TL;DR: The current state of a system that recognizes printed text of various fonts and sizes for the Roman alphabet is described, which combines several techniques in order to improve the overall recognition rate.
Abstract: We describe the current state of a system that recognizes printed text of various fonts and sizes for the Roman alphabet. The system combines several techniques in order to improve the overall recognition rate. Thinning and shape extraction are performed directly on a graph of the run-length encoding of a binary image. The resulting strokes and other shapes are mapped, using a shape-clustering approach, into binary features which are then fed into a statistical Bayesian classifier. Large-scale trials have shown better than 97 percent top choice correct performance on mixtures of six dissimilar fonts, and over 99 percent on most single fonts, over a range of point sizes. Certain remaining confusion classes are disambiguated through contour analysis, and characters suspected of being merged are broken and reclassified. Finally, layout and linguistic context are applied. The results are illustrated by sample pages.
TL;DR: The prediction of faster processing with a regular font, as compared with a mixed font, was examined in three experiments requiring the recognition of four-letter strings and showed that the disadvantage for a mixture of fonts is related to how much the representational system must be adjusted to handle the different fonts.
Abstract: The representation of visual information about letters is proposed to be highly systematic, involving not only abstract information that is invariant across type faces (or fonts), but also a number of parameters whose values are determined by the current font Th e system exploits regularities that are characteristic of letters and fonts by becoming tune d to the details of the font Thi s should result in efficient letter perception when the stimuli are regular (when all of the letters are of a consistent font), but not when the stimuli are irregular (when the letters are from a variety of fonts) Th e prediction of faster processing with a regular font, as compared with a mixed font, was examined in three experiments requiring the recognition of four-letter strings Experiment 1 confirmed the prediction, and Experiment 2 replicated the effect with the number of "features" equated across conditions. Experiment 3 showed that the disadvantage for a mixture of fonts is related to how muc h the representational system must be adjusted to handle the different fonts A central issue in cognitive an d perceptual psychology is how familiar objects are represented and perceived. This article concerns the representation and perception o f letters, which are representative of more complex objects in that they vary considerably in appearance from instance to instance: The actual form of a letter depends on the type face, or font. Therefore, it is necessary for models of letter perception to specify how the perceptual system maps letters of differen t fonts onto the appropriate abstract letter codes. The purpose of the present research was twofold, first, to begin developing a new kind of model of letter perception, one that uses the idea of a structural network, and, second, to test this model against a class of simpler models by examining how perceptual representations might become systematically tuned for a particular font. A useful general approach to the recognition problem is to
TL;DR: This paper introduces a syntactic omni-font character recognition system that features scale-invariance and user-definable sensitivity to tilt orientation, and a structural pattern-matching approach is employed.
Abstract: This paper introduces a syntactic omni-font character recognition system. The "omni-font" attribute reflects the wide range of fonts that fall within the class of characters that can be recognized. This includes hand-printed characters as well. A structural pattern-matching approach is employed. Essentially, a set of loosely constrained rules specify pattern components and their interrelationships. The robustness of the system is derived from the orthogonal set of pattern descriptors, location functions, and the manner in which they are combined to exploit the topological structure of characters. By virtue of the new pattern description language, PDL, developed in this paper, the user may easily write rules to define new patterns for the system to recognize. The system also features scale-invariance and user-definable sensitivity to tilt orientation.
TL;DR: Transition-encoded font image information is a form of information flagging all pixel displacement locations upon a scan line of a raster scan image generator, nominally a laser printer as mentioned in this paper.
Abstract: Transition-encoded font image information is a form of information flagging all pixel displacement locations upon a scan line of a raster scan image generator, nominally a laser printer, whereat black to white, and white to black, transitions occur during generation of the one scan line of a font image. Transition-encoded information for plural, superimposed, font images may be combined prior to generation of a synthesis image. In this combination of transition-encoded information for generating superimposed font images the flags representing transitions must not be overwritten, there being a limit that a single scan line pixel position must either transit black to white, transit white to black, or maintain the state of the previous pixel. The combining of transition encoded information encodes transitions for these one or ones of addresses (pixels) wherein two flags would otherwise overlap so that the total transitions within the combined font information are preserved. This is accomplished by ripple shifting the position(s) of the transition(s) flag(s) within the second font image information which have identical positional correspondence with transition(s) flag(s) within the first font image information so that the flag(s) are slightly relocated within the combined font image information.
TL;DR: In this article, the authors propose a temporary memory device for storing a dot-font form font called a font cash, which can be used to store a font with high quality at high speed.
