JP5881380B2 - Word recognition device, word recognition program, and paper sheet processing device provided with word recognition device - Google Patents

Description

  Embodiments described herein relate generally to a word recognition apparatus, a word recognition program, and a paper sheet processing apparatus including the word recognition apparatus.

  2. Description of the Related Art Conventionally, a paper sheet processing apparatus such as a mail sorting machine that processes paper sheets such as mail has been put into practical use. Such a paper sheet processing apparatus takes in paper sheets that have been input into the input unit one by one, and acquires an image from the paper sheets. The paper sheet processing apparatus includes a word recognition device. The word recognition device recognizes a word on a paper sheet based on an image acquired from the paper sheet. The paper sheet processing apparatus identifies an address or other information written on the paper sheet based on the recognition result, and sorts the paper sheet into a predetermined sorting pocket.

  As a method of recognizing a word, an analytical method (Analytic Approach) and an overall method (Holistic Approach) are generally known. The analytical method and the overall method have a complementary relationship. For this reason, the word recognition apparatus can recognize a word with higher accuracy by using both the analytical method and the overall method.

  When recognizing a word by an analytical method, the word recognition device extracts a plurality of cutting point candidates based on the word image, and generates a plurality of character candidates including overlapping portions based on the extracted cutting point candidates. . Furthermore, the word recognition device calculates an evaluation value of the analytical method using the posterior probability ratio, and selects a correct combination from a plurality of character candidates based on the evaluation value.

  Further, when recognizing a word by the overall method, the word recognizing device recognizes the word by using, for example, a hidden Markov model (HMM: Hidden Markov Model).

Japanese Patent No. 4601835

  As a method of using both the analytical method and the overall method, recognition is first performed by the overall method, the word is divided into characters based on the recognition result, and each divided character is recognized by the analytical method. Thus, there is a method of verifying the word recognition result. However, when a misperception occurs in recognition by the overall method, the word recognition device fails to recognize characters, so that the word recognition device cannot obtain sufficient accuracy.

  Further, when the analytical method and the overall method are used in combination, the approximation error of the evaluation value may increase. In such a case, there is a problem that the word recognition device cannot recognize words with sufficient accuracy.

  Therefore, an object of the present invention is to provide a word recognition device that can recognize words with higher accuracy, a word recognition program, and a paper sheet processing device including the word recognition device.

  A word recognition device according to an embodiment includes a word dictionary that stores a plurality of words, an image receiving unit that receives an image including a word, a word image extracting unit that extracts a word image for each word from the image, and the word Character candidate extraction means for extracting character candidates from an image, analytical matching means for recognizing the character candidates and calculating a first evaluation value for each word stored in the word dictionary, An overall matching unit that calculates a feature probability of appearing and calculates a second evaluation value based on the word image, a word model for each word stored in the word dictionary, and the feature probability; Correction means for correcting the first evaluation value with the first correction parameter and correcting the second evaluation value with a second correction parameter set in advance; and the corrected first evaluation value; Integrated evaluation value calculating means for calculating the third evaluation value by integrating the corrected second evaluation value, and output means for outputting the third evaluation value calculated by the integrated evaluation value calculating means And.

FIG. 1 is a diagram for explaining an example of a paper sheet processing apparatus according to an embodiment. FIG. 2 is a diagram for describing an example of a word recognition device according to an embodiment. FIG. 3 is a diagram for explaining processing of the word recognition device according to the embodiment. FIG. 4 is a diagram for explaining processing of the word recognition device according to the embodiment. FIG. 5 is a diagram for explaining processing of the word recognition device according to the embodiment. FIG. 6 is a diagram for explaining the processing of the word recognition device according to the embodiment. FIG. 7 is a diagram for describing an example of a word recognition device according to an embodiment. FIG. 8 is a diagram for describing an example of a word recognition device according to an embodiment. FIG. 9 is a diagram for explaining processing of the word recognition device according to the embodiment. FIG. 10 is a diagram for explaining processing of the word recognition device according to the embodiment. FIG. 11 is a diagram for explaining processing of the word recognition device according to the embodiment. FIG. 12 is a diagram for explaining processing of the word recognition device according to the embodiment.

  Hereinafter, a paper sheet processing apparatus and a word recognition apparatus according to an embodiment will be described in detail with reference to the drawings.

FIG. 1 shows a configuration example of a paper sheet processing apparatus 100 according to an embodiment.
The paper sheet processing apparatus 100 reads an image from the paper sheet, recognizes destination information and a stamping position from the read image, and stamps the paper sheet to classify the paper sheet. The sheet processing apparatus 100 includes a supply unit 200, a separation roller 210, a conveyance path 220, an image reading unit 400, a stamping unit 460, a printing unit 470, a main control unit 500, a sorting processing unit 300, a word recognition unit 600, and an operation unit. 700, a display unit 800, and an input / output unit 900.

  The main control unit 500 controls the operation of each unit of the paper sheet processing apparatus 100 in an integrated manner. The main control unit 500 includes a CPU, a buffer memory, a program memory, a nonvolatile memory, and the like. The CPU performs various arithmetic processes. The buffer memory temporarily stores the results of calculations performed by the CPU. The program memory and the nonvolatile memory store various programs executed by the CPU, control data, and the like. The main control unit 500 can perform various processes by executing a program stored in the program memory by the CPU.

  The supply unit 200 stocks the paper sheets 1 to be taken into the paper sheet processing apparatus 100. The supply unit 200 collectively receives the stacked paper sheets 1.

  The separation roller 210 is installed at the lower end of the supply unit 200, for example. When the paper sheet 1 is loaded into the supply unit 200, the separation roller 210 contacts the lower end of the loaded paper sheet 1 in the stacking direction. The separation roller 210 rotates to take the sheets 1 set in the supply unit 200 one by one from the lower end in the stacking direction into the sheet processing apparatus 100.

  For example, the separation roller 210 takes in one sheet 1 every rotation. Thereby, the separation roller 210 can take in the paper sheets 1 at a constant pitch. The paper sheet 1 taken in by the separation roller 210 is introduced into the conveyance path 220.

  The conveyance path 220 is a conveyance unit that conveys the paper sheet 1 to each unit in the paper sheet processing apparatus 100. The conveyance path 220 includes a conveyance belt (not shown) and a drive pulley (not shown). The conveyance path 220 drives a drive pulley by a drive motor (not shown). The conveyor belt is operated by a driving pulley.

  The transport path 220 transports the paper sheet 1 taken in by the separation roller 210 in the direction of arrow a (transport direction a) at a constant speed by the transport belt. Note that the side closer to the separation roller 210 in the conveyance path 220 will be described as an upstream side, and the opposite side will be described as a downstream side.

