquely identifies the user of the scanner), scanning context information (e.g., time of scan, location of scan, etc.) and/or scanned information (such as a text string) that is used to uniquely identify the document being scanned.
• scanning context information e.g., time of scan, location of scan, etc.
• scanned information such as a text string
• the operating environment may include more or less components.
• the network 220 may be a corporate intranet, the public Internet, a mobile phone network or some other network, or any interconnection of the above.
• the devices may all be operable in accordance with well-known commercial transaction and communication protocols (e.g., Internet Protocol (IP)).
• IP Internet Protocol
• the functions and capabilities of scanning device 202, computer 212, and mobile station 216 may be wholly or partially integrated into one device.
• scanning device, computer, and mobile station can refer to the same device depending upon whether the device incorporates functions or capabilities of the scanning device 202, computer 212 and mobile station 216.
• some or all of the functions of the search engines 232, document sources 234, user account services 236, markup services 238 and other network services 239 may be implemented on any of the devices and/or other devices not shown.
• the capture device may capture text using an optical scanner that captures image data from the rendered document, or using an audio recording device that captures a user's spoken reading of the text, or other methods. Some embodiments of the capture device may also capture images, graphical symbols and icons, etc., including machine readable codes such as barcodes.
• the device may be exceedingly simple, consisting of little more than the transducer, some storage, and a data interface, relying on other functionality residing elsewhere in the system, or it may be a more full-featured device. For illustration, this section describes a device based around an optical scanner and with a reasonable number of features.
• Scanners are well known devices that capture and digitize images. An offshoot of the photocopier industry, the first scanners were relatively large devices that captured an entire document page at once. Recently, portable optical scanners have been introduced in convenient form factors, such as a pen-shaped handheld device.
• the portable scanner is used to scan text, graphics, or symbols from rendered documents.
• the portable scanner has a scanning element that captures text, symbols, graphics, etc, from rendered documents.
• rendered documents include documents that have been displayed on a screen such as a CRT monitor or LCD display.
• FIG. 3 is a block diagram of an embodiment of a scanner 302.
• the scanner 302 comprises an optical scanning head 308 to scan information from rendered documents and convert it to machine-compatible data, and an optical path 306, typically a lens, an aperture or an image conduit to convey the image from the rendered document to the scanning head.
• the scanning head 308 may incorporate a Charge-Coupled Device (CCD), a Complementary Metal Oxide Semiconductor (CMOS) imaging device, or an optical sensor of another type.
• CCD Charge-Coupled Device
• CMOS Complementary Metal Oxide Semiconductor
• a microphone 310 and associated circuitry convert the sound of the environment (including spoken words) into machine-compatible signals, and other input facilities exist in the form of buttons, scroll-wheels or other tactile sensors such as touch-pads 314.
• Feedback to the user is possible through a visual display or indicator lights 332, through a loudspeaker or other audio transducer 334 and through a vibrate module 336.
• the scanner 302 comprises logic 326 to interact with the various other components, possibly processing the received signals into different formats and/or interpretations.
• Logic 326 may be operable to read and write data and program instructions stored in associated storage 330 such as RAM, ROM, flash, or other suitable memory. It may read a time signal from the clock unit 328.
• the scanner 302 also includes an interface 316 to communicate scanned information and other signals to a network and/or an associated computing device.
• the scanner 302 may have an on-board power supply 332.
• the scanner 302 may be powered from a tethered connection to another device, such as a Universal Serial Bus (USB) connection.
• USB Universal Serial Bus
• a reader may scan some text from a newspaper article with scanner 302. The text is scanned as a bit-mapped image via the scanning head 308.
• Logic 326 causes the bit-mapped image to be stored in memory 330 with an associated time-stamp read from the clock unit 328.
• Logic 326 may also perform optical character recognition (OCR) or other post-scan processing on the bit-mapped image to convert it to text.
• OCR optical character recognition
• Logic 326 may optionally extract a signature from the image, for example by performing a convolution-like process to locate repeating occurrences of characters, symbols or objects, and determine the distance or number of other characters, symbols, or objects between these repeated elements.
• the reader may then upload the bit-mapped image (or text or other signature, if post-scan processing has been performed by logic 326) to an associated computer via interface 316.
• a reader may capture some text from an article as an audio file by using microphone 310 as an acoustic capture port.
• Logic 326 causes audio file to be stored in memory 328.
• Logic 326 may also perform voice recognition or other post-scan processing on the audio file to convert it to text.
• the reader may then upload the audio file (or text produced by post-scan processing performed by logic 326) to an associated computer via interface 316.
• End-to-end feedback can be applied by performing an approximation of the recognition or interpretation, identifying a set of one or more candidate matching documents, and then using information from the possible matches in the candidate documents to further refine or restrict the recognition or interpretation.
• Candidate documents can be weighted according to their probable relevance (for example, based on then number of other users who have scanned in these documents, or their popularity on the Internet), and these weights can be applied in this iterative recognition process.
• the system may also employ other methods of discovering the document and location, such as by using watermarks or other special markings on the rendered document.
• the identity of the user may be determined from a unique identifier associated with a capturing device, and/or biometric or other supplemental information (speech patterns, fingerprints, etc.).
• the search query can be constructed taking into account the types of errors likely to occur in the particular capture method used.
• One example of this is an indication of suspected errors in the recognition of specific characters; in this instance a search engine may treat these characters as wildcards, or assign them a lower priority.
• the capturing device may not be in communication with the search engine or corpus at the time of the data capture. For this reason, information helpful to the offline use of the device may be downloaded to the device in advance, or to some entity with which the device can communicate. In some cases, all or a substantial part of an index associated with a corpus may be downloaded. This topic is discussed further in Section 15.3.
• the queries may be saved and transmitted to the rest of the system at such a time as communication is restored.
• multiple queries may be launched in response to a single capture, either in sequence or in parallel.
• queries may be sent in response to a single capture, for example as new words are added to the capture, or to query multiple search engines in parallel.
• the system sends queries to a special index for the current document, to a search engine on a local machine, to a search engine on the corporate network, and to remote search engines on the Internet.
• search engine that handles traditional online queries also to handle those originating from rendered documents.
• Conventional search engines may be enhanced or modified in a number of ways to make them more suitable for use with the described system.
• the search engine and/or other components of the system may create and maintain indices that have different or extra features.
• the system may modify an incoming paper-originated query or change the way the query is handled in the resulting search, thus distinguishing these paper-originated queries from those coming from queries typed into web browsers and other sources.
• the system may take different actions or offer different options when the results are returned by the searches originated from paper as compared to those from other sources.
• indexing Often, the same index can be searched using either paper-originated or traditional queries, but the index may be enhanced for use in the current system in a variety of ways.
• Extra fields can be added to such an index that will help in the case of a paper-based search.
• the first example is a field indicating that the document is known to exist or be distributed in paper form.
• the system may give such documents higher priority if the query comes from paper.
• Another important example may be recording information about the layout of a specific rendering of a document.
• the index may include information about where the line breaks and page breaks occur, which fonts were used, any unusual capitalization.
• the index may also include information about the proximity of other items on the page, such as images, text boxes, tables and advertisements.
• semantic information that can be deduced from the source markup but is not apparent in the paper document, such as the fact that a particular piece of text refers to an item offered for sale, or that a certain paragraph contains program code, may also be recorded in the index.
• indexing in the knowledge of the capture method is the knowledge of the type of capture likely to be used.
• a search initiated by an optical scan may benefit if the index takes into account characters that are easily confused in the OCR process, or includes some knowledge of the fonts used in the document.
• the query is from speech recognition, an index based on similar-sounding phonemes may be much more efficiently searched.
• An additional factor that may affect the use of the index in the described model is the importance of iterative feedback during the recognition process. If the search engine is able to provide feedback from the index as the text is being captured, it can greatly increase the accuracy of the capture.
• the system stores the appropriate offset or signature information in an index. 4.1.3. Multiple indices [0080] Lastly, in the described system, it may be common to conduct searches on many indices. Indices may be maintained on several machines on a corporate network. Partial indices may be downloaded to the capture device, or to a machine close to the capture device. Separate indices may be created for users or groups of users with particular interests, habits or permissions. An index may exist for each filesystem, each directory, even each file on a user's hard disk. Indexes are published and subscribed to by users and by systems. It will be important, then, to construct indices that can be distributed, updated, merged and separated efficiently.
• a search engine may take different actions when it recognizes that a search query originated from a paper document.
• the engine might handle the query in a way that is more tolerant to the types of errors likely to appear in certain capture methods, for example.
• queries from a capture device can reach the engine by a different channel or port or type of connection than those from other sources, and can be distinguished in that way.
• some embodiments of the system will route queries to the search engine by way of a dedicated gateway.
• the search engine knows that all queries passing through the dedicated gateway were originated from a paper document.
• Section 13 describes a variety of different factors which are external to the captured text itself, yet which can be a significant aid in identifying a document. These include such things as the history of recent scans, the longer-term reading habits of a particular user, the geographic location of a user and the user's recent use of particular electronic documents. Such factors are referred to herein as "context.”
• search engine may keep track of a user's scanning history, and may also cross-reference this scanning history to conventional keyboard-based queries. In such cases, the search engine maintains and uses more state information about each individual user than do most conventional search engines, and each interaction with a search engine may be considered to extend over several searches and a longer period of time than is typical today.
• Some of the context may be transmitted to the search engine in the search query (Section 3.3), and may possibly be stored at the engine so as to play a part in future queries. Lastly, some of the context will best be handled elsewhere, and so becomes a filter or secondary search applied to the results from the search engine.
• the described system can emit and use not only information about documents as a whole, but also information about sub-regions of documents, even down to individual words. Many existing search engines concentrate simply on locating a document or file that is relevant to a particular query. Those that can work on a finer grain and identify a location within a document will provide a significant benefit for the described system. 4.3. Returning the results [0089] The search engine may use some of the further information it now maintains to affect the results returned.
• the system may also return certain documents to which the user has access only as a result of being in possession of the paper copy (Section 7.4).
• the search engine may also offer new actions or options appropriate to the described system, beyond simple retrieval of the text.
• the described system In addition to performing the capture-search-retrieve process, the described system also associates extra functionality with a document, and in particular with specific locations or segments of text within a document. This extra functionality is often, though not exclusively, associated with the rendered document by being associated with its electronic counterpart. As an example, hyperlinks in a web page could have the same functionality when a printout of that web page is scanned. In some cases, the functionality is not defined in the electronic document, but is stored or generated elsewhere.
• markup This layer of added functionality is referred to herein as "markup.”
• the markup may include human-readable content, but is often invisible to a user and/or intended for machine use. Examples include options to be displayed in a popup-menu on a nearby display when a user captures text from a particular area in a rendered document, or audio samples that illustrate the pronunciation of a particular phrase.
• Markup data may be created or supplied by the author of the document, or by the user, or by some other party.
• Markup data may be attached to the electronic document or embedded in it. It may be found in a conventional location (for example, in the same place as the document but with a different filename suffix). Markup data may be included in the search results of the query that located the original document, or may be found by a separate query to the same or another search engine. Markup data may be found using the original captured text and other capture information or contextual information, or it may be found using already-deduced information about the document and location of the capture. Markup data may be found in a location specified in the document, even if the markup itself is not included in the document.
• the markup may be largely static and specific to the document, similar to the way links on a traditional html web page are often embedded as static data within the html document, but markup may also be dynamically generated and/or applied to a large number of documents.
• An example of dynamic markup is information attached to a document that includes the up-to-date share price of companies mentioned in that document.
• An example of broadly applied markup is translation information that is automatically available on multiple documents or sections of documents in a particular language.
• the system may add certain items to a pop-up menu (e.g., a link to an online bookstore) whenever the user captures the word "book," or the title of a book, or a topic related to books.
• a pop-up menu e.g., a link to an online bookstore
• digital counterpart documents or indices are consulted to determine whether a capture occurred near the word "book," or the title of a book, or a topic related to books - and the system behavior is modified in accordance with this proximity to keyword elements.
• markup enables data captured from non-commercial text or documents to trigger a commercial transaction.
• Annotations are another type of electronic information that may be associated with a document. For example, a user can attach an audio file of his/her thoughts about a particular document for later retrieval as voice annotations. As another example of a multimedia annotation, a user may attach photographs of places referred to in the document. The user generally supplies annotations for the document but the system can associate annotations from other sources (for example, other users in a work group may share annotations).
• markup data may often be supplied by third parties, such as by other readers of the document. Online discussions and reviews are a good example, as are community-managed information relating to particular works, volunteer-contributed translations and explanations.
