• 专利附录
  • 专利说明
  • 权利要求
  • 类似技术
  • 同族专利
  • 引用文献
  • 法律状态
  • 优先权
    • 专利摘要:
    A line segmentation method begins by determining a first starting point coordinate and generating a list of potential character widths dependent on a first maximum character width stored in a database and features of the part of the line of text corresponding to the maximum character width. The method determines a second portion of the text line corresponding to the first start point coordinate and the first width on the list of potential character widths. A classification method is applied on the second part giving an error probability for the first width and a candidate character. The error probability is compared to a first threshold determined by a compromise between speed and accuracy, and if the error probability corresponding to the first width is less than the threshold value, the candidate character is selected as the signifying character that a segment is known.
    公开号:BE1025503B1
    申请号:E2015/5242
    申请日:2015-04-15
    公开日:2019-03-27
    发明作者:Frédéric Collet；Jordi Hautot；Michel Dauw；Meulenaere Pierre De；Olivier Dupont；Günter Hensges
    申请人:I.R.I.S.；
    IPC主号:
    专利说明:

    Optical character recognition (OCR) systems are known. These systems automatically convert a paper document into a searchable text document. OCR systems are typically composed of three main stages: line segmentation, feature extraction and character classification. But, as illustrated in Figure 1, feature extraction is often presented as part of the character classification. In this way, starting from an image of a character string, known optical character recognition systems first apply line segmentation to become images of individual characters and then a step of classifying characters is executed to identify characters. Although character classification techniques have become extremely robust in recent years, line segmentation still remains a critical step in OCR, especially in the case of Asian texts.
    There are different approaches to line segmentation (which is also often called character segmentation). The image representing a line of text is broken down into individual sub-images which constitute the images of the characters. Different methods can be used to segment a line. A known method of line segmentation is the detection of inter-character breaks or word breaks (adapted to Latin characters) as a way of isolating individual characters. This is described for example in WO / 2011/128777 and WO / 2011/126755.
    BE2015 / 5242
    Another method of segmenting lines, described for example in WO / 2011/142977, uses cutting lines which are then processed to identify the lines which separate characters. Still other methods, such as for example in EP0138445B1, assume a constant pitch between characters.
    The line segmentation methods described above are known as dissection methods. This type of process is less effective for text composed of Asian text and Asian text combined with Latin text because in this type of text there is often no break or unclear between characters and that the characters Asians do not consist of a single connected component but more often than not of several connected components (for example the radicals for Chinese characters).
    Another type of line segmentation process is based on the recognition of components in the image which correspond to classes in a particular alphabet. However, such methods require long calculation times.
    A third type of segmentation technique uses a combination of the first two and is known as the over-segmentation process. The image is over-segmented with different dissection methods as illustrated in FIG. 2. Several plausible segmentation solutions are analyzed by the same character classification methods or by different methods and the best segmentation solution is then chosen. When segmentation becomes difficult, as is the case for example with Asian characters, many possible segmentation solutions are evaluated, which leads to extremely long computation times for analyzing the input image of the string.
    Disclosure of invention
    BE2015 / 5242
    It is an object of the present invention to provide a method for segmenting characters in a character string image which allows fast and exact segmentation of a line.
    These objects are achieved according to the invention with a method for segmenting characters in a character string image showing the technical characteristics of the first independent claim. The method for segmenting characters in a character string image according to the invention comprises the steps consisting in:
    a) determining a first starting point coordinate of a pixel contrasting with the background,
    b) generating a list of potential character widths depending on a maximum character width and characteristics of the part of the character string image corresponding to the maximum character width,
    c) determining a second part of the character string image corresponding to the first starting point coordinate and to the first width,
    d) applying the classification process to the second part of the character string image giving a probability of error for the first width and a candidate character,
    e) compare the probability of error to a first threshold determined by a compromise between speed and accuracy; and
    f) select the candidate character as the character corresponding to the first width if the probability of error corresponding to the first width is less than the threshold value.