  The image reading unit 400 acquires an image from the paper sheet 1 conveyed by the conveyance path 220. The image reading unit 400 includes, for example, illumination and an optical sensor. The illumination irradiates light onto the paper sheet 1 conveyed by the conveyance path 220. The optical sensor includes a light receiving element such as a Charge Coupled Device (CCD) and an optical system (lens). The optical sensor receives the reflected light reflected by the paper sheet 1 by the optical system, forms an image on the CCD, and acquires an electrical signal (image). The image reading unit 400 acquires an entire image of the paper sheet 1 by continuously acquiring images from the paper sheet 1 conveyed by the conveyance path 220. The image reading unit 400 supplies the acquired image to the main control unit 500. Note that the image reading unit 400 may include a video camera or the like.

  Based on the image received from the image reading unit 400, the main control unit 500 performs processing for determining the transport destination of the paper sheet 1. For this purpose, the main control unit 500 identifies an address (destination information) as a destination by causing the word recognition unit 600 to recognize a word in the image. The main control unit 500 generates an image such as a two-dimensional code or a barcode based on the destination information, and supplies the generated image to the printing unit 470.

  Further, the main control unit 500 specifies the position where a stamp or the like on the paper sheet 1 is pasted. Furthermore, the main control unit 500 controls the operation of the stamp unit 460 based on the specified stamp position.

  The stamp unit 460 stamps, for example, a stamp such as a date stamp on the paper sheet 1 based on the control of the main control unit 500. That is, the main control unit 500 controls the stamping unit 460 so as to stamp the stamp at a position overlapping the stamping position. For example, the stamp unit 460 stamps the stamp so as to be a split.

  The printing unit 470 prints an image such as a two-dimensional code or a barcode based on the control of the main control unit 500. That is, the printing unit 470 prints an image such as a two-dimensional code or a barcode supplied from the main control unit 500. For example, the printing unit 470 prints the above image with an ink containing a phosphor that emits excitation light when irradiated with ultraviolet rays.

  The sorting processing unit 300 sorts and stacks the sheets 1 based on the control of the main control unit 500. The sorting processing unit 300 includes, for example, a plurality of gates and stackers such as a first gate 310, a first stacker 320, a second gate 330, and a second stacker 340. Further, the sorting processing unit 300 further includes a plurality of gates and a plurality of stackers. A stacker is provided for each destination information, for example. A gate is provided for each stacker.

  The main control unit 500 can sort the sheets 1 by controlling each gate of the sorting processing unit 300. Thereby, the sorting processing unit 300 can stack the paper sheets 1 in different stackers for each piece of destination information of the paper sheets 1.

  The first gate 310 and the second gate 330 are provided downstream of the image reading unit 400, the stamping unit 460, and the printing unit 470 in the conveyance path 220. The first gate 310 and the second gate 330 each operate based on the control of the main control unit 500. The main control unit 500 controls the first gate 310 and the second gate 330 according to the destination information recognized by the above processing.

  The first gate 310 switches the transport destination of the paper sheet 1 between the first stacker 320 and the second gate 330. The second gate 330 switches the transport destination of the paper sheet 1 between the second stacker 340 and another stacker.

  The main control unit 500 supplies the image received from the image reading unit 400 to the word recognition unit 600 so that the word recognition unit 600 can recognize words in the image.

  The word recognition unit 600 recognizes a word in the received image. The word recognition unit 600 outputs the recognition result to the main control unit 500. The main control unit 500 identifies destination information and the like based on the recognition result by the word recognition unit 600.

  In addition, the main control unit 500 includes a memory that holds an image of the paper sheet 1 for which destination information could not be specified. In addition, the sorting processing unit 300 includes a stacker that accumulates the paper sheets 1 whose destination information could not be specified.

  The operation unit 700 receives various operation inputs from the operator through the operation unit. The operation unit 700 generates an operation signal based on an operation input by the operator, and transmits the generated operation signal to the main control unit 500.

  For example, the paper sheet processing apparatus 100 may include a VCS (Video Coding System) function. That is, the main control unit 500 of the paper sheet processing apparatus 100 causes the display unit 800 to display an image of the paper sheet 1 whose destination information could not be specified. The paper sheet processing apparatus 100 causes the operator to read the image of the paper sheet 1 displayed on the display unit 800 and causes the operation unit 700 to input destination information. Thereby, the paper sheet processing apparatus 100 can acquire correct destination information.

  The display unit 800 displays various screens based on the control of the main control unit 500. For example, the display unit 800 displays various operation guidance and processing results for the operator. Further, as described above, the display unit 800 may be configured to display an image of the paper sheet 1 for which the destination information is not specified. Note that the operation unit 700 and the display unit 800 may be integrally formed as a touch panel.

  The input / output unit 900 transmits / receives data to / from an external device connected to the paper sheet processing apparatus 100 or a storage medium. For example, the input / output unit 900 includes a disk drive, a USB connector, a LAN connector, or another interface capable of transmitting and receiving data. The paper sheet processing apparatus 100 can acquire data from an external device connected to the input / output unit 900 or a storage medium. In addition, the paper sheet processing apparatus 100 can transmit the processing result to an external device connected to the input / output unit 900 or a storage medium.

FIG. 2 shows an example of the configuration of the word recognition unit 600 according to an embodiment.
The word recognition unit 600 includes an image receiving unit 601, a word extraction unit 602, a character candidate extraction unit 603, a character recognition unit 604, a feature extraction unit 605, an analytical matching unit 610, an overall matching unit 620, a feature probability calculation unit 630, VCS 640, word image storage unit 641, model learning unit 642, model storage unit 643, word model generation unit 644, word dictionary 645, prior probability calculation unit 651, prior probability storage unit 652, prior probability input unit 653, recognition unit 660, And a learning unit 680.

  The operation of the word recognition unit 600 is roughly divided into a recognition phase and a learning phase. First, the recognition phase will be described.

  The word recognizing unit 600 performs analytical matching, performs overall matching, calculates feature probabilities, and integrates these results and prior probabilities for each word. Thereby, the word recognition part 600 can calculate the evaluation value (posterior probability) for every word.

  In general, in pattern recognition, it is optimal to assign a pattern to a category having the maximum posterior probability. That is, when a pattern belongs to a category having the maximum posterior probability, the identification error is minimized. That is, the word recognizing unit 600 can identify one word and transmit it to the main control unit 500 by outputting the word for which the highest posterior probability is calculated as a recognition result. For example, the word recognition unit 600 may be configured to output an evaluation value for each word to the main control unit 500 as a recognition result. In this case, the main control unit 500 can specify the destination information in consideration of the posterior probabilities of a plurality of words and combinations with other words.

For example, when all character recognition results in word candidates in analytical matching are Y and features extracted from an image used for overall matching are X, the word recognition unit 600 uses the following formula 1 to generate the word w The posterior probability P (w | Y, X) is calculated.

  The left side of Equation 1 shows a posterior probability of a word in a case where a set of character recognition results by analytical matching and a set of feature extraction results used for overall matching are used as conditions. That is, the posterior probability P (w | Y, X) indicates an evaluation value for each word when analytical matching and overall matching are used together.