Abstract: PURPOSE: To develop a font with high quality at high speed, by providing a high-speed temporary memory device storing a dot-font form font called a font cash. CONSTITUTION: When a communication controller 101 receives a character output command, a main controller 102 analizes the command and requests a dot form font of a predetermined font to a font data controller 106. The font data controller 106 checks whether or not the dot form font is stored in a font cash 117. If it is stored, the dot font is returned to the source of the request as the result. If not stored, the font data controller 106 determines a font for use among font data 108W111. A font converter 112 converts the designated font to a dot font form, stores it in an area of the font cash 117, and returns the dot form font to the source of the request as the result. The main controller 102 transfers the dot form font to a specified area of an image recorder 103. A printing controller 104 actually performs printing with use of the data stored in the image recorder 103. COPYRIGHT: (C)1989,JPO&Japio
TL;DR: A character font converter comprises font memory means for storing an original character font, periphery detection means for detecting peripheral black dots of the original character fonts, location calculation means for converting according to a specified relational expression the location of the peripherals detected by the periphery detection, convert pattern storage means for encoding the new location converted by the location calculation as black dots, and paint-over means for painting over the inside of the converted black dots as mentioned in this paper.
Abstract: A character font converter comprises font memory means for storing an original character font, periphery detection means for detecting peripheral black dots of the original character font, location calculation means for converting according to a specified relational expression the location of the peripheral black dots detected by the periphery detection means, convert pattern storage means for storing the new location converted by the location calculation means as black dots, and paint-over means for painting over the inside of the converted black dots.
TL;DR: In this paper, an output device is provided with a result deciding part 7 for adding such information as a size and a font of a character outputted from both a size deciding part 3 for deciding a character by obtaining the information of a size of the character from a character segmenting part 2, and an font deciding part 5 for determining a font (style of handwriting), and outputting its result.
Abstract: PURPOSE:To save information of a size and a style of handwriting of a character of an original document, and to execute a reproduction faithfully to some extent in case of reproducing a sentence from a result of recognition, by comparing a standard character size which has been designated in advance and a size of a character pattern which has been segmented actually, and adding its ratio to a result of recognition of its character. CONSTITUTION:This device is provided with a result deciding part 7 for adding such information as a size and a font of a character outputted from both a size deciding part 3 for deciding a size of a character by obtaining the information of a size of a character from a character segmenting part 2, and a font deciding part 5 for deciding a font (style of handwriting), and outputting its result, and an output device 8. For instance, by a font deciding method (a collation with a dictionary, of each font is executed, and a character font of the smallest distance is decided), a style of handwriting of its character is decided in the font deciding part 5, and information of four kinds consisting of a Ming- style, a Gothic style, a knurl style, and a textbook style is added.
TL;DR: In this article, the exclusive OR with the logical sum of a regular font pattern and four shift patterns obtained by shifting said font pattern to the up-anddown direction and the left-and right direction by one dot was used to make it possible to perform the printing of an outline font without applying excessive load to a host.
Abstract: PURPOSE:To make it possible to perform the printing of an outline font without applying excessive load to a host, by taking the exclusive OR with the logical sum of a regular font pattern and four shift patterns obtained by shifting said font pattern to the up-and-down direction and the left-and-right direction by one dot. CONSTITUTION:In performing printing by forming an outline font on the basis of a regular font pattern (a) constituted of a two-dimensional matrix, at first, this regular font pattern is shifted upwardly, downwardly, leftwardly and rightwardly by one shift to form an upper shift pattern, a lower shift pattern, a left shift pattern and a right shift pattern and the logical sum of these four shift patterns is taken to form a logical sum pattern (b). Next, when the exclusive OR with this logical sum pattern (b) and the regular font pattern (a) is taken, a pattern shown by a drawing (c) is obtained. This pattern is set as an outline font pattern to perform printing to enable the printing of an outline font.
TL;DR: A cooperative effort now under way among graphic arts professionals, representatives of industry, and national and international standards-making organizations to develop a public font standard along with the associated tools and services is described.
Abstract: Distribution of documents in electronic form, often called soft copy, permits more rapid sharing of information and the acceleration of new knowledge; however, barriers to soft-copy document distribution have arisen out of the diversity of the tools currently available in publishing technology. The authors address solving these problems by means of the development of a public font standard, and novel tools and services. They describe a cooperative effort now under way among graphic arts professionals, representatives of industry, and national and international standards-making organizations to develop such a standard along with the associated tools and services.
TL;DR: This paper treats problems and proposes a solution for handling of fonts for the XPS-700 Xerox Publishing System, which uses a variety of different typefaces.
TL;DR: In this article, an additional registration of the character code (reject code) of an input character was performed by providing a storage means for a font code, a displaying means for storage content, and a deciding means for the input character font.