• markup By analyzing the data captured from documents by several or all users of the system, markup can be generated based on the activities and interests of a community.
• An example might be an online bookstore that creates markup or annotations that tell the user, in effect, "People who enjoyed this book also enjoyed " The markup may be less anonymous, and may tell the user which of the people in his/her contact list have also read this document recently.
• Other examples of datastream analysis are included in Section 14.
• Markup based on external events and data sources [00106] Markup will often be based on external events and data sources, such as input from a corporate database, information from the public Internet, or statistics gathered by the local operating system.
• Data sources may also be more local, and in particular may provide information about the user's context - his/her identity, location and activities.
• the system might communicate with the user's mobile phone and offer a markup layer that gives the user the option to send a document to somebody that the user has recently spoken to on the phone.
• the identity of the user will be known. Sometimes this will be an “anonymous identity,” where the user is identified only by the serial number of the capture device, for example. Typically, however, it is expected that the system will have a much more detailed knowledge of the user, which can be used for personalizing the system and to allow activities and transactions to be performed in the user's name.
• the user can be presented with a "Life Library," a record of everything s/he has read and captured. This may be simply for personal interest, but may be used, for example, in a library by an academic who is gathering material for the bibliography of his next paper.
• the user may wish to make the library public, such as by publishing it on the web in a similar manner to a weblog, so that others may see what s/he is reading and finds of interest.
• the capture can be stored in the library and can be processed later, either automatically or in response to a user request.
• a user can also subscribe to new markup services and apply them to previously captured scans.
• a record of a user's past captures is also useful for the system. Many aspects of the system operation can be enhanced by knowing the user's reading habits and history. The simplest example is that any scan made by a user is more likely to come from a document that the user has scanned in the recent past, and in particular if the previous scan was within the last few minutes it is very likely to be from the same document. Similarly, it is more likely that a document is being read in start-to-finish order. Thus, for English documents, it is also more likely that later scans will occur farther down in the document. Such factors can help the system establish the location of the capture in cases of ambiguity, and can also reduce the amount of text that needs to be captured.
• Scanner as payment, identity and authentication device
• this device may be used as a key that identifies the user and authorizes certain actions.
• the device may be embedded in a mobile phone or in some other way associated with a mobile phone account.
• a scanner may be associated with a mobile phone account by inserting a SIM card associated with the account into the scanner.
• the device may be embedded in a credit card or other payment card, or have the facility for such a card to be connected to it. The device may therefore be used as a payment token, and financial transactions may be initiated by the capture from the rendered document.
• the scanner may also be associated with a particular user or account through the process of scanning some token, symbol or text associated with that user or account.
• scanner may be used for biometric identification, for example by scanning the fingerprint of the user.
• the system may identify the user by matching the voice pattern of the user or by requiring the user to speak a certain password or phrase.
• the capture device When the capture device is used to identify and authenticate the user, and to initiate transactions on behalf of the user, it is important that communications between the device and other parts of the system are secure. It is also important to guard against such situations as another device impersonating a scanner, and so- called "man in the middle" attacks where communications between the device and other components are intercepted.
• An advantage of the described system is that there is no need to alter the traditional processes of creating, printing or publishing documents in order to gain many of the system's benefits. There are reasons, though, that the creators or publishers of a document - hereafter simply referred to as the "publishers" - may wish to create functionality to support the described system.
• the system allows for printed documents to have an associated electronic presence.
• Conventionally publishers often ship a CD-ROM with a book that contains further digital information, tutorial movies and other multimedia data, sample code or documents, or further reference materials.
• some publishers maintain web sites associated with particular publications which provide such materials, as well as information which may be updated after the time of publishing, such as errata, further comments, updated reference materials, bibliographies and further sources of relevant data, and translations into other languages. Online forums allow readers to contribute their comments about the publication.
• the described system allows such materials to be much more closely tied to the rendered document than ever before, and allows the discovery of and interaction with them to be much easier for the user.
• the system can automatically connect the user to digital materials associated with the document, and more particularly associated with that specific part of the document.
• the user can be connected to online communities that discuss that section of the text, or to annotations and commentaries by other readers. In the past, such information would typically need to be found by searching for a particular page number or chapter.
• Some publishers may have mailing lists to which readers can subscribe if they wish to be notified of new relevant matter or when a new edition of the book is published.
• the user can register an interest in particular documents or parts of documents more easily, in some cases even before the publisher has considered providing any such functionality.
• the reader's interest can be fed to the publisher, possibly affecting their decision about when and where to provide updates, further information, new editions or even completely new publications on topics that have proved to be of interest in existing books.
• Printed marks with special meaning or containing special data [00128] Many aspects of the system are enabled simply through the use of the text already existing in a document. If the document is produced in the knowledge that it may be used in conjunction with the system, however, extra functionality can be added by printing extra information in the form of special marks, which may be used to identify the text or a required action more closely, or otherwise enhance the document's interaction with the system. The simplest and most important example is an indication to the reader that the document is definitely accessible through the system. A special icon might be used, for example, to indicate that this document has an online discussion forum associated with it.
• Such symbols may be intended purely for the reader, or they may be recognized by the system when scanned and used to initiate some action. Sufficient data may be encoded in the symbol to identify more than just the symbol: it may also store information, for example about the document, edition, and location of the symbol, which could be recognized and read by the system.
• the printed document is a gateway to extra materials and functionality, access to such features can also be time-limited. After the expiry date, a user may be required to pay a fee or obtain a newer version of the document to access the features again.
• the paper document will, of course, still be usable, but will lose some of its enhanced electronic functionality. This may be desirable, for example, because there is profit for the publisher in receiving fees for access to electronic materials, or in requiring the user to purchase new editions from time to time, or because there are disadvantages associated with outdated versions of the printed document remaining in circulation. Coupons are an example of a type of commercial document that can have an expiration date.
• Section 10.5 discusses the use of the system's statistics to influence compensation of authors and pricing of advertisements.
• the system deduces the popularity of a publication from the activity in the electronic community associated with it as well as from the use of the paper document. These factors may help publishers to make decisions about what they will publish in future. If a chapter in an existing book, for example, turns out to be exceedingly popular, it may be worth expanding into a separate publication.
• An important aspect of the described system is the ability to provide to a user who has access to a rendered copy of a document access to an electronic version of that document.
• a document is freely available on a public network or a private network to which the user has access.
• the system uses the captured text to identify, locate and retrieve the document, in some cases displaying it on the user's screen or depositing it in their email inbox.
• a document will be available in electronic form, but for a variety of reasons may not be accessible to the user. There may not be sufficient connectivity to retrieve the document, the user may not be entitled to retrieve it, there may be a cost associated with gaining access to it, or the document may have been withdrawn and possibly replaced by a new version, to name just a few possibilities.
• the system typically provides feedback to the user about these situations.
• the degree or nature of the access granted to a particular user may be different if it is known that the user already has access to a printed copy of the document.
• Access to the document may be restricted to specific users, or to those meeting particular criteria, or may only be available in certain circumstances, for example when the user is connected to a secure network.
• Section 6 describes some of the ways in which the credentials of a user and scanner may be established.
• Documents that are not freely available to the general public may still be accessible on payment of a fee, often as compensation to the publisher or copyright- holder.
• the system may implement payment facilities directly or may make use of other payment methods associated with the user, including those described in Section 6.2.
• Electronic documents are often transient; the digital source version of a rendered document may be available now but inaccessible in future.
• the system may retrieve and store the existing version on behalf of the user, even if the user has not requested it, thus guaranteeing its availability should the user request it in future. This also makes it available for the system's use, for example for searching as part of the process of identifying future captures.
• a trusted "document escrow" service can retrieve the document on behalf of the user, such as upon payment of a modest fee, with the assurance that the copyright holder will be fully compensated in future if the user should ever request the document from the service.
• Variations on this theme can be implemented if the document is not available in electronic form at the time of capture.
• the user can authorize the service to submit a request for or make a payment for the document on his/her behalf if the electronic document should become available at a later date.
• payment may be waived, reduced or satisfied based on the user's existing association with another account or subscription. Subscribers to the printed version of a newspaper might automatically be entitled to retrieve the electronic version, for example.
• association may not be quite so direct: a user may be granted access based on an account established by their employer, or based on their scanning of a printed copy owned by a friend who is a subscriber.
• OCR Optical Character Recognition
• a scanning device for use with the described system will often be small, portable, and low power.
• the scanning device may capture only a few words at a time, and in some implementations does not even capture a whole character at once, but rather a horizontal slice through the text, many such slices being stitched together to form a recognizable signal from which the text may be deduced.
• the scanning device may also have very limited processing power or storage so, while in some embodiments it may perform all of the OCR process itself, many embodiments will depend on a connection to a more powerful device, possibly at a later time, to convert the captured signals into text. Lastly, it may have very limited facilities for user interaction, so may need to defer any requests for user input until later, or operate in a "best-guess" mode to a greater degree than is common now.
• OCR optical character recognition
• the OCR process can be informed by the contents of the document corpus as it progresses, potentially offering substantially greater recognition accuracy.
• Such a connection will also allow the device to inform the user when sufficient text has been captured to identify the digital source.
• the font or a rendering of it, may be downloaded to the device to help with the recognition.
• While component characters of a text fragment may be the most recognized way to represent a fragment of text that may be used as a document signature, other representations of the text may work sufficiently well that the actual text of a text fragment need not be used when attempting to locate the text fragment in a digital document and/or database, or when disambiguating the representation of a text fragment into a readable form.
• Other representations of text fragments may provide benefits that actual text representations lack. For example, optical character recognition of text fragments is often prone to errors, unlike other representations of captured text fragments that may be used to search for and/or recreate a text fragment without resorting to optical character recognition for the entire fragment. Such methods may be more appropriate for some devices used with the current system.
• characterizations of text fragments may include, but are not limited to, word lengths, relative word lengths, character heights, character widths, character shapes, character frequencies, token frequencies, and the like.
• the offsets between matching text tokens i.e., the number of intervening tokens plus one are used to characterize fragments of text.
• each token i.e., the particular letter, digit or symbol, or group of these, that corresponds to the token image
• the offset number is the distance (number of tokens) to the next occurrence of the same token. If the token is unique within the text string, the offset is zero (0).
• the sequence of token offsets thus generated is a signature that can be used to identify the scanned text.
• the token offsets determined for a string of scanned tokens are compared to an index that indexes a corpus of electronic documents based upon the token offsets of their contents (Section 4.1.2).
• the token offsets determined for a string of scanned tokens are converted to text, and compared to a more conventional index that indexes a corpus of electronic documents based upon their contents
• Font/character "self-recognition" Conventional template-matching OCR compares scanned images to a library of character images. In essence, the alphabet is stored for each font and newly scanned images are compared to the stored images to find matching characters. The process generally has an initial delay until the correct font has been identified. After that, the OCR process is relatively quick because most documents use the same font throughout. Subsequent images can therefore be converted to text by comparison with the most recently identified font library.
• the user may be offered that document for purchase either in paper or electronic form.
• the user may also be offered related documents, such as those quoted or otherwise referred to in the paper document, or those on a similar subject, or those by the same author.
• the capture of text may be linked to other commercial activities in a variety of ways.
• the captured text may be in a catalog that is explicitly designed to sell items, in which case the text will be associated fairly directly with the purchase of an item (Section 18.2).
• the text may also be part of an advertisement, in which case a sale of the item being advertised may ensue.
• the user captures other text from which their potential interest in a commercial transaction may be deduced.
• a reader of a novel set in a particular country might be interested in a holiday there. Someone reading a review of a new car might be considering purchasing it.
• the user may capture a particular fragment of text knowing that some commercial opportunity will be presented to them as a result, or it may be a side-effect of their capture activities.
• the system allows for a new kind of advertising which is not necessarily explicitly in the rendered document, but is nonetheless based on what people are reading.
• advertisements In a traditional paper publication, advertisements generally consume a large amount of space relative to the text of a newspaper article, and a limited number of them can be placed around a particular article.
• advertising can be associated with individual words or phrases, and can selected according to the particular interest the user has shown by capturing that text and possibly taking into account their history of past scans.