  The left side of Equation 1 can be expanded like the right side by Bayes' theorem. Furthermore, by assuming that the result of the analytical matching and the result of the overall matching are independent from each other, the right side of the first stage of Equation 1 can be approximated as shown on the right side of the second stage.

  Note that P (Y | w) / P (Y) in Equation 1 indicates the result of analytical matching (a posteriori probability ratio). In addition, P (X | w) in Equation 1 indicates the overall matching result (likelihood). Furthermore, P (X) in Formula 1 indicates the calculation result of the feature probability. Furthermore, P (w) indicates the prior probability for the word w.

  The word recognition unit 600 can calculate the a posteriori probability for each word by calculating each of the above terms and calculating Formula 1.

  First, analytical matching will be described. The image receiving unit 601 of the word recognition unit 600 receives the image of the paper sheet 1 (paper sheet image) from the main control unit 500. FIG. 3 shows an example of a paper sheet image. As shown in FIG. 3, the image receiving unit 601 receives a paper sheet image including words written on the paper sheet 1. FIG. 3 shows an example in which a destination and the like are described using English words. However, the present embodiment can be applied even if the destination described on the paper sheet 1 is in Japanese or another language. The image receiving unit 601 transmits the received paper sheet image to the word extracting unit 602.

  The word extraction unit 602 extracts word candidates (word images) from the paper sheet image received by the image receiving unit 601. For example, the word extraction unit 602 identifies and extracts an area that is likely to be segmented as a word by performing image processing on the paper sheet image. FIG. 4 shows an example of word candidates. As shown in FIG. 4, the word extraction unit 602 extracts word candidates from the paper sheet image.

  For example, the word extraction unit 602 extracts word candidates by recognizing spaces between words, for example. In addition, for example, the word extraction unit 602 may be configured to extract word candidates by extracting “city”, “town”, or other keywords that serve as delimiters. Further, the process of extracting the word candidates is not limited to the above method, and any process may be used. The word extraction unit 602 transmits the extracted word candidates to the character candidate extraction unit 603 and the feature extraction unit 605.

  Character candidate extraction unit 603 extracts character candidates from word candidates. The character candidate extraction unit 603 identifies and extracts a region that is likely to be segmented as a character by performing image processing on the word candidate (word image). FIG. 5 shows an example of processing for extracting character candidates from word candidates. As illustrated in FIG. 5, the character candidate extraction unit 603 extracts a plurality of cut point candidates from the character candidates, and extracts a plurality of character candidates including overlapping portions based on the extracted cut point candidates. That is, the character candidate extraction unit 603 identifies a region that is highly likely to be recognized as one character and extracts it as a character candidate. The character candidate extraction unit 603 transmits the extracted character candidates to the character recognition unit 604.

  The character recognition unit 604 performs character recognition for each character candidate and acquires a character recognition result. That is, the character recognition unit 604 obtains a character recognition result by comparing a character candidate image with a character recognition dictionary prepared in advance. The character recognition unit 604 transmits the character recognition result for each character candidate to the analytical matching unit 610.

  The word dictionary 645 stores words to be recognized as a list. FIG. 6 shows an example of the word dictionary 645. When recognizing a word, the word recognition unit 600 selects a correct word from the list in the word dictionary 645. The word dictionary 645 supplies the word list to the analytical matching unit 610.

  The analytical matching unit 610 calculates the posterior probability ratio for each word stored in the word dictionary 645 based on the character recognition result for each character candidate transmitted from the character recognition unit 604. Accordingly, the analytical matching unit 610 searches for a correct path (route) of the plurality of character candidates extracted by the character candidate extraction unit 603.

For example, the i-th character of the word w is c _i , the serial number of the character candidate corresponding to the i-th character is f (i), and the character recognition result of the character candidate corresponding to the i-th character is y _{f ( i) When} the number of characters of the word w is N, the posterior probability ratio P (Y | w) / P (Y) of the word w is approximated as shown in Equation 2 below.

For example, when the target word is “ham”, c ₁ = “h”, c ₂ = “a”, and c ₃ = “m”. In this case, N = 3. In this case, P (y _{f (i)} | c _i ) / P (y _{f (i)} ) represents the posterior probability ratio of the i-th character.

  The analytical matching unit 610 calculates the posterior probability ratio P (Y | w) / P (Y) of the word w by multiplying the posterior probability ratio of the i-th character over i = 1 to N. be able to. In other words, the analytical matching unit 610 calculates the posterior probability ratio for each word by calculating the above Equation 2 based on the character recognition result for each word in the word list.

The analytical matching unit 610 includes a character probability calculation unit 611, a first calculation unit 612, and a second calculation unit 613. The character probability calculation unit 611 calculates the numerator of each factor on the right side of Equation 2. That is, the character probability calculation unit 611 calculates P (y _{f (i)} | c _i ) for each character of a certain word w.

The first calculation unit 612 calculates each factor on the right side of Equation 2. That is, the first calculation unit 612 calculates P (y _{f (i)} ) that is the denominator of the right side, and divides P (y _{f (i)} | c _i ) that is the numerator by the calculated value. Note that P (y _{f (i)} ) is the probability that the character recognition result y _{f (i)} will appear.

  The second calculation unit 613 calculates the right side of Equation 2. That is, the second calculation unit 613 multiplies all the factors on the right side of Formula 2 that is the calculation result of the first calculation unit 612. Thereby, the analytical matching unit 610 can calculate the posterior probability ratio P (Y | w) / P (Y) of the word w. The analytical matching unit 610 outputs the calculated posterior probability ratio P (Y | w) / P (Y) to the recognition unit 660.

  Next, overall matching will be described. The feature extraction unit 605 in FIG. 2 receives the image of the word candidate from the word extraction unit 602 as described above. The feature extraction unit 605 extracts a feature X that is a set of vectors based on the received word candidate images.

  For example, the feature extraction unit 605 extracts the feature X using the brightness gradient information after blurring the word candidate image as a 128-dimensional vector. The feature extraction unit 605 extracts a plurality of features while shifting a region of interest (a region of interest) in the word candidate image from left to right in the image.

  The feature extraction unit 605 may be configured to use the density value of the pixel of the word candidate image as a feature. Furthermore, the feature extraction unit 605 may be configured to use a pattern density value obtained by further simplifying a word candidate image as a feature.

  Through the above processing, the feature extraction unit 605 extracts a plurality of features from one word candidate image. The feature extraction unit 605 outputs the extracted feature X to the overall matching unit 620 and the feature probability calculation unit 630.

  The model storage unit 643 stores a character model for each character or a word model for each word. The model storage unit 643 may be configured to store a word model corresponding to each word in the word dictionary 645. The model storage unit 643 may be configured to store a word model corresponding to an arbitrary word in the word dictionary 645.