    An advantage of this method is that line segmentation and character classification are a character-by-character process. This has a huge advantage in reduced computation time because the number of steps required to execute the
    BE2015 / 5242 line segmentation and character classification of a character string image is seriously reduced. The result is an increase in the speed and accuracy of the process.
    In other embodiments according to the present invention, the method further comprises the step of comparing the probability of error with a second threshold value greater than the first threshold value; and wherein the step of comparing the probability of error to the first threshold value is performed only if the probability of error is less than the second threshold value.
    The second threshold value has the advantage that it allows rapid screening of candidates who have no chance of giving a positive result.
    In another embodiment according to the present invention, the method further comprises the step of calculating the starting point for the next character if the probability of error corresponding to the first width is less than the second threshold value, and keep the calculated starting point of the next character in memory.
    In another embodiment according to the present invention, the method further comprises the step of updating the statistical values of characters contained in a database if the probability of error corresponding to the first width is less than the first value threshold.
    This database contains information on the maximum and average sizes of characters in the text and reference characters. These values are used when estimating character widths in generating the list of potential character widths to improve the speed and accuracy of the process.
    In another embodiment of the present invention, the list of potential character widths is sorted from most likely to least likely, the most likely width being that which is the widest width containing a maximum number of connected components which
    BE2015 / 5242 are not greater than an estimated maximum width for a character stored in the database.
    In another embodiment according to the present invention, the two least probable widths from the list of potential character widths are an average overall width and half the average overall width, the average overall width being the height of the image. of character string for a first character in the character string image and the average global width being calculated on the basis of a previous average global width and an average character width stored in the database for a subsequent character in the string image.
    The advantage of this is that the average overall width will identify Asian characters, while half of the average overall width will identify Latin characters because the size of Asian characters is approximately twice the size of Latin characters. and because the line segmentation method can in this way be applied to Latin characters, Asian characters and a combination thereof.
    In another embodiment according to the present invention, if the probability of error corresponding to the previous width of the list of potential character widths is greater than the second threshold value, the method further comprises the steps consisting in:
    a) determining a second part of the character string image corresponding to the starting point coordinate and to the next width of the list;
    b) applying the classification method to the second part of the character string image giving a probability of error for this width and a candidate character;
    c) compare the probability of error with the threshold value stored in the database;
    BE2015 / 5242
    d) repeat steps a), b) and c) until the probability of error corresponding to this width is less than the threshold value or until all the widths contained in the candidate width list potential have been dealt with;
    e) select the candidate character as the character corresponding to the width if the probability of error corresponding to the first width is less than the first threshold value.
    Line segmentation and character classification are combined and performed one after the other until no solution has been found and until a solution is found. This reduces the number of steps required to perform such a process and also improves the accuracy of the process.
    In another embodiment according to the present invention, the character string image is a vertical character string image and all the widths are heights.
    Asian characters can be written in rows but also in columns. The process is certainly not limited to lines and can easily be adapted to columns by just changing the width of the characters in height and vice versa.
    In another embodiment, the method further includes the step of updating a statistical character database with the average overall width value upon a successful iteration.
    In another embodiment according to the present invention, the step consisting in generating a list of potential character widths is based on data retrieved from a database which contains reference characters for a given point size, the width of the largest reference characters, the average width of the reference characters and the size of the average space between reference characters.
    BE2015 / 5242
    In another embodiment of the present invention, the database further contains estimates of statistical values of the characters, the database being updated on each successful iteration.
    In another embodiment of the present invention, the maximum character width is a maximum character width for Asian characters.
    In another embodiment of the present invention, a computer program product includes a medium usable by a computer in which control logic is stored to cause a computer device to segment a character string image into an input image , the control logic comprising:
    a) first program code means readable by command to determine a first starting point coordinate of a pixel contrasting with the background,
    b) second command-readable program code means for generating a list of potential character widths dependent on a maximum character width and characteristics of the portion of the character string image corresponding to the maximum character width ,
    c) third command-readable program code means for determining a second part of the character string image corresponding to the first starting point coordinate and to the first width on the list of potential character widths,
    d) fourth means of program code readable by command to apply a classification method on the second part of the image of character string giving a probability of error for the first width and a candidate character,
    BE2015 / 5242
    e) fifth means of program code readable by command to compare the probability of error with a first threshold determined by a compromise between speed and accuracy; and
    f) sixth command-readable program code means for selecting the candidate character as the character corresponding to the first width if the probability of error corresponding to the first width is less than the threshold value.