• the system may gather a large amount of information about other aspects of a user's context for its own use (Section 13); estimates of the geographical location of the user are a good example. Such data can also be used to tailor the advertising presented to a user of the system. 10.5. Models of compensation
• the system enables some new models of compensation for advertisers and marketers.
• the publisher of a printed document containing advertisements may receive some income from a purchase that originated from their document. This may be true whether or not the advertisement existed in the original printed form; it may have been added electronically either by the publisher, the advertiser or some third party, and the sources of such advertising may have been subscribed to by the user.
• the "Life Library" or scan history described in Sections 6.1 and 16.1 can be an extremely valuable source of information about the interests and habits of a user. Subject to the appropriate consent and privacy issues, such data can inform offers of goods or services to the user. Even in an anonymous form, the statistics gathered can be exceedingly useful. 10.7. Sale/information at later date .when available) [00185] Advertising and other opportunities for commercial transactions may not be presented to the user immediately at the time of text capture. For example, the opportunity to purchase a sequel to a novel may not be available at the time the user is reading the novel, but the system may present them with that opportunity when the sequel is published.
• a user may capture data that relates to a purchase or other commercial transaction, but may choose not to initiate and/or complete the transaction at the time the capture is made.
• data related to captures is stored in a user's Life Library, and these Life Library entries can remain "active" (i.e., capable of subsequent interactions similar to those available at the time the capture was made).
• a user may review a capture at some later time, and optionally complete a transaction based on that capture. Because the system can keep track of when and where the original capture occurred, all parties involved in the transaction can be properly compensated.
• the author who wrote the story - and the publisher who published the story - that appeared next to the advertisement from which the user captured data can be compensated when, six months later, the user visits their Life Library, selects that particular capture from the history, and chooses "Purchase this item at Amazon" from the pop-up menu (which can be similar or identical to the menu optionally presented at the time of the capture).
• OSs Modern Operating Systems
• other software packages have many characteristics that can be advantageously exploited for use with the described system, and may also be modified in various ways to provide an even better platform for its use.
• New and upcoming file systems and their associated databases often have the ability to store a variety of metadata associated with each file.
• this metadata has included such things as the ID of the user who created the file, the dates of creation, last modification, and last use.
• Newer file systems allow such extra information as keywords, image characteristics, document sources and user comments to be stored, and in some systems this metadata can be arbitrarily extended. File systems can therefore be used to store information that would be useful in implementing the current system. For example, the date when a given document was last printed can be stored by the file system, as can details about which text from it has been captured from paper using the described system, and when and by whom.
• OS Support for capture devices [00191] As the use of capture devices such as pen scanners becomes increasingly common, it will become desirable to build support for them into the operating system, in much the same way as support is provided for mice and printers, since the applicability of capture devices extends beyond a single software application. The same will be true for other aspects of the system's operation. Some examples are discussed below. In some embodiments, the entire described system, or the core of it, is provided by the OS. In some embodiments, support for the system is provided by Application Programming Interfaces (APIs) that can be used by other software packages, including those directly implementing aspects of the system.
• APIs Application Programming Interfaces
• recognition components As recognition components become part of the OS, they can take better advantage of other facilities provided by the OS. Many systems include spelling dictionaries, grammar analysis tools, internationalization and localization facilities, for example, all of which can be advantageously employed by the described system for its recognition process, especially since they may have been customized for the particular user to include words and phrases that he/she would commonly encounter.
• an optical scan or other capture occurs and is presented to the OS, it may have a default action to be taken under those circumstances in the event that no other subsystem claims ownership of the capture.
• An example of a default action is presenting the user with a choice of alternatives, or submitting the captured text to the OS's built-in search facilities.
• OS has default action for particular documents or document types
• the OS may have a standard action that it will take when that particular document, or a document of that class, is scanned.
• Applications and other subsystems may register with the OS as potential handlers of particular types of capture, in a similar manner to the announcement by applications of their ability to handle certain file types.
• Markup data associated with a rendered document, or with a capture from a document can include instructions to the operating system to launch specific applications, pass applications arguments, parameters, or data, etc.
• gestures are discussed, particularly in the case of optical scanning, where particular movements made with a handheld scanner might represent standard actions such as marking the start and end of a region of text.
• a typical use of the system may be for the user to scan an area of a paper document, and for the system to open the electronic counterpart in a software package that is able to display or edit it, and cause that package to scroll to and highlight the scanned text (Section 12.2.1).
• the first part of this process, finding and opening the electronic document is typically provided by the OS and is standard across software packages.
• the second part, however - locating a particular piece of text within a document and causing the package to scroll to it and highlight it - is not yet standardized and is often implemented differently by each package.
• the availability of a standard API for this functionality could greatly enhance the operation of this aspect of the system.
• the system may wish to perform a variety of operations upon that text. As an example, the system may request the surrounding text, so that the user's capture of a few words could result in the system accessing the entire sentence or paragraph containing them. Again, this functionality can be usefully provided by the OS rather than being implemented in every piece of software that handles text.
• the system uses the application pop-up menus traditionally associated with clicking the right mouse button on some text. The system inserts extra options into such menus, and causes them to be displayed as a result of activities such as scanning a paper document.
• the OS keeps a simple record of when any document was printed and by whom.
• the OS takes one or more further actions that would make it better suited for use with the system. Examples include:
• An OS often maintains certain categories of folders or files that have particular significance.
• a user's documents may, by convention or design, be found in a "My Documents" folder, for example.
• Standard file-opening dialogs may automatically include a list of recently opened documents.
• Categories may be enhanced or augmented in ways that take into account a user's interaction with paper versions of the stored files. Categories such as “My Printed Documents” or “My Recently-Read Documents” might usefully be identified and incorporated in its operations.
• markup Since important aspects of the system are typically provided using the "markup" concepts discussed in Section 5, it would clearly be advantageous to have support for such markup provided by the OS in a way that was accessible to multiple applications as well as to the OS itself. In addition, layers of markup may be provided by the OS, based on its own knowledge of documents under its control and the facilities it is able to provide.
• OS DRM facilities An increasing number of operating systems support some form of "Digital Rights Management": the ability to control the use of particular data according to the rights granted to a particular user, software entity or machine. It may inhibit unauthorized copying or distribution of a particular document, for example.
• the user interface of the system may be entirely on a PC, if the capture device is relatively dumb and is connected to it by a cable, or entirely on the device, if it is sophisticated and with significant processing power of its own. In some cases, some functionality resides in each component. Part, or indeed all, of the system's functionality may also be implemented on other devices such as mobile phones or
• a handheld scanner may have a variety of ways of providing feedback to the user about particular conditions. The most obvious types are direct visual, where the scanner incorporates indicator lights or even a full display, and auditory, where the scanner can make beeps, clicks or other sounds. Important alternatives include tactile feedback, where the scanner can vibrate, buzz, or otherwise stimulate the user's sense of touch, and projected feedback, where it indicates a status by projecting onto the paper anything from a colored spot of light to a sophisticated display.
• Important immediate feedback that may be provided on the device includes:
• the device may provide a variety of ways for the user to provide input in addition to basic text capture. Even when the device is in close association with a host machine that has input options such as keyboards and mice, it can be disruptive for the user to switch back and forth between manipulating the scanner and using a mouse, for example.
• the handheld scanner may have buttons, scroll/jog-wheels, touch- sensitive surfaces, and/or accelerometers for detecting the movement of the device. Some of these allow a richer set of interactions while still holding the scanner.
• the system in response to scanning some text, the system presents the user with a set of several possible matching documents.
• the user uses a scroll- wheel on the side of the scanner is to select one from the list, and clicks a button to confirm the selection. 12.1.3. Gestures
• the user can indicate a large region of text by scanning the first few words in conventional left-to-right order, and the last few in reverse order, i.e. right to left.
• the user can also indicate the vertical extent of the text of interest by moving the scanner down the page over several lines.
• a backwards scan might indicate cancellation of the previous scan operation.
• the device may be used to capture text when it is out of contact with other parts of the system.
• a very simple device may simply be able to store the image or audio data associated with the capture, ideally with a timestamp indicating when it was captured.
• the various captures may be uploaded to the rest of the system when the device is next in contact with it, and handled then.
• the device may also upload other data associated with the captures, for example voice annotations associated with optical scans, or location information.
• More sophisticated devices may be able to perform some or all of the system operations themselves despite being disconnected. Various techniques for improving their ability to do so are discussed in Section 15.3. Often it will be the case that some, but not all, of the desired actions can be performed while offline. For example, the text may be recognized, but identification of the source may depend on a connection to an Internet-based search engine. In some embodiments, the device therefore stores sufficient information about how far each operation has progressed for the rest of the system to proceed efficiently when connectivity is restored.
• the operation of the system will, in general, benefit from immediately available connectivity, but there are some situations in which performing several captures and then processing them as a batch can have advantages. For example, as discussed in Section 13 below, the identification of the source of a particular capture may be greatly enhanced by examining other captures made by the user at approximately the same time. In a fully connected system where live feedback is being provided to the user, the system is only able to use past captures when processing the current one. If the capture is one of a batch stored by the device when offline, however, the system will be able to take into account any data available from later captures as well as earlier ones when doing its analysis.
• a scanner will often communicate with some other device, such as a PC, PDA, phone or digital camera to perform many of the functions of the system, including more detailed interactions with the user.
• the document may be retrieved from local storage or a remote location. (Section 8)
• the application may scroll to, highlight, move the insertion point to, or otherwise indicate the location of the capture. (Section 11.3)
• Annotations may be associated with the document or the captured text. These may come from immediate user input, or may have been captured earlier, for example in the case of voice annotations associated with an optical scan. (Section 19.4)
• Markup may be examined to determine a set of further possible operations for the user to select.
• the scanner device projects a popup menu onto the paper document.
• a user may select from such menus using traditional methods such as a keyboard and mouse, or by using controls on the capture device (Section 12.1.2), gestures (Section 12.1.3), or by interacting with the computer display using the scanner (Section 12.2.4).
• the popup menus which can appear as a result of a capture include default items representing actions which occur if the user does not respond - for example, if the user ignores the menu and makes another capture. 12.2.3. Feedback on disambiguation
• the system provides a real-time display of the documents or the locations found, for example in list, thumbnail-image or text-segment form, and for the number of elements in that display to reduce in number as capture continues.
• the system displays thumbnails of all candidate documents, where the size or position of the thumbnail is dependent on the probability of it being the correct match.
• the text captured will occur in many documents and will be recognized to be a quotation.
• the system may indicate this on the screen, for example by grouping documents containing a quoted reference around the original source document.
• Some optical scanners may be able to capture text displayed on a screen as well as on paper. Accordingly, the term rendered document is used herein to indicate that printing onto paper is not the only form of rendering, and that the capture of text or symbols for use by the system may be equally valuable when that text is displayed on an electronic display.
• the user of the described system may be required to interact with a computer screen for a variety of other reasons, such as to select from a list of options. It can be inconvenient for the user to put down the scanner and start using the mouse or keyboard. Other sections have described physical controls on the scanner (Section 12.1.2) or gestures (Section 12.1.3) as methods of input which do not require this change of tool, but using the scanner on the screen itself to scan some text or symbol is an important alternative provided by the system. [00255] In some embodiments, the optics of the scanner allow it to be used in a similar manner to a light-pen, directly sensing its position on the screen without the need for actual scanning of text, possibly with the aid of special hardware or software on the computer.
• Another example of useful context is the user's geographical location.
• a user in Paris is much more likely to be reading Le Monde than the Seattle Times, for example.
• the timing, size and geographical distribution of printed versions of the documents can therefore be important, and can to some degree be deduced from the operation of the system.
• the time of day may also be relevant, for example in the case of a user who always reads one type of publication on the way to work, and a different one at lunchtime or on the train going home.
• Section 14 covers the analysis of the data stream resulting from paper- based searches, but it should be noted here that statistics about the popularity of documents with other readers, about the timing of that popularity, and about the parts of documents most frequently scanned are all examples of further factors which can be beneficial in the search process.
• the system brings the possibility of Google-type page-ranking to the world of paper.
• One example is connecting one user with others who have related interests. These may be people already known to the user.
• the system may ask a university professor, "Did you know that your colleague at XYZ University has also just read this paper?"
• the system may ask a user, "Do you want to be linked up with other people in your neighborhood who are also how reading Jane Eyre?"
• Such links may be the basis for the automatic formation of book clubs and similar social structures, either in the physical world or online.
• Section 10.6 has already mentioned the idea of offering products and services to an individual user based on their interactions with the system.