  The word model generation unit 644 generates a word model corresponding to each word in the word dictionary 645 using the character model and the word model stored in the model storage unit 643. The word model generation unit 644 outputs the generated word model to the overall matching unit 620.

  For example, the word model generation unit 644 reads a character model stored in the model storage unit 643 and generates a word model by connecting the character models according to the words in the word dictionary 645. When a word model corresponding to a word in the word dictionary 645 is stored in the model storage unit 643, the word model generation unit 644 directly uses the word model stored in the model storage unit 643 as an overall matching unit 620. Output to.

  The overall matching unit 620 calculates a likelihood P (X | w) based on the feature X extracted by the feature extraction unit 605 and the word model output from the word model generation unit 644. The likelihood P (X | w) is a probability that the feature X extracted by the feature extraction unit 605 is output from the word model output from the word model generation unit 644. The likelihood P (X | w) is the same as the numerator of the second factor on the right side of Equation 1.

  The overall matching unit 620 calculates a likelihood P (X | w) by using a Viterbi algorithm.

  The Viterbi algorithm is an algorithm that calculates a state sequence having the highest possibility (likelihood) of outputting a given array when model parameters are known. That is, the Viterbi algorithm is a dynamic programming algorithm that searches for a sequence of hidden events that result in feature X.

  The overall matching unit 620 calculates a likelihood P (X | w) as a probability that the feature X appears using the Viterbi algorithm with the word model output from the word model generation unit 644 as a known parameter. That is, the likelihood P (X | w) indicates the probability that the feature X appears from the word model corresponding to the word w. The overall matching unit 620 outputs the calculated likelihood P (X | w) to the recognition unit 660.

Next, calculation of the feature probability will be described. The feature probability calculation unit 630 in FIG. 2 calculates the feature probability P (X) based on the feature X extracted by the feature extraction unit 605 and the parameters stored in advance. As described above, the feature extraction unit 605 extracts the feature X, which is a set of vectors, based on the word candidate images. This feature X has T feature vectors x ₁ , x ₂ , x ₃ ... X _T. In this case, the feature vector _xt indicates the t-th feature vector.

Assuming the above, the feature probability calculation unit 630 calculates the feature probability P (X) based on the following Equation 3.

That is, the feature probability calculation unit 630 can calculate the feature probability P (X) by approximating that all feature vectors _xt are independent and calculating the above Equation 3.

The feature probability calculation unit 630 calculates P (x _t ) from t = 1 to T based on parameters stored in advance.

The above parameters are parameters calculated by learning based on a plurality of word images. This parameter indicates the probability distribution formed by the feature vectors x ₁ , x ₂ , x ₃ ... Of the feature X extracted based on the word image. When modeled with a mixed Gaussian distribution, the parameter may be a mixing ratio, an average vector, or a covariance matrix of each Gaussian distribution.

The feature probability calculation unit 630 calculates the feature probability P (X) by multiplying all the calculated P (x _t ). The feature probability calculation unit 630 outputs the calculated feature probability P (X) to the recognition unit 660.

  Note that the posterior probability ratio P (Y | w) / P (Y), which is the first factor in the above-described Equation 1, is referred to as an analytical matching evaluation value. Further, P (X | w) / P (X), which is the second factor of the above-described Equation 1, is referred to as an overall matching evaluation value.

The recognizing unit 660 integrates the calculation results of the analytical matching unit 610, the overall matching unit 620, and the feature probability calculation unit 630, and calculates an integrated evaluation value (a posteriori probability). The posterior probability P (w | Y, X) calculated by the above mathematical formula 1 may have a large approximation error when the length of the character string of the word changes. For this reason, the recognizing unit 660 calculates a corrected posterior probability (integrated evaluation value) using Equation 4 obtained by introducing the correction parameter α and the correction parameter β into Equation 1.

As indicated by Equation 4, the correction parameter α corrects the evaluation value of analytical matching. Specifically, the correction parameter α raises the analytical matching evaluation value to the power of α. Also,
As shown by Equation 4, the correction parameter β corrects the evaluation value of the overall matching. Specifically, the correction parameter β raises the overall matching evaluation value to the power of β.

  FIG. 7 shows an example of the configuration of the recognition unit 660. The recognition unit 660 includes a first correction evaluation value calculation unit 661, a first correction parameter storage unit 662, a second correction evaluation value calculation unit 663, a second correction parameter storage unit 664, and an integrated evaluation value calculation unit 665. Is provided.

The first correction evaluation value calculation unit 661 calculates {P (Y | w) / P (Y)} ^α of Expression 4 using the correction parameter α. That is, the first correction evaluation value calculation unit 661 corrects the evaluation value of analytical matching with the correction parameter α.

  The first correction parameter storage unit 662 is a storage unit that stores the correction parameter α. The first correction parameter storage unit 662 receives and stores the correction parameter α selected by the learning unit 680 described later.

The second correction evaluation value calculation unit 663 calculates {P (X | w) / P (X)} ^β in Expression 4 using the correction parameter β. That is, the second correction evaluation value calculation unit 663 corrects the evaluation value of the overall matching with the correction parameter β.

  Each evaluation value before correction (that is, an evaluation value for analytical matching and an evaluation value for overall matching) increases exponentially. For this reason, it is desirable not only to improve accuracy but also to prevent overflowing during calculation, to use correction values obtained by logarithm on both sides of Equation 4.

  The second correction parameter storage unit 664 is a storage unit that stores the correction parameter β. The second correction parameter storage unit 664 receives and stores the correction parameter β selected by the learning unit 680 described later.

  The integrated evaluation value calculation unit 665 includes the analytical matching evaluation value corrected by the first correction evaluation value calculation unit 661, the overall matching evaluation value corrected by the second correction evaluation value calculation unit 663, and An integrated evaluation value obtained by integrating the prior probabilities stored in the prior probability storage unit 652 is calculated. The integrated evaluation value calculation unit 665 calculates the right side in a state where the logarithm of both sides of Equation 4 is taken. By taking the logarithm, the integrated evaluation value calculation unit 665 can represent each factor on the right side as the sum of logarithms.

  Prior probability storage unit 652 stores prior probability P (w) for each word as a table. The prior probability P (w) is a probability indicating the frequency with which a certain word is described in the paper sheet 1. By adjusting this value and creating a table, the posterior probability P (w | Y, X) of a word inappropriate as an address can be suppressed.

  For example, a barcode on the paper sheet 1 may be recognized as a word such as “11111111”. Even in such a case, by setting a low value as a prior probability P (w) in a word such as “11111111” in advance, the word recognition unit 600 can determine the posterior probability P (w |) of the word “11111111”. Y, X) can be prevented from being calculated as a high value. That is, by setting in advance a low value as the prior probability P (w) for a word that is easily misrecognized, it is possible to prevent the word recognition unit 600 from causing erroneous recognition.

  For example, when the appearance frequency of all the words is uniform, the prior probability P (w) may be a constant value.