    Brief description of the drawings
    The invention will be explained in more detail by means of the following description and the accompanying drawings.
    Figure 1 shows the different steps in an optical character recognition process according to the prior art.
    FIG. 2 illustrates a type of line segmentation in the prior art known as over-segmentation.
    FIG. 3 shows a method of segmenting lines according to an embodiment of the invention.
    FIG. 4 illustrates a method of segmenting lines with a statistical character database.
    Modes of implementing the invention
    The present invention will be described in relation to particular embodiments and with reference to certain drawings, but the invention is however not limited thereto, being limited only by the claims. The drawings described are only schematic and are not limiting. In the drawings, the size of some of the elements may be exaggerated and not drawn to scale for illustrative purposes. The dimensions and the relative dimensions do not necessarily correspond to the actual functional realizations (reductions to practice) of the invention.
    Furthermore, the terms first, second, third and the like in the description and in the claims are used to make the
    BE2015 / 5242 distinction between similar elements and not necessarily to describe a sequential or chronological order. The terms are interchangeable under the appropriate circumstances and the embodiments of the invention can function in sequences other than those described or illustrated in this document.
    In addition, the terms top, bottom, above, below and the like in the description and claims are used for descriptive purposes and not necessarily to describe relative positions. The terms thus used are interchangeable in the appropriate circumstances and the embodiments of the invention described in this document can function in other orientations than those described or illustrated in this document.
    Furthermore, the various embodiments, although qualified as preferred, should be interpreted as exemplary ways in which the invention can be implemented rather than as limiting the scope of the invention.
    The term comprising, used in the claims, should not be interpreted as being limited to the means or steps listed below; it does not exclude other elements or stages. It must be interpreted as specifying the presence of the elements, whole numbers, stages or components mentioned to which reference is made, but does not exclude the presence or addition of one or more other elements, whole numbers, stages or components or groups. of these. Therefore, the scope of the expression a device comprising A and B should not be limited to devices comprising only the components A and B, on the contrary, as regards the present invention, the only listed components of the device are A and B, and the claim should also be interpreted to include equivalents of these components.
    Referring to Figure 3, Figure 3 illustrates a flow diagram of an optical character recognition (OCR) method according to an embodiment of the present invention. The process input is a picture
    BE2015 / 5242 of character string 110. In a first step, a segmentation of lines is carried out on the image of character string 110. Preliminary information on the potential widths of the character analyzed is calculated. Preliminary information on the potential widths of the character allows a new sequence of steps which improves the speed of the OCR process. Although over-segmentation is still used, all the potential solutions (210, 220, 230) should not be systematically analyzed by the OCR method. Potential solutions are generated using a list of candidate character widths 310 and are sorted from most likely to least likely. The OCR method first analyzes the most probable solution 210. If a condition on the measurement error is satisfied 320, the character is classified 150, the other potential solutions are rejected and the next character is analyzed. If the condition on the measurement error is not satisfied 330, the next most likely solution is analyzed. This process is repeated iteratively until no character has been successfully classified or until all potential solutions have been assessed.
    The method, as described here, is applied to segment a line of text. However, the same process can be used to segment a column of text as is often the case for Asian text.
    As described above, a list of candidate character widths 310, ordered from the most likely to the least likely occurrences, is generated before the analysis of a character image. The generation of this list of candidate character widths will be described later in the request. The list contains □ 2 of the candidate widths, the first widths being widths for which no cut must be made in the character string image 110 to extract a character and the last two widths being widths for which a cut must be performed so
    BE2015 / 5242 to isolate and extract a character in the character string image 110.