• Current online booksellers for example, often make recommendations to a user based on their previous interactions with the bookseller. Such recommendations become much more useful when they are based on interactions with the actual books.
• the user will also not just be capturing some text, but will be causing some action to occur as a result. It might be emailing a reference to the document to an acquaintance, for example. Even in the absence of information about the identity of the user or the recipient of the email, the knowledge that somebody considered the document worth emailing is very useful.
• a capture device for use with the system needs little more than a way of capturing text from a rendered version of the document.
• this capture may be achieved through a variety of methods including taking a photograph of part of the document or typing some words into a mobile phone keypad.
• This capture may be achieved using a small hand-held optical scanner capable of recording a line or two of text at a time, or an audio capture device such as a voice-recorder into which the user is reading text from the document.
• the device used may be a combination of these - an optical scanner which could also record voice annotations, for example - and the capturing functionality may be built into some other device such as a mobile phone, PDA, digital camera or portable music player.
• Input and output Many of the possibly beneficial additional input and output facilities for such a device have been described in Section 12.1. They include buttons, scroll- wheels and touch-pads for input, and displays, indicator lights, audio and tactile transducers for output. Sometimes the device will incorporate many of these, sometimes very few. Sometimes the capture device will be able to communicate with another device that already has them (Section 15.6), for example using a wireless link, and sometimes the capture functionality will be incorporated into such other device (Section 15.7).
• the device implements the majority of the system itself. In some embodiments, however, it often communicates with a PC or other computing device and with the wider world using communications facilities.
• these communications facilities are in the form of a general- purpose data network such as Ethernet, 802.11 or UWB or a standard peripheral- connecting network such as USB, IEEE-1394 (Firewire), BluetoothTM or infra-red.
• a wired connection such as Firewire or USB
• the capture device may appear to a connected machine to be a conventional peripheral such as a USB storage device.
• the device may in some circumstances “dock” with another device, either to be used in conjunction with that device or for convenient storage.
• Sections 3.5 and 12.1.4 have raised the topic of disconnected operation.
• the device can still be useful, though the functionality available will sometimes be reduced.
• the device can record the raw image or audio data being captured and this can be processed later. For the user's benefit, however, it can be important to give feedback where possible about whether the data captured is likely to be sufficient for the task in hand, whether it can be recognized or is likely to be recognizable, and whether the source of the data can be identified or is likely to be identifiable later. The user will then know whether their capturing activity is worthwhile.
• the raw data can still be stored so that, at the very least, the user can refer to them later.
• the user may be presented with the image of a scan, for example, when the scan cannot be recognized by the OCR process.
• the SimpleScanner has a scanning head able to read pixels from the page as it is moved along the length of a line of text. It can detect its movement along the page and record the pixels with some information about the movement. It also has a clock, which allows each scan to be time-stamped. The clock is synchronized with a host device when the SimpleScanner has connectivity. The clock may not represent the actual time of day, but relative times may be determined from it so that the host can deduce the actual time of a scan, or at worst the elapsed time between scans.
• the SimpleScanner does not have sufficient processing power to perform any OCR itself, but it does have some basic knowledge about typical word- lengths, word-spacings, and their relationship to font size. It has some basic indicator lights which tell the user whether the scan is likely to be readable, whether the head is being moved too fast, too slowly or too inaccurately across the paper, and when it determines that sufficient words of a given size are likely to have been scanned for the document to be identified.
• the SimpleScanner has a USB connector and can be plugged into the USB port on a computer, where it will be recharged. To the computer it appears to be a USB storage device on which time-stamped data files have been recorded, and the rest of the system software takes over from this point.
• the SuperScanner also depends on connectivity for its full operation, but it has a significant amount of on-board storage and processing which can help it make better judgments about the data captured while offline.
• the captured pixels are stitched together and passed to an OCR engine that attempts to recognize the text.
• a number of fonts including those from the user's most-read publications, have been downloaded to it to help perform this task, as has a dictionary that is synchronized with the user's spelling-checker dictionary on their PC and so contains many of the words they frequently encounter.
• Also stored on the scanner is a list of words and phrases with the typical frequency of their use - this may be combined with the dictionary. The scanner can use the frequency statistics both to help with the recognition process and also to inform its judgment about when a sufficient quantity of text has been captured; more frequently used phrases are less likely to be useful as the basis for a search query.
• the full index for the articles in the recent issues of the newspapers and periodicals most commonly read by the user are stored on the device, as are the indices for the books the user has recently purchased from an online bookseller, or from which the user has scanned anything within the last few months.
• the titles of several thousand of the most popular publications which have data available for the system are stored so that, in the absence of other information the user can scan the title and have a good idea as to whether or not captures from a particular work are likely to be retrievable in electronic form later.
• the system informs user that the captured data has been of sufficient quality and of a sufficient nature to make it probable that the electronic copy can be retrieved when connectivity is restored. Often the system indicates to the user that the scan is known to have been successful and that the context has been recognized in one of the on-board indices, or that the publication concerned is known to be making its data available to the system, so the later retrieval ought to be successful.
• the SuperScanner docks in a cradle connected to a PC's Firewire or USB port, at which point, in addition to the upload of captured data, its various onboard indices and other databases are updated based on recent user activity and new publications. It also has the facility to connect to wireless public networks or to communicate via Bluetooth to a mobile phone and thence with the public network when such facilities are available.
• Some embodiments of the system use a scanner that scans in contact with the paper, and which, instead of lenses, uses an image conduit a bundle of optical fibers to transmit the image from the page to the optical sensor device.
• a scanner that scans in contact with the paper, and which, instead of lenses, uses an image conduit a bundle of optical fibers to transmit the image from the page to the optical sensor device.
• Such a device can be shaped to allow it to be held in a natural position; for example, in some embodiments, the part in contact with the page is wedge-shaped, allowing the user's hand to move more naturally over the page in a movement similar to the use of a highlighter pen.
• the conduit is either in direct contact with the paper or in close proximity to it, and may have a replaceable transparent tip that can protect the image conduit from possible damage.
• the scanner may be used to scan from a screen as well as from paper, and the material of the tip can be chosen to reduce the likelihood of damage to such displays.
• some embodiments of the device will provide feedback to the user during the scanning process which will indicate through the use of light, sound or tactile feedback when the user is scanning too fast, too slow, too unevenly or is drifting too high or low on the scanned line.
• the capture device may form an important part of identification and authorization for secure transactions, purchases, and a variety of other operations. It may therefore incorporate, in addition to the circuitry and software required for such a role, various hardware features that can make it more secure, such as a smartcard reader, RFID, or a keypad on which to type a PIN.
• the scanning head may also be able to read a fingerprint.
• the voice pattern of the user may be used.
• the device is able to form an association with other nearby devices to increase either its own or their functionality.
• it uses the display of a nearby PC or phone to give more detailed feedback about its operation, or uses their network connectivity.
• the device may, on the other hand, operate in its role as a security and identification device to authenticate operations performed by the other device. Or it may simply form an association in order to function as a peripheral to that device.
• An interesting aspect of such associations is that they may be initiated and authenticated using the capture facilities of the device. For example, a user wishing to identify themselves securely to a public computer terminal may use the scanning facilities of the device to scan a code or symbol displayed on a particular area of the terminal's screen and so effect a key transfer. An analogous process may be performed using audio signals picked up by a voice-recording device.
• the functionality of the capture device is integrated into some other device that is already in use.
• the integrated devices may be able to share a power supply, data capture and storage capabilities, and network interfaces. Such integration may be done simply for convenience, to reduce cost, or to enable functionality that would not otherwise be available.
• Some examples of devices into which the capture functionality can be integrated include:
• an existing peripheral such as a mouse, a stylus, a USB “webcam” camera, a BluetoothTM headset or a remote control
• another processing/storage device such as a PDA, an MP3 player, a voice recorder, a digital camera or a mobile phone
• the phone hardware is not modified to support the system, such as where the text capture can be adequately done through voice recognition, where they can either be processed by the phone itself, or handled by a system at the other end of a telephone call, or stored in the phone's memory for future processing.
• voice recognition can either be processed by the phone itself, or handled by a system at the other end of a telephone call, or stored in the phone's memory for future processing.
• Many modern phones have the ability to download software that could implement some parts of the system.
• voice capture is likely to be suboptimal in many situations, however, for example when there is substantial background noise, and accurate voice recognition is a difficult task at the best of times.
• the audio facilities may best be used to capture voice annotations.
• the camera built into many mobile phones is used to capture an image of the text.
• the phone display which would normally act as a viewfinder for the camera, may overlay on the live camera image information about the quality of the image and its suitability for OCR, which segments of text are being captured, and even a transcription of the text if the OCR can be performed on the phone.
• the phone is modified to add dedicated capture facilities, or to provide such functionality in a clip-on adaptor or a separate Bluetooth- connected peripheral in communication with the phone.
• the phone has connectivity with the wider world, which means that queries can be submitted to remote search engines or other parts of the system, and copies of documents may be retrieved for immediate storage or viewing.
• a phone typically has sufficient processing power for many of the functions of the system to be performed locally, and sufficient storage to capture a reasonable amount of data. The amount of storage can also often be expanded by the user. Phones have reasonably good displays and audio facilities to provide user feedback, and often a vibrate function for tactile feedback. They also have good power supplies.
• the Life Library (see also Section 6.1.1) is a digital archive of any important documents that the subscriber wishes to save and is a set of embodiments of services of this system. Important books, magazine articles, newspaper clippings, etc., can all be saved in digital form in the Life Library. Additionally, the subscriber's annotations, comments, and notes can be saved with the documents.
• the Life Library can be accessed via the Internet and World Wide Web.
• the system creates and manages the Life Library document archive for subscribers.
• the subscriber indicates which documents the subscriber wishes to have saved in his life library by scanning information from the document or by otherwise indicating to the system that the particular document is to be added to the subscriber's Life Library.
• the scanned information is typically text from the document but can also be a barcode or other code identifying the document.
• the system accepts the code and uses it to identify the source document. After the document is identified the system can store either a copy of the document in the user's Life Library or a link to a source where the document may be obtained.
• One embodiment of the Life Library system can check whether the subscriber is authorized to obtain the electronic copy. For example, if a reader scans text or an identifier from a copy of an article in the New York Times (NYT) so that the article will be added to the reader's Life Library, the Life Library system will verify with the NYT whether the reader is subscribed to the online version of the NYT; if so, the reader gets a copy of the article stored in his Life Library account; if not, information identifying the document and how to order it is stored in his Life Library account.
• NYT New York Times
• the system maintains a subscriber profile for each subscriber that includes access privilege information.
• Document access information can be compiled in several ways, two of which are: 1) the subscriber supplies the document access information to the Life Library system, along with his account names and passwords, etc., or 2) the Life Library service provider queries the publisher with the subscriber's information and the publisher responds by providing access to an electronic copy if the Life Library subscriber is authorized to access the material. If the Life Library subscriber is not authorized to have an electronic copy of the document, the publisher provides a price to the Life Library service provider, which then provides the customer with the option to purchase the electronic document.
• the Life Library service provider either pays the publisher directly and bills the Life Library customer later or the Life Library service provider immediately bills the customer's credit card for the purchase.
• the Life Library service provider would get a percentage of the purchase price or a small fixed fee for facilitating the transaction.
• the system can archive the document in the subscriber's personal library and/or any other library to which the subscriber has archival privileges. For example, as a user scans text from a printed document, the Life Library system can identify the rendered document and its electronic counterpart. After the source document is identified, the Life Library system might record information about the source document in the user's personal library and in a group library to which the subscriber has archival privileges.
• Group libraries are collaborative archives such as a document repository for: a group working together on a project, a group of academic researchers, a group web log, etc.
• the life library can be organized in many ways: chronologically, by topic, by level of the subscriber's interest, by type of publication (newspaper, book, magazine, technical paper, etc.), where read, when read, by ISBN or by Dewey decimal, etc.
• the system can learn classifications based on how other subscribers have classified the same document. The system can suggest classifications to the user or automatically classify the document for the user.
• annotations may be inserted directly into the document or may be maintained in a separate file.
• the article is archived in his Life Library with the scanned text highlighted.
• the article is archived in his Life Library along with an associated annotation file (thus leaving the archived document unmodified).
• Embodiments of the system can keep a copy of the source document in each subscriber's library, a copy in a master library that many subscribers can access, or link to a copy held by the publisher.
• the Life Library stores only the user's modifications to the document (e.g., highlights, etc.) and a link to an online version of the document (stored elsewhere). The system or the subscriber merges the changes with the document when the subscriber subsequently retrieves the document.
• the source document and the annotation file are provided to the subscriber and the subscriber combines them to create a modified document.