As shown in Equation 4, the integrated evaluation value calculation unit 665 is corrected by the analytical matching evaluation value corrected by the first correction evaluation value calculation unit 661 and the second correction evaluation value calculation unit 663. The overall matching evaluation value is multiplied by the prior probability P (w), and the integrated evaluation value {P (Y | w) / P (Y)} ^α · {P (X | w) / P (X)} ^β · P (w) is calculated. The integrated evaluation value calculation unit 665 outputs the calculated integrated evaluation value. Thereby, the word recognition unit 600 can output the integrated evaluation value to the main control unit 500.

  Through the above processing, the main control unit 500 can acquire a recognition result (evaluation value) for each word. The main control unit 500 can specify the destination information in consideration of the posterior probabilities of a plurality of words and combinations with other words. For example, the main control unit 500 can infer a combination of words suitable as destination information.

  Further, as described above, the word recognition unit 600 corrects the evaluation value of analytical matching with the correction parameter α, and corrects the evaluation value of overall matching with the correction parameter β. The word recognition unit 600 multiplies the corrected evaluation value of analytical matching, the evaluation value of overall matching, and the prior probability to calculate an integrated evaluation value. The integrated evaluation value calculation unit 665 outputs the calculated integrated evaluation value. As a result, the word recognition unit 600 can output a more accurate integrated evaluation value to the main control unit 500.

  Note that the word recognition unit 600 may be configured to output the word for which the highest posterior probability P (w | Y, X) is calculated to the main control unit 500 as a recognition result. In this case, the word recognition unit 600 can identify one word as a recognition result and transmit it to the main control unit 500.

  Further, as described above, the word recognition unit 600 may be configured to output the posterior probability P (w | Y, X) for each word to the main control unit 500 as a recognition result. In this case, the main control unit 500 can specify the destination information in consideration of the posterior probabilities P (w | Y, X) of a plurality of words and combinations with other words.

Next, the learning phase will be described.
The VCS 640 shown in FIG. 2 is a module for causing the operator of the paper sheet processing apparatus 100 to input correct destination information of the paper sheet 1 whose destination information has not been recognized by the word recognition unit 600, for example. The VCS 640 includes, for example, an operation unit 700 and a display unit 800 illustrated in FIG. Further, for example, the word recognition unit 600 may be configured to include a module capable of operation and display as the VCS 640 separately from the operation unit 700 and the display unit 800.

  The VCS 640 displays an image of the paper sheet 1 whose destination information could not be specified. The VCS 640 causes the operator to read the displayed image of the paper sheet 1 and input destination information. For example, the VCS 640 causes the operator to input a correct word for each word candidate. As a result, the VCS 640 can associate the word image with the correct destination information (correct answer).

  The VCS 640 outputs the word image and correct destination information (correct answer) to the word image storage unit 641 and the prior probability calculation unit 651.

  First, word model learning will be described. The word image storage unit 641 stores the word image input by the VCS 640 and the correct answer in association with each other.

  The model learning unit 642 learns either or both of each character model and each word model using the word image stored in the word image storage unit 641 and its correct answer.

  The model learning unit 642 performs model learning using, for example, a Baum-Welch algorithm. The Baumwelch algorithm is an algorithm that searches for unknown parameters in a hidden Markov model. The Baumwelch algorithm can estimate model parameters from the sequence output by the model.

  For example, the model learning unit 642 generates a model by the Baum Welch algorithm using the word image stored in the word image storage unit 641 and its correct answer. The model learning unit 642 outputs the generated model to the model storage unit 643. The model storage unit 643 stores the received model.

  The model learning unit 642 may be configured to update a model already stored in the model storage unit 643.

  Next, learning of prior probabilities will be described. The prior probability calculation unit 651 counts the frequency for each word based on the correct destination information of the word image input by the VCS 640. That is, the prior probability calculation unit 651 calculates the prior probability P (w) for each word by counting and counting the number of words included in the destination information for each word. The prior probability calculation unit 651 stores the calculated prior probability P (w) for each word in the prior probability storage unit 652.

  Prior probability input section 653 can change prior probability P (w) stored in prior probability storage section 652. The prior probability input unit 653 rewrites the prior probability P (w) stored in the prior probability storage unit 652 to a value corresponding to the operation based on, for example, an operation input by the operation unit 700 illustrated in FIG.

  The prior probability input unit 653 is a value corresponding to the prior probability P (w) stored in the prior probability storage unit 652 based on an operation input by a module that can be operated separately from the operation unit 700. The configuration may be rewritten as follows.

  Thereby, a low value can be set as the prior probability P (w) for the above-described words that are easily misrecognized. Thereby, it is possible to prevent the word recognition unit 600 from causing erroneous recognition.

  Next, learning of the correction parameter α and the correction parameter β will be described. In this learning process, the image itself acquired from the paper sheet 1 may be used, or an evaluation value before correction in the above recognition process may be used. In this example, a method using an evaluation value before correction will be described.

  Note that the analytical matching unit 610 shown in FIG. 2 outputs the calculated posterior probability ratio P (Y | w) / P (Y) to the learning unit 680. Also, the overall matching unit 620 illustrated in FIG. 2 outputs the calculated likelihood P (X | w) to the learning unit 680. Also, the feature probability calculation unit 630 illustrated in FIG. 2 outputs the calculated feature probability P (X) to the learning unit 680.

  The analytical matching unit 610, the overall matching unit 620, and the feature probability calculation unit 630 output an ID (word image ID) indicating the word image of the word used for calculating the evaluation value to the learning unit 680 together with the evaluation value. To do. The learning unit 680 receives the analytical matching evaluation value, the global matching evaluation value, and the word image ID from the analytical matching unit 610, the global matching unit 620, and the feature probability calculation unit 630, respectively.

  FIG. 8 shows an example of the configuration of the learning unit 680. The recognition unit 660 includes a first evaluation value storage unit 681, a second evaluation value storage unit 682, an online recognition result storage unit 683, a VCS result storage unit 684, a simulation unit 685, and a parameter selection unit 686.

  The first evaluation value storage unit 681 stores the analytical matching evaluation value output from the analytical matching unit 610 and the word image ID in association with each other.

  The second evaluation value storage unit 682 stores the overall matching evaluation value output from the global matching unit 620 and the feature probability calculation unit 630 and the word image ID in association with each other.

  The online recognition result storage unit 683 stores a word character string selected as a recognition result and a word image ID when processed by a postal sorting machine or the like and automatically recognized normally. The word character string and the word image ID selected as the recognition result when the automatic recognition can be normally performed are supplied to the main control unit 500 via the network, for example. The online recognition result storage unit 683 receives the word character string and the word image ID selected as the recognition result when the automatic recognition has been normally performed from the main control unit 500 and stores it.

  Further, as described above, the VCS 640 generates a correct recognition result of a word that has not been automatically recognized described on the paper sheet 1 based on an operation input by the operator. The VCS 640 outputs the correct word recognition result (correct answer) and the word image ID indicating the word image of the correct word to the VCS result storage unit 684. The VCS result storage unit 684 stores the correct word recognition result (correct answer) supplied from the VCS 640 and the word image ID.