    The starting points are the coordinates S which define the position of the new character image to be analyzed. A list of initial starting points is created at the start of the algorithm, where the first initial starting point in the list corresponds to the first black pixel on the left in the image. Other predefined starting points correspond to the end of the line or to the pixel on the far right. Other starting points are added to the starting point list during the OCR process. The process ensures that all starting points in the list are processed.
    A character image is entirely defined by a starting point coordinate and a width related to a list of connected components. The height of the line is the same for all characters. At the end of the OCR process, the character is classified.
    As soon as a potential solution is created, a character classification method 140 is applied to the potential solution to determine if a character can be classified for that potential solution. In one embodiment of the invention, the character classification method 140 is based on Gabor functions.
    A character classification method 140, according to an embodiment of the invention, requires two inputs:
    - the starting point coordinate of character W 3 ^. The starting point coordinate is the coordinate of the first pixel of a character at the bottom left of the character to be analyzed,
    - a candidate width taken from the list of candidate character widths for the character S
    The result is a probability of error which is used to calculate the character 1 caractère. The probability of error Wagest compared to two threshold parameters: a threshold having a low probability of error ^^ get a threshold having a high probability of error âr »The values of
    BE2015 / 5242 and Sqœ can be adjusted according to the speed in relation to the accuracy requirements. In a preferred embodiment of the invention, the values of §S = jet är ^ are set to œæ = 20% and §) □ == 99.9%. The threshold with a low probability of error ^^ defines the condition for having a character classified successfully.
    A method of line segmentation according to an embodiment of the invention uses a statistical character database 400 as illustrated in FIG. 4. The elements of the database will now be listed. A more detailed description of how each of the elements is used will follow later in the application. The database contains:
    - a library of reference sizes (height and width) for Asian and Latin characters, and for a selected point size, stored in the memory,
    - the maximum reference size for Asian and Latin characters for the selected point size, stored in the memory, respectively Shiir ^ and Earthquake
    - the average inter-character reference space, the same for Asian and Latin text, for the selected point size,
    - the maximum estimated width of the Asian and Latin characters in the text being analyzed: respectively and ifegtmo
    - the average inter-character space for Asian and Latin characters in the text being analyzed, §
    - the local estimate of the width of Asian and Latin characters W respectively and ^ a , which represents the width of the corresponding reference character, calculated only for characters which have been classified. It is a measure of the point size of the character calculated using the effective width and value of the character S
    BE2015 / 5242
    - the local estimate of the width of Asian and Latin characters 1¾, which represents the width of the corresponding reference character, calculated only for characters which have been classified. The value of is a running average of the local estimates previously measured Set is therefore a more accurate measure of the average point size of the character. This value is more reliable because it is more tolerant of incorrectly classified characters.
    Referring to Figure 4, Figure 4 shows a flow diagram of the line segmentation process according to an embodiment of the invention. The process is illustrated for a segmentation of character ë ^. The list 310 of all ^ | D 2 candidate character widths for the character ë ^ is generated and a first candidate character width is taken from the list of candidate character widths. These two values, and are the inputs 410 for the character classification process 140 in step 420. The result of step 420 is the probability of error
    Depending on the value of two options are possible. If the probability of error ^ g is less than the threshold on the error of having a high probability of error ^ 7 æ421, the character ëg is a potential solution. The character ë & corresponding to the first candidate width ^ j, is then kept in memory and the starting point of the following character is calculated and added to the list of starting points to be processed if necessary: l ^ + = 425. If, in in addition, the probability of error ^ gest also below the threshold on error to have a low probability of error ^ §æ422, the character S can be considered as successfully classified and the statistical database of characters 400 is updated as will be explained later in the description. The method can go to the next starting point S ^ + D 405 to determine the next character ë ^ + D without processing the other widths for the point of
    BE2015 / 5242 current start If the probability of error ^ gest greater than the threshold on the error of having a low probability of error 423, the character classification process is executed with the following candidate width ^^ 430, as described below.
    If however the probability of error ^ œ is higher than the threshold on the error to have a probability of error S) L æ424 high, the character corresponding to the candidate width 1, is not kept in memory and no new point is calculated.