• the system combines the two files prior to presenting them to the subscriber.
• the annotation file is an overlay to the document file and can be overlaid on the document by software in the subscriber's computer.
• Subscribers to the Life Library service pay a monthly fee to have the system maintain the subscriber's archive. Alternatively, the subscriber pays a small amount (e.g., a micro-payment) for each document stored in the archive. Alternatively, the subscriber pays to access the subscriber's archive on a per-access fee. Alternatively, subscribers can compile libraries and allow others to access the materials/annotations on a revenue share model with the Life Library service provider and copyright holders. Alternatively, the Life Library service provider receives a payment from the publisher when the Life Library subscriber orders a document (a revenue share model with the publisher, where the Life Library service provider gets a share of the publisher's revenue).
• a small amount e.g., a micro-payment
• subscribers can compile libraries and allow others to access the materials/annotations on a revenue share model with the Life Library service provider and copyright holders.
• the Life Library service provider receives a payment from the publisher when the Life Library subscriber orders a document (a revenue share model with the publisher, where
• the Life Library service provider acts as an intermediary between the subscriber and the copyright holder (or copyright holder's agent, such as the Copyright Clearance Center, a.k.a. CCC) to facilitate billing and payment for copyrighted materials.
• the Life Library service provider uses the subscriber's billing information and other user account information to provide this intermediation service. Essentially, the Life Library service provider leverages the pre-existing relationship with the subscriber to enable purchase of copyrighted materials on behalf of the subscriber.
• the Life Library system can store excerpts from documents. For example, when a subscriber scans text from a paper document, the regions around the scanned text are excerpted and placed in the Life Library, rather than the entire document being archived in the life library. This is especially advantageous when the document is long because preserving the circumstances of the original scan prevents the subscriber from re-reading the document to find the interesting portions. Of course, a hyperlink to the entire electronic counterpart of the paper document can be included with the excerpt materials.
• the system also stores information about the document in the Life Library, such as author, publication title, publication date, publisher, copyright holder (or copyright holder's licensing agent), ISBN, links to public annotations of the document, readrank, etc.
• information about the document is a form of paper document metadata.
• Third parties may create public annotation files for access by persons other than themselves, such the general public. Linking to a third party's commentary on a document is advantageous because reading annotation files of other users enhances the subscriber's understanding of the document.
• the system archives materials by class. This feature allows a Life Library subscriber to quickly store electronic counterparts to an entire class of paper documents without access to each paper document. For example, when the subscriber scans some text from a copy of National Geographic magazine, the system provides the subscriber with the option to archive all back issues of the National Geographic. If the subscriber elects to archive all back issues, the Life Library service provider would then verify with the National Geographic Society whether the subscriber is authorized to do so. If not, the Life Library service provider can mediate the purchase of the right to archive the National Geographic magazine collection.
• Life Saver A variation on, or enhancement of, the Life Library concept is the "Life Saver," where the system uses the text captured by a user to deduce more about their other activities.
• the scanning of a menu from a particular restaurant, a program from a particular theater performance, a timetable at a particular railway station, or an article from a local newspaper allows the system to make deductions about the user's location and social activities, and could construct an automatic diary for them, for example as a website.
• the user would be able to edit and modify the diary, add additional materials such as photographs and, of course, look again at the items scanned.
• Portable scanners supported by the described system have many compelling uses in the academic setting. They can enhance student/teacher interaction and augment the learning experience. Among other uses, students can annotate study materials to suit their unique needs; teachers can monitor classroom performance; and teachers can automatically verify source materials cited in student assignments.
• a child's interaction with a paper document, such as a book, is monitored by a literacy acquisition system that employs a specific set of embodiments of this system.
• the child uses a portable scanner that communicates with other elements of the literacy acquisition system.
• the literacy acquisition system includes a computer having a display and speakers, and a database accessible by the computer.
• the scanner is coupled with the computer (hardwired, short range RF, etc.).
• the literacy acquisition system compares the scanned text with the resources in its database to identify the word.
• the database includes a dictionary, thesaurus, and/or multimedia files (e.g., sound, graphics, etc.).
• multimedia files e.g., sound, graphics, etc.
• the system uses the computer speakers to pronounce the word and its definition to the child.
• the word and its definition are displayed by the literacy acquisition system on the computer's monitor.
• Multimedia files about the scanned word can also be played through the computer's monitor and speakers. For example, if a child reading "Goldilocks and the Three Bears" scanned the word "bear," the system might pronounce the word "bear” and play a short video about bears on the computer's monitor. In this way, the child learns to pronounce the written word and is visually taught what the word means via the multimedia presentation.
• the literacy acquisition system provides immediate auditory and/or visual information to enhance the learning process.
• the child uses this supplementary information to quickly acquire a deeper understanding of the written material.
• the system can be used to teach beginning readers to read, to help children acquire a larger vocabulary, etc.
• This system provides the child with information about words with which the child is unfamiliar or about which the child wants more information.
• the system compiles personal dictionaries. If the reader sees a word that is new, interesting, or particularly useful or troublesome, the reader saves it (along with its definition) to a computer file. This computer file becomes the reader's personalized dictionary. This dictionary is generally smaller in size than a general dictionary so can be downloaded to a mobile station or associated device and thus be available even when the system isn't immediately accessible.
• the personal dictionary entries include audio files to assist with proper word pronunciation and information identifying the paper document from which the word was scanned.
• the system creates customized spelling and vocabulary tests for students. For example, as a student reads an assignment, the student may scan unfamiliar words with the portable scanner. The system stores a list of all the words that the student has scanned. Later, the system administers a customized spelling/vocabulary test to the student on an associated monitor (or prints such a test on an associated printer).
• Teachers can use the system to detect plagiarism or to verify sources by scanning text from student papers and submitting the scanned text to the system. For example, a teacher who wishes to verify that a quote in a student paper came from the source that the student cited can scan a portion of the quote and compare the title of the document identified by the system with the title of the document cited by the student. Likewise, the system can use scans of text from assignments submitted as the student's original work to reveal if the text was instead copied.
• capturing text from an academic textbook links students or staff to more detailed explanations, further exercises, student and staff discussions about the material, related example past exam questions, further reading on the subject, recordings of the lectures on the subject, and so forth. (See also Section 7.1.)
• the system is used to teach foreign languages. Scanning a Spanish word, for example, might cause the word to be read aloud in Spanish along with its definition in English.
• the system provides immediate auditory and/or visual information to enhance the new language acquisition process.
• the reader uses this supplementary information to acquire quickly a deeper understanding of the material.
• the system can be used to teach beginning students to read foreign languages, to help students acquire a larger vocabulary, etc.
• the system provides information about foreign words with which the reader is unfamiliar or for which the reader wants more information.
• Reader interaction with a paper document, such as a newspaper or book, is monitored by a language skills system.
• the reader has a portable scanner that communicates with the language skills system.
• the language skills system includes a computer having a display and speakers, and a database accessible by the computer.
• the scanner communicates with the computer (hardwired, short range RF, etc.).
• the database includes a foreign language dictionary, thesaurus, and/or multimedia files (sound, graphics, etc.).
• the system compares the scanned text with the resources in its database to identify the scanned word. After the word has been identified, the system uses the computer speakers to pronounce the word and its definition to the reader.
• the word and its definition are both displayed on the computer's monitor.
• Multimedia files about grammar tips related to the scanned word can also be played through the computer's monitor and speakers. For example, if the words "to speak” are scanned, the system might pronounce the word “hablar,” play a short audio clip that demonstrates the proper Spanish pronunciation, and display a complete list of the various conjugations of "hablar.” In this way, the student learns to pronounce the written word, is visually taught the spelling of the word via the multimedia presentation, and learns how to conjugate the verb. The system can also present grammar tips about the proper usage of "hablar" along with common phrases.
• the user scans a word or short phrase from a rendered document in a language other than the user's native language (or some other language that the user knows reasonably well).
• the system maintains a prioritized list of the user's "preferred" languages.
• the system identifies the electronic counterpart of the rendered document, and determines the location of the scan within the document.
• the system also identifies a second electronic counterpart of the document that has been translated into one of the user's preferred languages, and determines the location in the translated document corresponding to the location of the scan in the original document.
• the system identifies a small region (e.g., a paragraph) that includes the corresponding location of the scanned location.
• the corresponding translated location is then presented to the user. This provides the user with a precise translation of the particular usage at the scanned location, including any slang or other idiomatic usage that is often difficult to accurately translate on a word-by-word basis.
• a user researching a particular topic may encounter all sorts of material, both in print and on screen, which they might wish to record as relevant to the topic in some personal archive.
• the system would enable this process to be automatic as a result of scanning a short phrase in any piece of material, and could also create a bibliography suitable for insertion into a publication on the subject.
• Fee-based searching and indexing Conventional Internet search engines typically provide free search of electronic documents, and also make no charge to the content providers for including their content in the index.
• the system provides for charges to users and/or payments to search engines and/or content providers in connection with the operation and use of the system.
• subscribers to the system's services pay a fee for searches originating from scans of paper documents.
• a stockbroker may be reading a Wall Street Journal article about a new product offered by Company X.
• the stockbroker uses the system to search special or proprietary databases to obtain premium information about the company, such as analyst's reports.
• the system can also make arrangements to have priority indexing of the documents most likely to be read in paper form, for example by making sure all of the newspapers published on a particular day are indexed and available by the time they hit the streets.
• Content providers may pay a fee to be associated with certain terms in search queries submitted from paper documents. For example, in one embodiment, the system chooses a most preferred content provider based on additional context about the provider (the context being, in this case, that the content provider has paid a fee to be moved up the results list). In essence, the search provider is adjusting paper document search results based on pre-existing financial arrangements with a content provider. See also the description of keywords and key phrases in Section 5.2.
• Such content may be protected by a firewall and thus not generally indexable by third parties.
• the content provider may nonetheless wish to provide an index to the protected content.
• the content provider can pay a service provider to provide the content provider's index to system subscribers.
• a law firm may index all of a client's documents. The documents are stored behind the law firm's firewall.
• the law firm wants its employees and the client to have access to the documents through the portable scanner so it provides the index (or a pointer to the index) to the service provider, which in turn searches the law firm's index when employees or clients of the law firm submit paper-scanned search terms via their portable scanners.
• the law firm can provide a list of employees and/or clients to the service provider's system to enable this function or the system can verify access rights by querying the law firm prior to searching the law firm's index.
• the index provided by the law firm is only of that client's documents, not an index of all documents at the law firm.
• the service provider can only grant the law firm's clients access to the documents that the law firm indexed for the client.
• the search function revenue can be generated from paid subscriptions from the scanner users, but can also be generated on a per-search charge.
• the content delivery revenue can be shared with the content provider or copyright holder (the service provider can take a percentage of the sale or a fixed fee, such as a micropayment, for each delivery), but also can be generated by a "referral" model in which the system gets a fee or percentage for every item that the subscriber orders from the online catalog and that the system has delivered or contributed to, regardless of whether the service provider intermediates the transaction.
• the system service provider receives revenue for all purchases that the subscriber made from the content provider, either for some predetermined period of time or at any subsequent time when a purchase of an identified product is made.
• Consumers may use the portable scanner to make purchases from paper catalogs.
• the subscriber scans information from the catalog that identifies the catalog. This information is text from the catalog, a bar code, or another identifier of the catalog.
• the subscriber scans information identifying the products that s/he wishes to purchase.
• the catalog mailing label may contain a customer identification number that identifies the customer to the catalog vendor. If so, the subscriber can also scan this customer identification number.
• the system acts as an intermediary between the subscriber and the vendor to facilitate the catalog purchase by providing the customer's selection and customer identification number to the vendor.
• a consumer scans paper coupons and saves an electronic copy of the coupon in the scanner, or in a remote device such as a computer, for later retrieval and use.
• An advantage of electronic storage is that the consumer is freed from the burden of carrying paper coupons.
• a further advantage is that the electronic coupons may be retrieved from any location.
• the system can track coupon expiration dates, alert the consumer about coupons that will expire soon, and/or delete expired coupons from storage.
• An advantage for the issuer of the coupons is the possibility of receiving more feedback about who is using the coupons and when and where they are captured and used. 19.
• the system may be used to auto-populate an electronic document that corresponds to a paper form.
• a user scans in some text or a barcode that uniquely identifies the paper form.
• the scanner communicates the identity of the form and information identifying the user to a nearby computer.
• the nearby computer has an Internet connection.
• the nearby computer can access a first database of forms and a second database having information about the user of the scanner (such as a service provider's subscriber information database).
• the nearby computer accesses an electronic version of the paper form from the first database and auto-populates the fields of the form from the user's information obtained from the second database.