Abstract: PURPOSE:To prevent the designation error of a font, and to exactly and easily attain an additional registration in the additional registration of a character by providing a storage means for a font code, a displaying means for a storage content, and a deciding means for an input character font, etc CONSTITUTION:In the additional registration of the character code (reject code) of an input character, first of all, an additional registration processing program 24b on a main memory 24 is executed with a CPU14 Next, a character code file part 20 is searched, and the presence/absence of a reject character is investigated, and when it is present, the pattern information of the first reject character is read out from a reject character buffer 24a, then being displayed on a CRT26 Then, by indicating an automatic process, the font code corresponding to a prescribed number of input characters just prior at the reject character in process is read out from a font code file part 22, and is converted to a character code, thence displayed on the CRT26 In this way, an operator can easily designate the font of the reject character, and the designation error of the font can be prevented, and an exact and easy additional registration can be performed
TL;DR: In this article, a font to be displayed is temporarily transferred from a font memory 3' to the image memory 4. The contents of the memory 5 are displayed on a bit map display 8.
Abstract: PURPOSE:To display characters having optional size at real time by temporarily transferring a font to be displayed to an image memory and executing the editing processing of the font in the image memory. CONSTITUTION:A font to be displayed is temporarily transferred from a font memory 3' to the image memory 4. An image processor 7 expands/contracts the font at an optional magnification in accordance with the contents of a character string buffer 21 or changes its font on the basis of logical operation and then transfers the processed font to a display memory 5 on the basis of a character interval and a line interval specified by the buffer 21. The contents of the memory 5 are displayed on a bit map display 8.
TL;DR: In this paper, an external character font pattern received from an external device 30 is not intact stored into external character fonts memory 17. But when the external character pattern is different from the dot resolving power of a printer 40, it is subjected to density conversion in density converter 19 and then stored into internal font memory 17, therefore, when printing data is thereafter inputted from the external devices 30, the printer 40 reads out an externalcharacter font pattern of the dot resolution power suitable to itself and edits image data as it is.
Abstract: PURPOSE: To shorten the printing time by subjecting an original external character font pattern to density conversion into a conversion external character font pattern before an external character font pattern is stored into memory. CONSTITUTION: An external character font pattern received from an external device 30 is not intact stored into external character font memory 17. When the external character font pattern is different from the dot resolving power of a printer 40, it is subjected to density conversion in density converter 19 and then stored into external character font memory 17. Therefore, when printing data is thereafter inputted from the external device 30, the printer 40 reads out an external character font pattern of the dot resolving power suitable to itself and edits image data as it is. That is, conversion actions are not necessary thereafter, resulting in reduction of the printing time. COPYRIGHT: (C)1989,JPO&Japio
TL;DR: In this paper, the authors proposed to eliminate the need to have two fonts for normal upright and 90 deg-conversion of the same letters without lowering the printing speeds even with slow-speed processors by forming a 90 deg.-conversion font during the power-on time of a print controller.
Abstract: PURPOSE:To eliminate the needs to have two fonts for normal upright and 90 deg.-conversion of the same letters without lowering the printing speeds even with slow-speed processors by forming 90 deg.-conversion font during the power-on time of a print controller. CONSTITUTION:When a print controller is powered on, a processor 2 reads upright font possessed in advance out from a font memory 5a, converts into a 90 deg.-conversion font and then stores the result in a font memory 5b. From a host CPU information, fonts for upright and 90 deg.-conversion pringing are read out of the font memories 5a and 5b, and stored in a full-bit memory 3. Printing information for one page is converted and when the storage is ended, the information is sent by a parallel-series conversion circuit 4 to the printer for printing. Since 90 deg.-conversion font is formed during the power-on period, the slow processing speed of the processor 2 is allowed and the cost can thus be reduced. The needs to have two fonts for upright and 90 deg.-conversion can be eliminated.
TL;DR: In this paper, a Chinese character code is read from a buffer memory and stored in a CCS memory, and an address at this time is designated from a control code discriminating circuit for instructing the start and completion of a printing.
Abstract: PURPOSE:To print satisfactorily a character through a coded transmission such as a character broadcasting or a CAPTAIN by printing and displaying a character font by the use of the separate character font from a CRT such as a word processor. CONSTITUTION:When a Chinese character code is read from a buffer memory 5, the Chinese character code is supplied to a Chinese character code memory 12 and stored. An address at this time is designated from a control code discriminating circuit 6. When a switch 11 for instructing the start and the completion of a printing is operated, a reading address of the memory 12 changes correspondingly to a printing speed of a printer of the word processor 2 and the printer prints the Chinese character displayed on the CRT by the Chinese character of another font.