    The character classification process is performed with the following candidate width 430. There are again two options depending on the value of ^ æg Si ^ gest less than ^ 7 ^ 431, the character SI is memorized with the width the starting point of the next character is calculated and added to the list of points start to process if necessary 405 and if ^ æ is also less than ^^ g432, the statistical character database is updated 400. If ^ g is however greater than ^^ get / or §7 ^ (435, 433), the character classification method is executed with the next candidate width i] 1, ^ D until all the widths in the list have been processed (^ = or until a character has been successfully classified ( ^ g <1§ ^).
    For □ 1, the same process is repeated but the width ^ + D is now such that a first cut is made for the width value 440. If no character has been classified with a low probability of error (443 or 445) ^ g <^^ g for § = 1 then the process is repeated for ^ lu 2 where ^ + D = Sfeaj iD , D 450 and different ways are again possible such as 451 with 452, 451 with 453 or 454.
    In order not to analyze all the solutions of the over-segmentation, the list of all the ^ | D 2 candidate widths {§} for the character S (¾) is generated as follows: the candidate widths are sorted from most likely to least probable and the number of candidate widths varies
    BE2015 / 5242 from one character to another, depending on the geometry of the potential character measured with the number of connected elements. It is assumed, on the basis of observations, that the width of Asian characters is common for most characters, except for a few characters which have a smaller width. According to an embodiment of the present invention, the most likely width corresponds to that which contains the largest set of connected components, not wider than the estimated width of the wider Asian character plus the estimated average space between characters (§ ).
    Characters may or may not touch. Non-touching characters have a higher probability of occurrence and should therefore be considered first.
    For the characters not touching, (no cut is necessary), the candidate width with the index S (^), calculated in pixels, is such that it is the W th largest width with a set S ( S ^ 0) of connected components smaller than the largest Asian character plus the estimated average space between characters (ë). The width Sa ^ connected components, the width has Sou less connected components and is such that%. D <S5
    The largest Asian character (% | gn, ^ a) and the estimated space between characters (ë) are evaluated in the statistical character database. There are S characters not touching possible.
    Clippings must be made if two adjacent characters touch, the characters will be cut in the most likely place which is calculated from the average overall width § ^ _ a of the character which can be found in the statistical database of characters updated to the previous iteration (SD 1) for the character The width with the index SD 1, □, corresponds to the sum of the average global width of Asian characters and the average space § The width with the index SD 2,% + □, corresponds to the sum of the width
    BE2015 / 5242 global average of Latin characters ^ -q / 2 and of average space § We assume that the width of Latin characters is half the width of Asian characters.
    To summarize, at each iteration, the list of candidate widths entered for the West character given by:
    width ^ me larger set of connected components so that B] = 1, ---, W; W ^ 0
    B = SH m H.
    B = SH / Ο π Ä where Shaw- S are values which come from the statistical character database updated each time a character has been classified (or ai ") ·
    The database contains a data structure which stores the character information extracted from the lines and a library of reference characters as well as statistical values on these characters. The only data structure is created at the start of the process, the structure is then empty. The data structure, stored in memory, is updated at each iteration and its structure is similar to a graph.
    All the parameters of the database are summarized in the following table:
    Individual characters Max Way Reference ^ (stored in the library, for each character of a point size selected) Text B^ 2_
    Table 1: Parameters of the statistical character database
    BE2015 / 5242 and the evaluation of the different parameters of the database will now be explained.
    The width of the largest Asian and Latin characters is evaluated as follows:
    □ = T “'□ Wia:, m = where the proportionality rate represents a conversion from point size of characters in the library to point size of characters in the text.
    The same is done for the medium size, respectively, of Asian characters and Latin characters:
    D n
    SH □ Ü, D, e x
    This value represents the local estimate of the width of the character Set is also used to evaluate the global estimate of the width of the characters in step S
    The overall character width estimate in step W is calculated using the following equation:
    _ ^ _ d * (Sd1) d ^ s
    where § £] _ □ is the overall estimate of the average width of the characters updated in step Sd 1, ^ is the local estimate of the average size of the characters in step WBest the index of the current step of the process and 1¾ is the height of the line (we assume that the Asian characters are square). The equation is valid for Asian and Latin characters. It is assumed that, for Latin characters, the overall estimate of the width is half the overall estimate of the Asian characters.