• the nearby computer then emails the completed form to the intended recipient.
• the computer could print the completed form on a nearby printer.
• the system has a portable scanner that contains the user's information, such as in an identity module, SIM, or security card.
• the scanner provides information identifying the form to the nearby PC.
• the nearby PC accesses the electronic form and queries the scanner for any necessary information to fill out the form.
• the system can be used to automatically populate electronic address books or other contact lists from paper documents. For example, upon receiving a new acquaintance's business card, a user can capture an image of the card with his/her cellular phone. The system will locate an electronic copy of the card, which can be used to update the cellular phone's onboard address book with the new acquaintance's contact information. The electronic copy may contain more information about the new acquaintance than can be squeezed onto a business card. Further, the onboard address book may also store a link to the electronic copy such that any changes to the electronic copy will be automatically updated in the cell phone's address book.
• the business card optionally includes a symbol or text that indicates the existence of an electronic copy.
• the cellular phone can use OCR and knowledge of standard business card formats to fill out an entry in the address book for the new acquaintance. Symbols may also aid in the process of extracting information directly from the image. For example, a phone icon next to the phone number on the business card can be recognized to determine the location of the phone number.
• the system can enhance the proofreading and editing process.
• One way the system can enhance the editing process is by linking the editor's interactions with a paper document to its electronic counterpart.
• the system will make the appropriate annotations or edits to an electronic counterpart of the paper document. For example, if the editor scans a portion of text and makes the "new paragraph" control gesture with the scanner, a computer in communication with the scanner would insert a "new paragraph” break at the location of the scanned text in the electronic copy of the document.
• a user can make voice annotations to a document by scanning a portion of text from the document and then making a voice recording that is associated with the scanned text.
• the scanner has a microphone to record the user's verbal annotations. After the verbal annotations are recorded, the system identifies the document from which the text was scanned, locates the scanned text within the document, and attaches the voice annotation at that point. In some embodiments, the system converts the speech to text and attaches the annotation as a textual comment.
• the system keeps annotations separate from the document, with only a reference to the annotation kept with the document.
• the annotations then become an annotation markup layer to the document for a specific subscriber or group of users.
• the system identifies the document, opens it using a software package, scrolls to the location of the scan and plays the voice annotation. The user can then interact with a document while referring to voice annotations, suggested changes or other comments recorded either by themselves or by somebody else.
• a markup layer associated with a paper document contains help menu information for the document. For example, when a user scans text from a certain portion of the document, the system checks the markup associated with the document and presents a help menu to the user. The help menu is presented on a display on the scanner or on an associated nearby display.
• the portable scanner is used to scan information from computer monitors and televisions.
• the portable optical scanner has an illumination sensor that is optimized to work with traditional cathode ray tube (CRT) display techniques such as rasterizing, screen blanking, etc.
• CTR cathode ray tube
• a voice capture device which operates by capturing audio of the user reading text from a document will typically work regardless of whether that document is on paper, on a display, or on some other medium.
• a public kiosk displays a dynamic session ID on its monitor.
• the kiosk is connected to a communication network such as the Internet or a corporate intranet.
• the session ID changes periodically but at least every time that the kiosk is used so that a new session ID is displayed to every user.
• the subscriber scans in the session ID displayed on the kiosk; by scanning the session ID, the user tells the system that he wishes to temporarily associate the kiosk with his scanner for the delivery of content resulting from scans of printed documents or from the kiosk screen itself.
• the scanner may communicate the Session ID and other information authenticating the scanner (such as a serial number, account number, or other identifying information) directly to the system.
• the scanner can communicate directly (where "directly" means without passing the message through the kiosk) with the system by sending the session initiation message through the user's cell phone (which is paired with the user's scanner via BluetoothTM).
• the scanner can establish a wireless link with the kiosk and use the kiosk's communication link by transferring the session initiation information to the kiosk (perhaps via short range RF such as BluetoothTM, etc.); in response, the kiosk sends the session initiation information to the system via its Internet connection.
• the system can prevent others from using a device that is already associated with a scanner during the period (or session) in which the device is associated with the scanner. This feature is useful to prevent others from using a public kiosk before another person's session has ended.
• the user scans a barcode on a monitor of a PC which s/he desires to use; in response, the system sends a session ID to the monitor that it displays; the user initiates the session by scanning the session ID from the monitor (or entering it via a keypad or touch screen or microphone on the portable scanner); and the system associates in its databases the session ID with the serial number (or other identifier that uniquely identifies the user's scanner) of his/her scanner so another scanner cannot scan the session ID and use the monitor during his/her session.
• the scanner is in communication (through wireless link such as BluetoothTM, a hardwired link such as a docking station, etc.) with a PC associated with the monitor or is in direct (i.e., w/o going through the PC) communication with the system via another means such as a cellular phone, etc.
• wireless link such as BluetoothTM, a hardwired link such as a docking station, etc.
• PC associated with the monitor
• FIG. 4 is a diagram showing the typical environment in which embodiments of the system operate.
• the system functions in a distributed computing environment 400 that includes a plurality of devices interconnected by a wireless network 401 , the Internet 402 or other networks (not shown). All these communications and connections are interconnected via suitable network connections using suitable network communication protocols.
• the servers and other devices communicate with each other in accordance with respective APIs, which form additional embodiment of the system.
• the devices and servers may communicate in accordance with open/standard protocols.
• the servers and other devices include an OCR device 411 or other text capture device that is used to capture text from a rendered document 412; a wireless device 421 and/or a user device 422 to which the text capture device uploads various representations of captured text and other user input, and via which the system can provide various types of feedback to the user; a user account server 431 and associated user account database 432 in which the system manages user account information for the user; a search engine server 441 and an associated search database 442 that the system uses to perform queries containing text captured from a rendered document to identify positions in electronic documents in which the captured text occurs; and a document server 451 and an associated document database 452 from which the system retrieves copies of documents determined to contain the captured text.
• OCR device 411 or other text capture device that is used to capture text from a rendered document 412
• a wireless device 421 and/or a user device 422 to which the text capture device uploads various representations of captured text and other user input, and via which the system can provide various types of feedback to the
• each server may actually comprise more than a single device in an actual system practicing embodiments of the system.
• the servers include file servers, database servers or a mixture that includes file servers and database servers. It will further be appreciated by those of ordinary skill in the art, that while the various servers may be discussed as separate devices, in other embodiments of the system the servers may reside on a single device.
• the scanner incorporates a magnetic sensor
• data can be coded in documents magnetically as well as optically, acoustically, factually, etc.
• the system achieves this by scanning the desired position within a document to identify a distinctive text "signature," where this signature provides information that can be used to locate the corresponding location within the original digital source document.
• the system sends this digital signature to a server that has access to database of electronic documents, which desirably includes an electronic version of the paper document in question (though, as explained below, useful outcomes can be obtained even when this is not the case).
• the server identifies the corresponding location (or locations) in the electronic source document, connecting it with the original scan of the paper document. Establishing this relationship enables numerous useful innovations related to the use of printed documents in a variety of contexts. Various embodiment of the system are discussed below.
• the system can be viewed as using auxiliary or augmenting information to transform document recognition into document navigation (e.g., finding a location and traversing portions of a document, and generating information for the traversal).
• auxiliary or augmenting information to transform document recognition into document navigation (e.g., finding a location and traversing portions of a document, and generating information for the traversal).
• the device can capture and transmit actual text (e.g., ASCII), or it can use the (offset-based) template-matching representation mentioned elsewhere.
• actual text e.g., ASCII
• offset-based template-matching representation mentioned elsewhere.
• a good example is the capital letter that usually begins an English sentence. Because these are rare, it will not in general be easy to use template matching and disambiguation to interpret these initial capitals.
• One alternative solution is to essentially ignore capitals - by ignoring the first character of each new paragraph, sentence, etc.
• a user may wish to indicate the following sentence, which occurs in a rendered document: "Take as an example this sentence.”
• a conventional OCR system would seek to understand and interpret the letter “T” - in order to ascertain whether the first was “Take” or "Make” or “Fake” or “Rake”, etc.
• the system is only looking for distinguishing reference features for navigation purposes.
• the system can simply drop the "T” and search the source document for "ake as an example” - where this phrase might be represented as letters, offsets, or in another form. So long as the remainder of the phrase constitutes an identifying signature, the interpretation of the initial character is not essential.
• bookmarks each indicating a location in the rendered document. Generally these can be used to find a location in a document at a later time.
• One simple but interesting application is to mark where the user last was reading in the document - very much in keeping with the traditional meaning of "bookmark". It can be useful for the user to have easy access to this information - where they left off in a book or other document. This data might appear on the user's PC or PDA or cellular telephone.
• the device itself indicates the last read location, such as by using a small LCD display in the device itself.
• the display is binary - for example, an LED which is either on or off. This LED might be on if the user scans text which they have already read, and off if they scan new text. In this way the user can "hunt" for where the user left off.
• FIG. 5 is a flow diagram showing steps typically performed by the system to implement bookmarks.
• the system receives text scanned by a user.
• the system preprocesses the text scanned in step 501.
• the system compares the scanned text to a document history maintained for the user.
• the system if the scanned text was previously scanned by the user, then the system continues in step 505 to return an indication of the previous scan, else the system continues in step 506 to locate the scan in a document.
• step 507 if the scan is located before the last bookmark, then the system continues in step 508 to return an indication of being previously read, else the facility continues in step 509 to return an indication of being previously unread.
• FIG. 6 is a diagram showing a scanning device that has an integrated highlighter and pen. It can be seen that scan device 600 contains both a highlighter 601 and a pen 602, which are optionally retractable.
• This overlay can thus be thought of as an abstract, virtual layer.
• This layer can then be merged with or "overlaid" on the source or reference version of the document. In one example, this would occur when the user views the reference document on their computer screen - where the overlaid data from the user's actions would appear on top of, or integrated into, the reference document. Note that the reference document does not need to be modified to be displayed in this way.
• the overlay information is combined with the reference or source document when the user prints the document.
• the system applies the overlay to — or merges it into — the source document when it is delivered electronically to the user. For example, the source document and overlay might be combined into a PDF document and emailed to the user.
• the user's overlaid information can be stored as a separate layer, the source document does not need to be changed. Thus many users can all mark and interact with a single copy of a document. The user's marks and notes are stored separately, so the original doesn't need to change.
• Each user's data is typically small compared to the underlying document.
• highlighting all that needs to be stored is the start and end location of the highlighted text in the document, and the highlight color.
• One way of storing this data is as character offsets from the start of the document.
• Another way is as an address like document:page:line.
• the system stores the actual x-y coordinates of the user's actions in the rendered document.
• Documents used with the system sometimes carry a special mark (e.g., barcode, etc.) that the user can scan to indicate which issue or copy of the document is being scanned. This additional identifying information enables the system to determine which document the user has, and thus how it is rendered.
• a special mark e.g., barcode, etc.
• the system may require the user to scan an identifying code or mark to enable the user to interact with the document using the system. This may be required before the user has used the device on other parts of the document.
• the system may allow the user to interact with the document, but require that they perform an identifying scan at some future point.
• the identifying scan may be optional; without it, the system may have additional ambiguity, i.e., less certainty about the specific document being used. With the extra scan, the system knows more about the specific document.
• the capture device provides an error indicator or signal (e.g., an LED or audible tone) which tells the user that this document either isn't recognized or isn't enabled - i.e., that an identifying scan is desired or required.
• an error indicator or signal e.g., an LED or audible tone
• This identifying scan may be used to specifically indicate which document the user has - for example, their local morning newspaper - which then allows the system to refer to a cached copy of the document or related dictionary for determining disambiguation and location of scans.
• the special marks that are scanned might be 1 or 2 dimensional barcodes, or a specific region of human-readable text, or otherwise coded data.
• a region of text within the rendered document is specifically marked (for example, by a margin mark, or highlighting, or underlining, or special colored ink) to indicate to the user that this region should be scanned for document recognition.
• these scans may be used to help the system know the user's context. In other cases they may enable or unlock features of the system which would not otherwise be available. For example, a purchase from a printed catalog might not be allowed unless the user scans the address label with their identifying code on it.
• the user may be able to scan a region of text specifically to establish their context - which location in which document.
• the terminal device may have a special switch or input to indicate this desired function (setting context).
• the user can perform a special gesture with the device that may indicate a context-setting reference scan function, such as by scanning the text in reverse.
• a motion or gesture with the device may indicate "erasing" or "undoing" a previous action.
• the motions and actions of the device can be used to indicate user intent.