TL;DR: In this article, the authors propose to enable patterns in various display forms to be used for displaying and printing, by providing displaying patterns and printing patterns in different dot densities, and providing a pattern data switching means.
Abstract: PURPOSE:To enable patterns in various display forms to be used for displaying and printing, by providing displaying patterns and printing patterns in different dot densities, and providing a pattern data switching means CONSTITUTION:A displaying font for one character stored in a font pack 1a is constituted of, for example, 12X12 dots, whereas a printing font for one character stored in the font pack 1a is constituted of, for example, 24X24 dots After data are inputted through an input device 3, a printing font corresponding to a displaying font displayed on a display device 4 is read from a printing pattern 1c to be sequentially outputted to a printer 5 only when it is discriminated that the operating mode is a printing mode Accordingly, by preparing memory packs in various display forms, patterns in various forms can be outputted to the display device and the printer
TL;DR: In this article, the bottom of a font matrix is used as a reference to make spaces between every character uniform and to enable the printing of legible slant characters by slanting a font pattern.
Abstract: PURPOSE:To make spaces between every character uniform and to enable printing of legible slant characters, by slanting a font pattern with the bottom of a font matrix as a reference. CONSTITUTION:Font data are read out from a character font ROM and 7 bit dot data are taken out, and 2 bit null data are added to the above to make 9 bit data and these data are stored in the font data. These data are slanted by logical operation. During the above operation, the bottome dots of the font matrix are used as the basis of the slanting, and the whole data are slanted and stored in a print buffer. 9 bits at each time are taken out from the print buffer, and printing is carried out to a printing medium with a printing head. Thus, printing of legible slant characters with equal spacing between the characters become feasible.
TL;DR: In this paper, a character recognition section uses an input stroke number from a stroke discrimination section 3 to select character pattern information stored in a character discrioinary memory 7, the input stroke information and the character patterns stored in the character dictionary memory 7 are compared, the character is specified and the recognition processing of the input character is applied depending on the discriminated font at a font discrimination section 5.
Abstract: PURPOSE:To reduce the fatigue of the user and to obtain a easy-to-use character recognition device by applying specific non-recognition display, multiple recognition display and input font unspecified display in addition to recognized character display. CONSTITUTION:A character recognition section 6 uses an input stroke number from a stroke discrimination section 3 to select character pattern information stored in a character discrioinary memory 7, the input stroke information and the character pattern information stored in the character dictionary memory 7 are compared, the inputted character is specified and the recognition processing of the input character is applied depending on the discriminated font at a font discrimination section 5. When the possibility of the recognition is low, an unrecognized code, when plural character candidates exist, a multiple recognition code, when there is a property hard of specification, a font unspecified code and when a character is recognized as one character, a candidate character code is generated and the result is fed to a picture converter 8.
TL;DR: In this article, a font selection table memory is used to select a font having a property same or nearest to the front of the font front to be used at the time of a printing order.
Abstract: PURPOSE:To attain versatility from the aspect of the arrangement of the output format of a document, in changing over a font, by making it possible to select whether the feed quantity of the font thereafter is made inherent to the font after changing over or unchangeable. CONSTITUTION:CPU4 controls the properties of the various fonts of a character pattern generator 6 on a font selection table memory 12 at first. Next, when a font to be used in indicated at the time of a printing order from a host calculator 1, CPU4 selects a font having a property same or nearest to that of the indicated front. In case of printing, if at the time of the pitch mode according to the property of the selected font in accordance with the pitch mode selected by a mode selection means (switch 17), a character pitch thereafter is controlled on the basis of the feed quantity HMI inherent to the selected font and, at the time of a pitch mode using a predetermined fixed pitch, the character pitch thereafter is controlled on the basis of fixed feed quantity HMI. By this method, the response to various needs of application soft is enabled.
TL;DR: The story of how the CHIO project evolved and how it learned by doing is told, which includes traditional MIS techniques and shows how these are applied in a production system.
Abstract: Karsten Manufacturing Corporation issues a “Certificate of Hole in One” (CHIO) to golfers who ace with PING (trademark) equipment. Formerly done manually, the certificate is now plotted automatically in less than two minutes, unattended, with two-color English Gothic and simplex italic Roman fonts, and a digitized signature, on gilded, preprinted, parchment forms. The plot file is produced from international input data received orally by telephone, or on tape from the “Golf Digest” clearing house. The APL2 application was implemented as an extension of the IBM GIAM graphics language and employs several new commands for loading and manipulating graphic characters from the Hershey font library. The paper explains various input panels and graphic displays. Technical obstacles including acceptable interactive and batch running-time-per-certificate are described, along with their solution. The method includes traditional MIS techniques and shows how these are applied in a production system.This is the story of how the CHIO project evolved and how we learned by doing.