    BE2015 / 5242
    Finally, the same proportionality is applied to estimate the inter-character space in the text § when the point size of the text is different from the point size of the reference characters:
    Dg, g □ g ggg.g.g
    This embodiment illustrates the case of a line segmentation method but the method is not limited to a line. Asian text can also be written in columns and the same process can also be used. In this case, the width of the character should be replaced by the height of the character, and the coordinate of the starting point is the coordinate (S | of the first pixel of a character at the top of the character string image.
    权利要求:
    Claims (15)
    [1]
    1. A method for segmenting characters in a character string image, the method comprising the steps of determining a first starting point coordinate of a pixel contrasting with a background, generating a list of dependent dependent widths of characters with a maximum character width and characteristics of the part of the character string image corresponding to the maximum character width, determining a second part of the character string image corresponding to the first starting point coordinate and at the first width on the list of potential character widths, apply the classification process on the second part of the character string image giving a probability of error for the first width and a candidate character, compare the probability d 'error at a first threshold determined by a compromise between speed and accuracy; and selecting the candidate character as the character corresponding to the first width if the probability of error corresponding to the first width is less than the threshold value.
    [2]
    2. The method of claim 1, further comprising the step of comparing the probability of error to a second threshold value greater than the first threshold value; and wherein the step of comparing the probability of error to the first threshold value is performed only if the probability of error is less than the second threshold value.
    BE2015 / 5242
    [3]
    The method of claim 1, further comprising the step of calculating the starting point for the next character if the probability of error corresponding to the first width is less than the second threshold value, and 'keeping the point of start calculated for the next character in memory.
    [4]
    The method of claim 1, further comprising the step of updating statistical values of characters contained in a database if the probability of error corresponding to the first width is less than the first threshold value.
    [5]
    5. Method according to claim 1, in which the list of potential character widths is sorted from most probable to least probable, in which the most probable width is such that it is the largest width containing a maximum number of connected components which are not greater than an estimated maximum width for a character.
    [6]
    6. Method according to claim 5, in which the two least probable widths from the list of potential character widths are an average overall width and half the average overall width, the average overall width being the height of the image of character string for a first character in the character string image and the average overall width being calculated on the basis of a previous average global width and an average character width for a subsequent character in the string image of characters.
    BE2015 / 5242
    [7]
    7. The method as claimed in claim 1, in which if the probability of error corresponding to the first width of the list of potential character widths is greater than the second threshold value, the method further comprises the steps consisting in:
    a) determining a third part of the character string image corresponding to the starting point coordinate and to the next width on the list of potential character widths;
    b) applying the classification method to the third part of the character string image giving a probability of error for this next width and a next candidate character;
    c) compare the probability of error for the next width to the first threshold value;
    d) repeat steps a), b) and c) until the probability of error corresponding to a width is less than the threshold value or until all the widths contained in the list of character widths potential have been dealt with;
    e) select the candidate character as the character corresponding to the width if the probability of error corresponding to the first width is less than the first threshold value.
    [8]
    8. The method of claim 1, wherein the character string image is a vertical character string image and all the widths are heights.
    [9]
    The method of claim 6, further comprising the step of updating a statistical character database with the overall average width value upon a successful iteration.
    [10]
    The method of claim 1, wherein the step of generating a list of potential character widths is based on
    BE2015 / 5242 data retrieved from a database which contains reference characters for a given point size, the width of the largest reference characters, the average width of the reference characters and the size of the average space between characters of reference.
    [11]
    11. The method of claim 10, wherein the database further contains estimates of statistical values of the characters.
    [12]
    The method of claim 11, wherein the database is updated to a successful iteration.
