Abstract: A method of post-processing character data from an optical character recognition (OCR) engine and apparatus to perform the method. This exemplary method includes segmenting the character data into a set of initial words. The set of initial words is word level processed to determine at least one candidate word corresponding to each initial word. The set of initial words is segmented into a set of sentences. Each sentence in the set of sentences includes a plurality of initial words and candidate words corresponding to the initial words. A sentence is selected from the set of sentences. The selected sentence is word disambiguity processed to determine a plurality of final words. A final word is selected from the at least one candidate word corresponding to a matching initial word. The plurality of final words is then assembled as post-processed OCR data.

Abstract: Document type comparison and classification using layout classification is accomplished by first segmenting a document page into blocks of text and white space. A grid of rows and columns, forming bins, is created on the page to intersect the blocks. Layout information is identified using a unique fixed length interval vector, to represent each row on the segmented document. By computing the Manhattan distance between interval vectors of all rows of two document pages and performing a warping function to determine the row to row correspondence, two documents may be compared by their layout. Furthermore, interval vectors may be grouped into N clusters with a cluster center, defined as the median of the interval vectors of the cluster, replacing each interval vector in its cluster. Using Hidden Markov Models, documents can be compared to document type models comprising rows represented by cluster centers and identified as belonging to one or more document types.

Abstract: An image processing method or device invented to reduce the ratio of erroneously recognized non-character elements in optical character recognition (OCR) regarding a color document that includes character images and other types of images, wherein the extracted character image data is checked to determine whether a color change exists in each character image, and wherein if no color change exists, the character image data is converted into character code data, but where a color change does exist, the character image data is not converted into character code data.

Abstract: The present invention is a computer-implemented method for calculating word accuracy. Word grouping accuracy values (260) are calculated (212) by using the character accuracy values (250) calculated by an OCR program present in a computer system. The present invention preferably uses these character accuracy values (250) to create a word grouping accuracy value (260). Various methods are employed to calculate the word accuracy (260), including binarizing the character accuracy values (250), modified averaging of the character accuracy values (250), and creating fuzzy visual displays of word grouping accuracy values (260). The calculated word grouping accuracy values (260) are then adjusted based upon known OCR strengths and weaknesses, and based upon comparisons to stored word lists and the application of language rules. In a system with multiple character recognition techniques, the system can compare the accuracy values (260) of different versions of the word groupings to find the most accurate version.

Abstract: Systems, apparatuses, and methods to relate images of words to a list of words are provided. A trellis based word decoder analyses a set of OCR characters and probabilities using a forward pass across a forward trellis and a reverse pass across a reverse trellis. Multiple paths may result, however, the most likely path from the trellises has the highest probability with valid links. A valid link is determined from the trellis by some dictionary word traversing the link. The most likely path is compared with a list of words to find the word closest to the most.

Abstract: The present invention provides a method and system for confirming uncertainly recognized words as reported by an Optical Character Recognition process by using spelling alternatives as search arguments for an Internet search engine. The measured number of hits for each spelling alternative is used to provide a confirmation measure for the most probable spelling alternative. Whenever the confirmation measure is inconclusive, a plurality of search strategies are used to reach a measured result comprising zero hits except for one spelling alternative that is used as the correct alternative.

Abstract: A method and system is provided for identifying a page layout of an image that includes textual regions. The textual regions are to undergo optical character recognition (OCR). The system includes an input component that receives an input image that includes words around which bounding boxes have been formed and a text identifying component that groups the words into a plurality of text regions. A reading line component groups words within each of the text regions into reading lines. A text region sorting component that sorts the text regions in accordance with their reading order.

Abstract: Characters are recognized by a conventional OCR apparatus and converted into outline font form. The system includes a recognition device for optically reading printed characters and recognizing those to obtain information on the recognized characters consisting of text code information and character layout information, an outline font table for retaining outline font data of characters, a character box enlarging function for enlarging enclosing rectangles of the recognized characters obtained by the recognition device by a ratio of an outline font character box to a black pixel component to be drawn in the character box while referring to the outline font table, and modifying the information on the recognized characters by using the enlarged enclosing rectangles as new character boxes of the outline font.

Abstract: Embodiments of the present invention disclose a dictionary lookup method and an electronic device that implements the dictionary lookup method. The dictionary lookup method allows a user to quickly obtain meanings and translations of words from electronic dictionaries while reading a text on a display screen of the electronic device, wherein reading text is utilized by performing an optical character recognition comprising of determining a set of base forms of each inflected recognized word. Advantageously, in one embodiment the meanings (e.g., the base forms) and translations may be displayed in a balloon, in a pop-up window, as subscript, as superscript, or in any other suitable manner when the user touches a word on the display screen, in one embodiment.

Abstract: A system and process of nearsighted (myopia) camera object detection involves detecting the objects through edge detection and outlining or thickening them with a heavy border. Thickening may include making the object bold in the case of text characters. The bold characters are then much more apparent and heavier weighted than the background. Thresholding operations are then applied (usually multiple times) to the grayscale image to remove all but the darkest foreground objects in the background resulting in a nearsighted (myopic) image. Additional processes may be applied to the nearsighted image, such as morphological closing, contour tracing and bounding of the objects or characters. The bound objects or characters can then be averaged to provide repositioning feedback for the camera user. Processed images can then be captured and subjected to OCR to extract relevant information from the image.