  Note that the names of the above storage units are for convenience, and may be a memory on a computer, or each may be stored separately as a file.

  The simulation unit 685 stores the evaluation value stored by the first evaluation value storage unit 681, the evaluation value stored by the second evaluation value storage unit 682, and the online recognition result storage unit 683. Based on the recognition result, the teaching result (correct answer) stored by the VCS result storage unit 684, and the word image ID stored in association with each evaluation value or teaching result, the correction parameter α and the correction Optimize parameter β. The simulation unit 685 includes an evaluation value (first evaluation value) stored in the first evaluation value storage unit 681 and an evaluation value (second evaluation value) stored in the second evaluation value storage unit 682. ) And the recognition result stored in the online recognition result storage unit 683 and the teaching result (correct answer) stored in the VCS result storage unit 684 are connected based on the word image ID, and are shown in FIG. Generate a list that looks like

FIG. 9 illustrates an example of processing of the simulation unit 685.
The simulation unit 685 associates the first evaluation value, the second evaluation value, the recognition result, and the teaching result with each word image ID, and generates an integrated list. In the example of each list in FIG. 9, the line indicated by “#” is a line that is commented out as a heading line.

  In the first evaluation value list stored by the first evaluation value storage unit 681, a pair of a word name registered in the database and an evaluation value for the word is associated with the word image ID for each word. Stored.

  For example, when there are 100 words registered in the DB, a pair of 100 words and an evaluation value is described after the ID number of the word image. Note that the configuration is the same as that of the first evaluation value list of the second evaluation value list stored in the second evaluation value storage unit 682.

  In addition, in the case where evaluation values are given to all words registered in the DB as in this case, the evaluation values are described in the order of DB registration even if the word “ham” is not described as a pair. These rules may be set in advance. Thereby, the description of the character string of the word registered in DB becomes unnecessary.

  Similarly, it is naturally possible to replace each registered word with an ID number by assigning an ID number instead of a word character string.

  In the example of the recognition result list (online recognition result list) stored by the online recognition result storage unit 683, the online recognition result is stored in association with the word image ID. “REJECT” shown in FIG. 9 indicates an automatic recognition result whose final recognition result is not selected.

  In the example of the teaching result list (VCS teaching result list) stored by the VCS result storage unit 684, the teaching result is stored in association with the word image ID. “No Process” shown in FIG. 9 indicates that the word of the word image is automatically recognized. In this case, since the correct answer is not input by the VCS 640, “No Process” is obtained. In this case as well, an ID number may be associated with the word image ID for each registered word instead of the character string.

  The integrated list includes a word image ID, a correct word name, and an evaluation value for each DB registered word in association with each other. The evaluation value for each DB registration word is a set of a DB registration name, a first evaluation value, and a second evaluation value.

  For example, in the word image ID “000001”, the online recognition result is “ham”. That is, the correct word name of the word image ID “000001” is “ham”. The word image ID “000001” is associated with the first evaluation value and the second evaluation value for each DB registered word.

  The word image ID “000002” has an online recognition result “REJECT”. For this reason, the VCS teaching result, not the online recognition result, is used as the correct word name. That is, the correct word name of the word image ID “000002” is “gotorgorg”. The word image ID “000002” is associated with the first evaluation value and the second evaluation value for each DB registered word.

  The simulation unit 685 identifies the correct word name for each word image ID, extracts the first evaluation value and the second evaluation value for each DB registered word, and associates them. As a result, the simulation unit 685 creates an integrated list.

  This integrated list may be created by a system different from the system that actually processed mail items, as long as it is a list necessary for creating the integrated list. The list itself may be received from a plurality of processing systems. But it is possible. In addition, for the sake of simplification, an example of an integrated list has been described in a format output to a single file. However, in reality, information may be virtually connected based on a word image ID on a program. no problem.

  Furthermore, the simulation unit 685 simulates the relationship between fluctuations in the correction parameters α and β and the integrated evaluation value after correction based on the integration list. That is, the simulation unit 685 substitutes values as the correction parameters α and β in Equation 4. The simulation unit 685 calculates an integrated evaluation value based on the mathematical formula 4 in which values are substituted into the correction parameters α and β, the first evaluation value, and the second evaluation value. The simulation unit 685 identifies one word based on the calculated integrated evaluation value. For example, the simulation unit 685 identifies the word for which the highest integrated evaluation value has been calculated. That is, the specified word is a recognition result by simulation.

  A correct word name is associated with each word image ID in the integrated list. The simulation unit 685 checks whether the word specified based on the integrated evaluation value matches the correct word name. The simulation unit 685 generates and outputs a display screen based on the fluctuations in the correction parameters α and β and the collation result. For example, the simulation unit 685 outputs the generated screen to the display 800 and displays it.

  FIG. 10 shows an example of a simulation result. The simulation result includes a word image ID, a correct word name, an integrated evaluation value calculated by the simulation, a simulation recognition result, and a matching result.

  10 shows a simulation result when the correction parameter α is “1.0” and the correction parameter β is “1.0”. The lower part of FIG. 10 shows a simulation result when the correction parameter α is “1.0” and the correction parameter β is “0.5”.

  As the correction parameters α and β, values generated based on an operation input by the operation unit 700 may be used. Further, values input by the input / output unit 900 may be used as the correction parameters α and β. Further, as the correction parameters α and β, values stored in advance in a memory in the paper sheet processing apparatus 100 may be used.

  The simulation by the simulation unit 685 may be configured to perform batch processing while varying the correction parameters α and β according to a preset variation pattern. The simulation by the simulation unit 685 may be configured to determine appropriate values while varying the correction parameters α and β according to a variation pattern set in advance for each batch process.

  The simulation unit 685 calculates an integrated evaluation value by simulation from a pre-correction evaluation value of each registered word for a certain word image, and specifies a word as a recognition result. As a result, the simulation unit 685 can display the difference between the recognition result and the correct answer on the display screen.

  The simulation unit 685 can calculate the recognition rate with the specified correction parameters α and β by performing the above-described simulation on all word images. The simulation unit 685 can calculate the recognition rates of the corrected correction parameters α and β by changing the correction parameters α and β. For example, the simulation unit 685 calculates a recognition rate for each set of the correction parameter α and the correction parameter β.

  The parameter selection unit 686 selects a specific value from the correction parameters α and β that have been changed as described above. For example, the parameter selection unit 686 employs the one having the best recognition rate as the correction parameters α and β. That is, the parameter selection unit 686 selects a set for which the best recognition rate is calculated from among a set of a plurality of correction parameters α and correction parameters β.

  For example, the parameter selection unit 686 generates a screen based on the difference between the recognition result and the correct answer, the recognition rate for each of the correction parameters α and β, and displays the screen on the display unit 800 as described above. Further, the parameter selection unit 686 may be configured to select a set of one correction parameter α and correction parameter β based on an operation input from the operation unit 700.