    [13]
    13. The method of claim 1, wherein the maximum character width is a maximum character width for Asian characters.
    [14]
    14. A computer program product comprising a medium usable by a computer in which control logic is stored to cause a computer device to segment a character string image in an input image, the control logic comprising:
    a) first program code means readable by command to determine a first starting point coordinate of a pixel contrasting with a background,
    b) second command-readable program code means for generating a list of potential character widths dependent on a maximum character width and characteristics of the portion of the character string image corresponding to the maximum character width ,
    c) third means of program code readable by command to determine a second part of the image of
    BE2015 / 5242 character string corresponding to the first starting point coordinate and to the first width on the list of potential character widths,
    d) fourth means of program code readable by
    5 command to apply a classification method on the second part of the character string image giving a probability of error for the first width and a candidate character,
    e) fifth means of program code readable by
    10 command to compare the probability of error with a first threshold determined by a compromise between speed and accuracy; and
    f) sixth means of command readable program code for selecting the candidate character as the character corresponding to the first width if the probability of error
    [15]
    15 corresponding to the first width is less than the threshold value.
    类似技术:
    公开号 | 公开日 | 专利标题
    BE1025503B1|2019-03-27|LINE SEGMENTATION METHOD
    EP3117369B1|2020-06-24|Detecting and extracting image document components to create flow document
    BE1022562A9|2017-02-07|Optical character recognition method
    BE1024194B1|2017-12-12|Method for identifying a character in a digital image
    WO2018010657A1|2018-01-18|Structured text detection method and system, and computing device
    US20190130232A1|2019-05-02|Font identification from imagery
    WO2017202232A1|2017-11-30|Business card content identification method, electronic device and storage medium
    EP1836651B1|2018-02-21|Method for searching, recognizing and locating a term in ink, and a corresponding device and computer program
    US10643094B2|2020-05-05|Method for line and word segmentation for handwritten text images
    CN111709420A|2020-09-25|Text detection method, electronic device and computer readable medium
    BE1025502B1|2019-03-27|SYSTEM AND METHOD FOR FORM RECOGNITION USING GABOR FUNCTIONS
    Ayesh et al.2017|A robust line segmentation algorithm for Arabic printed text with diacritics
    BE1026159B1|2020-05-08|IMAGE PROCESSING SYSTEM AND IMAGE PROCESSING METHOD
    EP2686782A1|2014-01-22|Method and device for recognizing and tagging of peaks, increases, or abnormal or exceptional variations in the throughput of a stream of digital documents
    Zhai et al.2016|Chinese image text recognition with BLSTM-CTC: a segmentation-free method
    BE1026039B1|2019-12-13|IMAGE PROCESSING METHOD AND IMAGE PROCESSING SYSTEM
    WO2016086877A1|2016-06-09|Text detection method and device
    JP2018152059A|2018-09-27|Inferring character stroke information from image
    Shetty et al.2015|Optical character recognition for alphanumerical character verification in video frames
    FR2880708A1|2006-07-14|Term e.g. typed character, searching method for digital handwritten document, involves converting handwritten data into intermediate data, in intermediate format, in form of segmentation graph, and searching terms on intermediate data
    CN113887375A|2022-01-04|Text recognition method, device, equipment and storage medium
    BE1020588A5|2014-01-07|METHOD OF RECOGNIZING FORMS, COMPUTER PROGRAM PRODUCT, AND MOBILE TERMINAL.