• the device can record that a particular document was scanned at a particular time, place, etc. - including which portions were scanned.
• the optics in the device may also be able of capturing and storing an image of a signature.
• documents may be provided with special marks or codes at one or more locations (perhaps including a unique identifier code for the entire document). These marks might include specially marked or indicated human readable text (e.g., a section of a legal document which is printed in bold, or underlined, etc.).
• a user can then scan various portions of a document to indicate that they have read it.
• they might optionally sign the document and scan in their signature with the terminal device.
• the device itself may incorporate writing means as shown in Figure 6 - in which case the user can both scan and sign with one device.
• a terminal device which can also include hardware and/or software capable of reading coded (e.g. barcode) information.
• coded e.g. barcode
• the terminal device might simply capture an image of the barcode to be read and forward this for interpretation.
• the device recognizes when the image it is scanning is 1 -dimensional, as in a barcode. For example, it may check (in software or electronically in hardware, or some combination of these) whether there is one axis which carries no information.
• a 1-d barcode has this property, being made up of parallel lines assumed to be parallel to the y-axis. In this case, only changes along the x-axis — that cross the lines, perpendicular to them, in the x direction — contain information.
• the device may have local intelligence (hardware and/or software) to reduce this scanned data by collapsing/ignoring the y-axis - i.e., it may be able to partially or completely read or interpret the code (e.g., before communicating with a server).
• local intelligence hardware and/or software
• De-skewing is the process of removing any artificial angular components from scanned or imaged data.
• a situation which often occurs with hand scanners is that the user holds them at an angle, rotated around the axis perpendicular to the page, so that data captured has an artificial angle or slant included. Note that the angle also may vary over time, for example as the user moves their hand across the page. It is helpful if this artificial skew is removed in one of the steps of processing the data or image.
• the system de-skews text by performing a mathematical transform on the data from which the skew angle can be readily determined. This transform can be applied locally so that variations in skew (e.g., across a single line of text) can be detected and measured locally.
• the template-matching (offset-based), convolution approach which can be used in the system has the ability to find matching objects, using earlier occurrences of these objects as templates.
• One interesting result of this ability is that any repeating objects can carry easily readable information - and the tokens which represent this information do not need to be defined in advance or known to the system.
• a document contains a string of 1's and 0's (e.g., a representation of binary data), such as: "100101001.”
• the system needn't recognize and understand the meaning of a "1" or a "0". Rather, it can interpret the sample string as, "an object of a first type, followed by two objects of a second type, followed by an object of the first type, ... etc.” This information in the sample string can just as well be represented by "abbababba” or by "011010110.”
• Data can be encoded using any number of distinct objects or symbols - and that spaces can be treated as one of these objects (if spaces are used, measured distance may be used to count adjacent spaces, or each space object may be limited to always having a count of 1 - e.g., such that "1 11 111 1" would be interpreted the same as "1 11 111 1”). From this perspective, a language written in the 26 character roman alphabet (assuming all lower case) is just a special instance of this encoding, where the number of symbols happens to be 26.
• the system represents the data in a sequence such as the above "011010110” example as offsets - the number of character positions separating recurring instances of each symbol.
• "011010110” is represented as 3,1,2,2,2,3,1,?,? - where each digit corresponds to a character in the original string, and the value of the digit is the distance or offset to the next occurrence of this same character.
• the system exploits the forbidden codes in decoding and/or representing the data.
• the facility exploits the forbidden codes by using them to store additional data, such as an exception code.
• additional data such as an exception code.
• the system may enter a special mode or routine, perhaps treating following codes specially.
• 1 - cant be a 2 (conflicts with 5) or a 3 (conflicts with 4)
• 5 - can't be a 2 (conflicts with 4)
• 5 - can't be a 4 or a 1 (conflicts with 4 and 5)
• 1 - can't be a 4 or a 3 (conflicts with 5 and 5)
• 1 - can't be a 3 or a 2 (conflicts with 5 and 5)
• Another use is to encode additional data.
• the system interprets forbidden codes as "escape sequences" which invoke special actions or processes, or read subsequent embedded data from the stream - after which the system can re-synchronize to the input stream (since the escape sequence can carry information about the sequence length, or this can be known or inferred by the system) and continue processing.
• forbidden codes Yet another use of these forbidden codes is to reduce the amount of information in the encoding of the data.
• One algorithm here might be, "if this is the first (smallest) forbidden code, treat it as the first allowed/valid code; if this is the second (next smallest) forbidden code, treat it as the second allowed/valid code". In this way smaller (forbidden) numbers can represent larger (allowed) codes - so the number of bits used to store or transmit the data is reduced.
• forbidden codes can be combined - for example, the first (smallest) forbidden code might be interpreted as an escape sequence, while higher codes are mapped to the next available valid codes. And other uses of these codes can be applied as well.
• sequence "abcbcbcabcbcbcabcbcbcbcbcbcbc" is first coded as offsets: "72222337222233?2222??”, then reduced to "74(2)2(3)74(2)2(3)?4(2)2(?)", then further reduced to
• the system drafts repeat counts are altogether, and only the objects themselves are stored or transmitted.
• "abcabcabcabcabc” can first be coded as offsets: "333333333333???” which in turn can be represented as count(object): 12(3)??? - which can be stored or transmitted as simply "+3???” - where here the "+” is an indicator that the object repeats.
• the system just stores or transmits "3", dropping any reference to a repeat count.
• the system does not require that this be scanned in its entirety in order to be recognized.
• the offsets from this sequence could be represented (counting spaces as objects) as: "31641153164115" This can be stored or transmitted as a count plus object, or as a "plurality indicator" ("+3164115" - see use of plus sign above), or without any indication of plurality Gust "3164115").
• These latter two examples are roughly akin to matching constructs in regular expressions.
• a separate system e.g., running on a remote server, can then search for, or consult an index looking for, sequences which match this compressed representation. To do so it would treat each object in the sequence as potentially occurring "1 or more times" - an, as with regular expressions, would implement code and algorithms to find these matches.
• a similar efficiency in storage and/or transmission can be obtained by noting that, when character offsets are employed, all scans of text end in unknown offsets (shown above as "???").
• Template matching and/or self-correlation use one instance of a token or object or character or symbol as a template to recognize subsequent occurrences of this same object.
• a token or object or character or symbol as a template to recognize subsequent occurrences of this same object.
• the system captures an image of the text and stores it in memory and/or transmits it.
• the system performs the template matching on-the-fly, by immediately calculating the offsets of matching objects and storing only the individual templates — and these may be discarded when the offsets are known.
• the system in some embodiments, as (or after) the user scans a portion of a horizontal line of text, the system "convolves" the line, i.e., effectively slides a copy of the line horizontally past itself, looking for regions which match well. At the start of this process it may be helpful to determine the baseline of the text, and to de-skew it - both are well-known techniques in the field of document imaging. However, note that there are ways to search for matching regions without either of these steps.
• This process may optionally choose to use connected-region analysis - wherein tokens/objects/characters/symbols are assumed to be comprised of "connected” (i.e., contiguous) pixels or ink. Thus we are in this instance looking for connected regions which match. Note this is related to our knowledge of whitespace, since by definition the region exterior to a connected region of ink is whitespace.
• the system uses (as another of several approaches) simple horizontal .extent. That is, the system pays attention to the horizontal width and position of matching regions of ink or pixels, with no or little attention to connected regions or whitespace - though these components can be introduced to assist the process.
• FIG. 7 is a flow diagram showing steps typically performed by the system to process text capture actions.
• the system receives text captured by the user.
• the system preprocesses the text scanned in step 701.
• the system identifies word and line boundaries in the scanned text.
• the system convolves the text as described above.
• the system uses delimiters to determine boundaries of unknown regions in the text.
• the system processes the scan to produce a representation of the captured text.
• step 707 the system searches for a matching text representation in the corpus of electronic documents.
• step 708 if the search of step 707 is successful, then the system continues in step 709 to return a successful search notice, else the system continues in step 710.
• step 710 if the search can be refined, then the system continues in step 711 , else the system continues in step 712 to return a failed search notice.
• step 711 the system indicates to the user that refinement is needed.
• step 701 the system continues in step 701 to receive additional text captured by the user in response to the indication of step 711.
• the system In matching two regions of text taken from rendered documents the system encounters the issue of "goodness of match.” Because all physical measurements include errors, the system's matching process will essentially never be exact. Therefore, in some embodiments, the system makes decisions as to how well one region matches another. Several tools can be used to accomplish this, many already well known in the fields of OCR, document imaging, and machine vision.
• One way to estimate fit or match used by the system in some embodiments is to first find the best alignment of the objects to be compared, then calculate a difference over the objects. For example, in the case of simple black and white pixels (no gray scale), the system simply finds the pixels which are on in one image/object, and off in the other image/object. The count of these "errors" is a rough estimate of fit.
• This count estimate can be improved by "normalizing it” by (dividing it by) the total number of pixels included.
• the system uses either:
• fit_error #_bad_pixels / #_pixels_in_x_y_region_compared
• fit_error #_bad_pixels / #_pixels_in_object
• the former considers the number of error pixels compared to the area being compared.
• the latter considers the number of error pixels compared to the number or matching, non-error pixels.
• the system adds various refinements to these techniques, or uses other matching techniques from OCR, etc.
• the physical distribution (e.g., x-y coordinates) of the error and matching pixels is of potential importance.
• the matching pixels are "concentrated” (i.e., contiguous, occurring close together), and the error pixels are "distributed” (i.e., non-contiguous, occurring far apart).
• the system compares the image of an "r" to that of an "n". Only a small part of the image contains error pixels - where the right hand side of the "n" descends to the base line, but the "r” does not.
• Figures 8A-8D illustrate how two characters of an alphabet often have approximately the same * relative * shape in multiple fonts.
• Figure 8A shows the letters "D", "C", and "L” in an Arial lower-case font
• Figure 8C shows these letters in a Times New Roman lower-case font. Even though the fonts are very different and distinct, the relationship between these characters * within * a particular font are approximately the same.
• a special case of self-recognition occurs when a set of symbols or tokens (i.e., a new font) is constructed with explicit intent that they be self- recognizable as described above. That is, given a subset of these symbols, the others can be derived or inferred. These symbols can be used to error-check each other, since their graphic components/design are interrelated. This redundancy also provides also robustness against noise.
• Figure 9 is a diagram that illustrates the approach used by the facility in some embodiments to learn an entirely new set of symbols using self recognition.
• the diagram shows a set 900 of symbols comprised of a 2 X 2 array of small "boxes".
• a baseline to establish vertical and horizontal spacing, there are 16 possible symbols: 4 single boxes in the four corners, 6 permutations involving 2 boxes, 4 permutations each having 3 boxes, and one symbol having all 4 boxes, and a whitespace symbol comprised of no boxes.
• Figure 10 is a diagram that shows a subset of the symbols shown in Figure 9 that self-define their vertical and horizontal extent. Each symbol of the subset 1000 is two boxes wide and two boxes high.
• Figure 11 is a diagram that shows relationships between some of the symbols shown in Figure 9.
• each of relationships 1101 and 1102 show a relationship by which one symbol of the set can be constructed out of two other symbols. Note that we can choose as a subset those symbols which have the best redundancy and/or error correction properties. Note that some symbols can be constructed at least two different ways from combinations of symbols in the set. rnn-17-i. *** *** ****** * [00475]
• Several dedicated bar code scanning devices aimed at the consumer market have achieved commercial failure, perhaps because barcode scanning alone does not offer enough value or function for broad consumer adoption.
• the combination of document marking/scanning and barcode scanning creates a combined function which will be of use and interest to a wide array of consumers.
• This combination of text and barcode imaging can be accomplished with many of the same components, hardware and software.
• one really interesting new ingredient is a scanning device which makes use of a reference or source copy of the rendered document being scanned, which enables the user's actions in the rendered document are interpreted with respect to, and mapped to, the source or reference document.
• Another means for locating the source document is if the marked version carries a machine-readable code which identifies the document and/or provides directions for retrieving this source document (e.g., a URL).
• This code might be a barcode, machine-readable font, or any machine-readable means for conveying this information.
• machine-readable document ID and document locator An interesting extension of the machine-readable document ID and document locator is for this data to include access information. That is, included in the machine readable code is information which allows the system to access the document, e.g., when it is password protected, behind a corporate firewall, etc. Note that an additional input might be required by the user or another individual which confirms the request to access this document.
• the system maintains the relationship between a user's notes and marks and the document content and features associated with these notes and marks. This can be important, for example, if the source document is re-rendered in a different style or format, and the system wants to re-display the user's mark at the correct location. As an example, if the user draws a line through a word of text, the system may want to show that line through the same text in subsequent renderings.
• One means to accomplish this is to "anchor" each user mark or group of marks with some recognizable feature in the document (e.g., a single word, punctuation mark, image, etc.). In some embodiments, the system does so by finding the nearest source-document feature (e.g., by geometric distance) and associate the mark with this feature.
• the system finds near-by features, and to weight or rank them, and then associate the user's marks with a high-rated feature.
• the system might examine all near-by words, and associate the mark with the word which is most relevant (e.g., a keyword rather than a stop-word, perhaps related to the topic of the source document text, etc.).
• This aspect of the system can employ some of the many techniques which are known for identifying important elements in a document.
• Notes and marks which have associated anchor points can thus be related to digital documents - such that the relevant note or mark can be accessed when viewing or editing the document (e.g., with a word processor). For example, all such notes might appear as a special symbol embedded in the document (and optionally stored within the document). The user might then mouse-over or click on these symbols to reveal the embedded or associated note or mark. Similarly, the user's notes might be turned on or off via a menu command.