  However, when the correction parameter is automatically changed, it is assumed that the parameter diverges. For this reason, it is desirable to finally determine whether or not the operator adopts the adjustment result by GUI or the like.

  In this example, the simulation unit 685 performs the simulation with the prior probability set to a constant value (1.0). However, when the distribution of prior probabilities differs for each word, the integrated evaluation value of each word changes according to the prior probability as shown in Equation 4. As a result, the recognition result in the simulation varies.

  Through the processing described above, the learning unit 680 can identify the optimal correction parameters α and β. The learning unit 680 outputs the correction parameters α and β selected by the parameter selection unit 686 and supplies the correction parameters α and β to the recognition unit 660.

  The recognition unit 660 stores the supplied correction parameter α in the first correction parameter storage unit 662. The recognizing unit 660 stores the supplied correction parameter β in the second correction parameter storage unit 664. Thereby, the recognition unit 660 can use the correction parameters α and β specified by the learning unit 680 for the recognition process. As a result, it is possible to provide a word recognition device that can recognize words with higher accuracy, a word recognition program, and a paper sheet processing device including the word recognition device.

  In the above-described embodiment, the learning unit 680 has been described to automatically select and output the correction parameters α and β, but is not limited to this configuration. The learning unit 680 displays the identified correction parameters α and β as candidates on the display unit 800 and the like, and determines whether to update the correction parameters α and β based on the operation input input to the operation unit 700. There may be. That is, the learning unit 680 determines whether to output the specified correction parameters α and β to the recognition unit 660 based on an operation input input to the operation unit 700. This can prevent the correction parameters α and β confirmed by the operator from being set.

  Further, the learning unit 680 may be configured to take a backup of the correction parameters α and β already set in the recognition unit 660 before outputting the specified correction parameters α and β to the recognition unit 660. Further, this backup may be performed by the recognition unit 660 before overwriting the correction parameters α and β.

  Further, the update of the correction parameters α and β does not necessarily have to be distributed to one system as in the case of data collection, and the same correction parameters α and β are distributed and updated to a plurality of systems. There may be. In this case, the paper sheet processing apparatus 100 supplies the correction parameters α and β to other paper sheet processing apparatuses connected via the network.

  In the above-described embodiment, the learning unit 680 performs the simulation based on the first evaluation value and the second evaluation value and specifies the correction parameters α and β. However, the configuration is not limited to this configuration. The learning unit 680 may be configured to learn the correction parameters α and β based on the word image itself.

  When learning the correction parameters α and β based on the word image itself, it is necessary to store the word image itself. For this reason, a load is applied in terms of capacity both in terms of communication and file storage. Moreover, since it is necessary to redo the calculation of the evaluation value, it is disadvantageous in processing time at the time of learning.

  On the other hand, when learning the correction parameters α and β based on the first evaluation value and the second evaluation value, the correct word name is specified based on the recognition result before correction. For this reason, there is a possibility that a misread is handled as a correct word name. In addition, since the operator cannot confirm the word image when approving the correction parameter after simulation, there is a problem that it is uncertain whether or not the correct word name is incorrect.

  When a simulation is performed using a word image, the word image is stored regardless of the process during recognition. That is, the word image extracted by the word extraction unit 602 is stored in the memory in the paper sheet processing apparatus 100. In this case, the stored word image is displayed on the display unit 800. The operator confirms the word image displayed on the display unit 800 and inputs the correct word name using the operation unit 700. The learning unit 680 generates a correct word name based on the operation input input by the operation unit 700.

  The simulation unit 685 can calculate the recognition rate for each set of correction parameters by causing the word image stored in the memory to be recognized using the set of correction parameters used in the recognition process. .

  The simulation unit 685 uses either the automatically recognized recognition result or the VCS teaching result as a correct word name necessary for the simulation.

  The simulation unit 685 performs simulation by changing the correction parameter. Thereby, as described above, the result of matching between the recognition result based on the simulation result and the correct word name is displayed on the screen. Thereby, the operator can confirm the final recognition result from which the difference came out. Furthermore, the simulation unit 685 can refer to the word image from the difference image list based on the operation input. As a result, the simulation unit 685 can cause the operator to check the word image having the difference. As a result, the simulation unit 685 can cause the operator to check whether the correct answer is misread or whether the misread is correct.

  FIG. 11 shows an example of a simulation result. In this example, the simulation unit 685 displays a button (SIM difference image) for displaying a word image.

  According to the example in FIG. 11, “Image 1” has the original result (recognition result by the recognition unit 660) “REJECT”. In this case, the correct word name input by the VCS 640 has already been input. For this reason, there is no need to change the correct word name. The new result (recognition result by simulation) is correctly recognized. For this reason, the determination changes from “REJ” to “correct answer”.

  On the other hand, “image 3” is different in the original result and the new result. The original result and the new result are calculated using different correction parameters. For this reason, it is necessary for the operator to determine which recognition result is correct.

  As described above, the simulation unit 685 displays a button (SIM difference image) for displaying a word image. For example, when “image 3” is clicked, the simulation unit 685 causes the display unit 800 to display a corresponding word image. Further, the simulation unit 685 causes the operator to input a correct word name by using the operation unit 700, whereby the simulation unit 685 can acquire the correct word name.

  In this way, the simulation unit 685 causes the operator to refer to other word images except the word image in which the correct word name is taught by the VCS 640, and causes the correct word name to be input. As a result, the simulation unit 685 fills in the “correct answer” column. As a result, the simulation unit 685 can create a correct list with no misreading with respect to a list in which a difference is found in the recognition result due to a change in the correction parameter.

  In addition, the simulation unit 685 indicates increase / decrease in the total number of recognitions and the number of misreads according to the operation of the aggregation button. Further, the simulation unit 685 determines whether or not to apply the corrected correction parameter based on the operation of the “OK button” and the “NG button”.

  The simulation unit 685 displays an increase / decrease in the total number of recognitions and the number of misreads on the display unit 800, and allows the operator to determine whether the corrected correction parameter is effective. The simulation unit 685 approves the change of the correction parameter when the “OK button” is operated. That is, the parameter selection unit 686 selects the correction parameters α and β used for the simulation. Further, the simulation unit 685 rejects the change of the correction parameter when the “NG button” is operated.

  When humans make judgments including misreading, it is necessary to make judgments according to the occasion. For this reason, the simulation unit 685 causes the display 800 to display a graph indicating the relationship between the fluctuation of the evaluation value correction parameter and the increase / decrease in the number of recognitions and the recognition rate, as shown in FIG. Further, the simulation unit 685 visualizes the increase / decrease in the number of misreads and the misread rate as a graph and displays the graph on the display 800. As a result, the simulation unit 685 can make the operator easily determine the effectiveness of the correction parameter.