    CN113850178A|2021-12-28|Video word cloud generation method and device, storage medium and electronic equipment
    BE1025134A1|2018-11-09|Method of identifying a character in a digital image
    EP2546801A1|2013-01-16|Method for detecting and correcting the direction of a document in a digital image
    同族专利:
    公开号 | 公开日
    WO2015158781A1|2015-10-22|
    BE1025503A1|2019-03-20|
    JP6693887B2|2020-05-13|
    CN106255979B|2019-07-12|
    US20150302598A1|2015-10-22|
    JP2017515222A|2017-06-08|
    CN106255979A|2016-12-21|
    KR102345498B1|2021-12-31|
    KR20170004983A|2017-01-11|
    US9183636B1|2015-11-10|
    引用文献:
    公开号 | 申请日 | 公开日 | 申请人 | 专利标题
    US4562594A|1983-09-29|1985-12-31|International Business Machines Corp. |Method and apparatus for segmenting character images|
    US5060277A|1985-10-10|1991-10-22|Palantir Corporation|Pattern classification means using feature vector regions preconstructed from reference data|
    JPH04270485A|1991-02-26|1992-09-25|Sony Corp|Printing character recognition device|
    JPH05128307A|1991-10-31|1993-05-25|Toshiba Corp|Character recognition device|
    US6041141A|1992-09-28|2000-03-21|Matsushita Electric Industrial Co., Ltd.|Character recognition machine utilizing language processing|
    CN1145872C|1999-01-13|2004-04-14|国际商业机器公司|Method for automatically cutting and identiying hand written Chinese characters and system for using said method|
    JP2001195544A|2000-01-07|2001-07-19|Fujitsu Ltd|Character segmenting device|
    US7734636B2|2005-03-31|2010-06-08|Xerox Corporation|Systems and methods for electronic document genre classification using document grammars|
    JP2007058803A|2005-08-26|2007-03-08|Canon Inc|Online hand-written character recognition device, and online hand-written character recognition method|
    JP4424309B2|2006-01-23|2010-03-03|コニカミノルタビジネステクノロジーズ株式会社|Image processing apparatus, character determination program, and character determination method|
    JP4662066B2|2006-07-12|2011-03-30|株式会社リコー|Image processing apparatus, image forming apparatus, image distribution apparatus, image processing method, program, and recording medium|
    JP4860574B2|2006-09-13|2012-01-25|株式会社キーエンス|Character segmentation device, method and program|
    CN101398894B|2008-06-17|2011-12-07|浙江师范大学|Automobile license plate automatic recognition method and implementing device thereof|
    CN101770576A|2008-12-31|2010-07-07|北京新岸线网络技术有限公司|Method and device for extracting characters|
    DE102009029186A1|2009-09-03|2011-03-10|BSH Bosch und Siemens Hausgeräte GmbH|Dishwasher with a fleet storage and associated method|
    US8385652B2|2010-03-31|2013-02-26|Microsoft Corporation|Segmentation of textual lines in an image that include western characters and hieroglyphic characters|
    US8571270B2|2010-05-10|2013-10-29|Microsoft Corporation|Segmentation of a word bitmap into individual characters or glyphs during an OCR process|
    US8606010B2|2011-03-18|2013-12-10|Seiko Epson Corporation|Identifying text pixels in scanned images|
    JP5075997B2|2011-03-30|2012-11-21|株式会社東芝|Electronic device, program, and character string recognition method|
    US8611662B2|2011-11-21|2013-12-17|Nokia Corporation|Text detection using multi-layer connected components with histograms|
    JP5547226B2|2012-03-16|2014-07-09|株式会社東芝|Image processing apparatus and image processing method|
    US9330070B2|2013-03-11|2016-05-03|Microsoft Technology Licensing, Llc|Detection and reconstruction of east asian layout features in a fixed format document|EP2836962A4|2012-04-12|2016-07-27|Tata Consultancy Services Ltd|A system and method for detection and segmentation of touching characters for ocr|
    US9798943B2|2014-06-09|2017-10-24|I.R.I.S.|Optical character recognition method|
    JP6352695B2|2014-06-19|2018-07-04|株式会社東芝|Character detection apparatus, method and program|
    CN106156766B|2015-03-25|2020-02-18|阿里巴巴集团控股有限公司|Method and device for generating text line classifier|
    US9836646B2|2015-10-15|2017-12-05|I.R.I.S.|Method for identifying a character in a digital image|
    CN110135426B|2018-02-09|2021-04-30|北京世纪好未来教育科技有限公司|Sample labeling method and computer storage medium|
    WO2020041448A1|2018-08-22|2020-02-27|Leverton Holding Llc|Text line image splitting with different font sizes|
    法律状态:
    2019-05-08| FG| Patent granted|Effective date: 20190327 |
    优先权:
    申请号 | 申请日 | 专利标题
    US14/254,096|US9183636B1|2014-04-16|2014-04-16|Line segmentation method|
    US14254096|2014-04-16|
    [返回顶部]