Landscapes

• Engineering & Computer Science (AREA)
• Theoretical Computer Science (AREA)
• Physics & Mathematics (AREA)
• Databases & Information Systems (AREA)
• General Engineering & Computer Science (AREA)
• General Physics & Mathematics (AREA)
• Data Mining & Analysis (AREA)
• Mathematical Physics (AREA)
• Computational Linguistics (AREA)
• Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
• Document Processing Apparatus (AREA)
• Processing Or Creating Images (AREA)
• Editing Of Facsimile Originals (AREA)

Priority Applications (3)

Applications Claiming Priority (186)

Publications (2)

Family

ID=37684666

Family Applications (1)

Country Status (4)

Cited By (10)

Families Citing this family (42)

Family Cites Families (19)

• 2005
  • 2005-04-19 WO PCT/US2005/013297 patent/WO2005101192A2/en active Application Filing
  • 2005-04-19 JP JP2007509565A patent/JP5102614B2/ja not_active Expired - Fee Related
  • 2005-04-19 EP EP05742065A patent/EP1759278A4/de not_active Ceased
• 2006
  • 2006-11-17 KR KR1020067024177A patent/KR101174536B1/ko active IP Right Grant
• 2011
  • 2011-11-14 JP JP2011248290A patent/JP5496987B2/ja active Active

Non-Patent Citations (1)

Also Published As

Similar Documents

Legal Events