Abstract: A method and system of recognizing handwritten words in scanned documents, wherein by processing a document containing handwriting, features for word localization are extracted from handwritten words contained in said document through basis points taken from a single curve of text lines. The method is independent of page orientation, and does not assume that the individual lines of handwritten text are parallel, and the method does not require that word regions be aligned with text line orientation wherein intra-word statistics are derived from sample pages rather than using a fixed threshold. The method has applications in digital libraries, handwriting tokenization, document management and OCR systems.

Abstract: A method for segmentation of characters in text that segments text into lines, words and slices and determines at least one of fixed pitch and proportional pitch prior to segmentation. The method computes histograms of the lines and defines widths of lobes of the histograms of the lines as the character pitches. In addition, the method further analyzes the character pitches; segments lines into words; computes histograms of the words and aggregating the histograms of the words at predetermined points. Moreover, the method segments the words; slicing them words into an upper slice and lower slice and further segments the upper slice and the lower slice. The results are then combined to provide for both coarse and fine segmentation that enhance the performance of character OCR for documents scanned as at least one of gray-scale images and color images.

Abstract: A method and system for preprocessing an image for Optical Character Recognition (OCR), wherein the image includes a plurality of columns is disclosed. Each column includes one or more of Arabic text and non-text items. The method includes determining a plurality of components associated with one or more of the Arabic text and the non-text items, wherein a component includes a set of connected pixels. On determining the plurality of components, a line height and a column spacing is determined for the plurality of components. The plurality of components are then associated with a column of the plurality of columns based on the line height and the column spacing. Subsequently, a set of characteristic parameters are calculated for each column and the plurality of components of each column are merged based on the set of characteristic parameters to form sub-words and words.

Abstract: A method for identifying words in a textual image undergoing optical character recognition includes receiving a bitmap of an input image which includes textual lines that have been segmented by a plurality of chop lines. The chop lines are each associated with a confidence level reflecting a degree to which the respective chop line properly segments the textual line into individual characters. One or more words are identified in one of the textual lines based at least in part on the textual lines and a first subset of the plurality of chop lines which have a chop line confidence level above a first threshold value. If the first word is not associated with a sufficiently high word confidence level, at least a second word in the textual line is identified based at least in part on a second subset of the plurality of chop lines which have a confidence level above a second threshold value lower than the first threshold value.

Abstract: A method for identifying words in a textual image undergoing optical character recognition includes receiving a bitmap of an input image which includes textual lines that have been segmented by a plurality of chop lines. The chop lines are each associated with a confidence level reflecting a degree to which the respective chop line properly segments the textual line into individual characters. One or more words are identified in one of the textual lines based at least in part on the textual lines and a first subset of the plurality of chop lines which have a chop line confidence level above a first threshold value. If the first word is not associated with a sufficiently high word confidence level, at least a second word in the textual line is identified based at least in part on a second subset of the plurality of chop lines which have a confidence level above a second threshold value lower than the first threshold value.

Abstract: An image document processing device extracts a character sequence image having M number of characters in an image document, divides the image into individual character images, extracts features of the individual character images, and based on the features, selects N (N is an integer more than 1) character images in the order of degree of matching from a font-feature dictionary for storing features of all character images according to fonts, and generates an M×N index matrix for the extracted character sequence. In searching, the device searches an index-information storage section with respect to each search character included in a search keyword in an input search expression, and extracts an image document including an index matrix including the search keyword. This provides an image document processing device and an image document processing method each allowing indexing not requiring user's operation and each allowing highly precise searching without OCR recognition.

Abstract: Systems and methods for character-by-character alignment of two character sequences (such as OCR output from a scanned document and an electronic version of the same document) using a Hidden Markov Model (HMM) in a hierarchical fashion are disclosed. The method may include aligning two character sequences utilizing multiple hierarchical levels. For each hierarchical level above a final hierarchical level, the aligning may include parsing character subsequences from the two character sequences, performing an alignment of the character subsequences, and designating aligned character subsequences as the anchors, the parsing and performing the alignment being between the anchors generated from an immediately previous hierarchical level if the current hierarchical level is below the first hierarchical level. For the final hierarchical level, the aligning includes performing a character-by-character alignment of characters between anchors generated from the immediately previous hierarchical level.

Abstract: Font-independent spotting of user-defined keywords in a scanned image. Word identification is based on features of the entire word without the need for segmentation or OCR, and without the need to recognize non-keywords. Font-independent character models are created using hidden Markov models (HMMs) and arbitrary keyword models are built from the character HMM components. Word or text line bounding boxes are extracted from the image, a set of features based on the word shape, (and preferably also the word internal structure) within each bounding box is extracted, this set of features is applied to a network that includes one or more keyword HMMs, and a determination is made. The identification of word bounding boxes for potential keywords includes the steps of reducing the image (say by 2.times.) and subjecting the reduced image to vertical and horizontal morphological closing operations.

Abstract: An optical character recognition process characterizes text lines in a textual image by their base-line, mean-line and x-height. The base-line for at least one text line in the image is determined by finding a parametric curve that maximizes a first fitness function that depends on the values of pixels through which the parametric curve passes and pixels below the parametric curve. The base-line corresponds to the parametric curve for which the first fitness function is maximized. The first fitness function is designed so that it increases with increasing lightless or brightness of pixels immediately below the parametric curve while also increasing with decreasing lightness of pixels through which the parametric curve passes. The mean-line is determined by incrementally shifting the base-line upward by predetermined amounts (e.g., a single pixel) until a second fitness function for the shifted base-line is maximized. The second fitness function is essentially the inverse of the first fitness function.