  For example, the simulation unit 685 temporarily stores the value obtained by changing the correction parameter and the simulation result at that time, and graphs the stored simulation result when the graph is displayed, as shown in FIG. A graph can be displayed.

  The correction parameter α and the correction parameter β approved by the above processing are output to the recognition unit 660. That is, the learning unit 680 outputs the correction parameters α and β selected by the parameter selection unit 686 and supplies the correction parameters α and β to the recognition unit 660. As a result, the correction parameter is updated.

  According to such a configuration, the word recognizing unit 600 can use an analytical technique (analytic matching) and an overall technique (global matching) in combination. Furthermore, correction can be performed using appropriate correction parameters for the analytical technique (analytic matching) and the overall technique (global matching). Thereby, the word recognizing unit 600 can calculate the posterior probability P (w | Y, X) with higher accuracy. As a result, it is possible to provide a word recognition apparatus, a word recognition program, and a paper sheet processing apparatus including a word recognition apparatus that can recognize words with higher accuracy.

  In the above embodiment, either the analytical matching or the overall matching may be performed first. In addition, when the word recognition unit 600 includes a configuration capable of processing the analytical matching and the global matching in parallel, the configuration may be such that the analytical matching and the global matching are processed in parallel.

  In the above embodiment, when the word recognition unit 600 specifies one word as the recognition result, the main control unit 500 uses the word for which the highest posterior probability P (w | Y, X) is calculated as the recognition result. However, the present invention is not limited to this configuration. Since the feature probability P (X) is constant when the word images are the same, the word recognition unit 600 calculates the posterior probability P (w | Y, X) with P (X) in Equation 1 as an arbitrary value. It may be configured to.

  In the embodiment described above, the word recognition unit 600 determines the posterior probability based on the posterior probability ratio P (Y | w) / P (Y), the likelihood P (X | w), and the prior probability P (w). Although described as a configuration for calculating P (w | Y, X), it is not limited to this configuration. For example, when it is not necessary to consider the prior probability or when the prior probability is a constant value, the word recognition unit 600 ignores the prior probability P (w) of Equation 1 or replaces it with a predetermined value, and the posterior The probability P (w | Y, X) may be calculated.

  In the above-described embodiment, the word recognition unit 600 has been described as a configuration in which the calculations of Formulas 1 to 4 are performed as they are, but the configuration is not limited thereto. The word recognition unit 600 may be configured to calculate by taking the logarithm of each term of Equations 1 to 3. By using the logarithm in this way, the part that was a multiplication is replaced with addition. Also, the part that was the division is replaced with subtraction.

  It should be noted that the functions described in the above embodiments are not limited to being configured using hardware, but can be realized by causing a computer to read a program describing each function using software. Each function may be configured by appropriately selecting either software or hardware.

  Note that the present invention is not limited to the above-described embodiment as it is, and can be embodied by modifying the constituent elements without departing from the scope of the invention in the implementation stage. Further, various inventions can be formed by appropriately combining a plurality of constituent elements disclosed in the embodiment. For example, some components may be deleted from all the components shown in the embodiment. Furthermore, you may combine suitably the component covering different embodiment.

  DESCRIPTION OF SYMBOLS 1 ... Paper sheets, 100 ... Paper sheet processing apparatus, 200 ... Supply part, 210 ... Separation roller, 220 ... Conveyance path, 300 ... Sorting processing part, 400 ... Image reading part, 500 ... Main control part, 600 ... Word Recognizing unit, 601 ... Image receiving unit, 602 ... Word extracting unit, 603 ... Character candidate extracting unit, 604 ... Character recognizing unit, 605 ... Feature extracting unit, 610 ... Analytical matching unit, 611 ... Character probability calculating unit, 612 ... First calculation unit, 613 ... second calculation unit, 620 ... global matching unit, 630 ... feature probability calculation unit, 640 ... VCS, 641 ... first word image storage unit, 642 ... model learning unit, 643 ... Model storage unit, 644 ... Word model generation unit, 645 ... Word dictionary, 651 ... Prior probability calculation unit, 652 ... Prior probability storage unit, 653 ... Prior probability input unit, 660 ... Recognition unit, 661 ... First correction evaluation value Calculation Unit, 662... First correction parameter storage unit, 663... Second correction evaluation value calculation unit, 664... Second correction parameter storage unit, 665... Integrated evaluation value calculation unit, 680. Evaluation value storage unit, 682 ... second evaluation value storage unit, 683 ... online recognition result storage unit 683, 684 ... VCD result storage unit, 685 ... simulation unit, 686 ... parameter selection unit, 700 ... operation unit, 800 ... Display unit, 900... Input / output unit.

Claims (10)

A word dictionary for storing multiple words,
An image receiving means for receiving an image including a word;
Word image extraction means for extracting a word image for each word from the image;
Character candidate extraction means for extracting character candidates from the word image;
An analytical matching means for recognizing the character candidates and calculating a first evaluation value for each word stored in the word dictionary;
An overall matching means for calculating a feature probability that a feature appears, and calculating a second evaluation value based on the word image, a word model for each word stored in the word dictionary, and the feature probability;
Correction means for correcting the first evaluation value with a first correction parameter set in advance, and correcting the second evaluation value with a second correction parameter set in advance;
Integrated evaluation value calculating means for calculating a third evaluation value by integrating the corrected first evaluation value and the corrected second evaluation value;
Output means for outputting the third evaluation value calculated by the integrated evaluation value calculating means;
A word recognition device comprising:   The word recognition device according to claim 1, wherein the correction unit raises the first evaluation value to a power by the first correction parameter and raises the second evaluation value to a power by the second correction parameter.   3. The word according to claim 2, wherein the correction unit corrects the first evaluation value with the first correction parameter as a logarithm and corrects the second evaluation value with the second correction parameter as a logarithm. Recognition device. Recognition means for recognizing a word based on the third evaluation value;
Teaching means for generating a correct word name of a word whose word is not specified by the recognition means based on an operation input;
Storage means for storing the word image and the recognition result or the correct word name by the recognition means in association with each other;
Simulation means for changing the first correction parameter and the second correction parameter and recognizing a word based on the word image and the changed first correction parameter and second correction parameter. When,
A correction parameter selection means for collating the recognition result by the simulation means with the recognition result by the recognition means or the correct word name, and specifying the first correction parameter and the second correction parameter;
The word recognition device according to claim 2, comprising: Recognition means for recognizing a word based on the third evaluation value;
Teaching means for generating a correct word name of a word whose word is not specified by the recognition means based on an operation input;
Storage means for storing the first evaluation value, the second evaluation value, the recognition result by the recognition means or the correct word name in association with each other;
The first correction parameter and the second correction parameter are changed, and the first evaluation value and the second evaluation value, and the changed first correction parameter and second correction parameter are changed. Simulating means for recognizing words based on
A parameter selection means for collating the recognition result by the simulation means with the recognition result by the recognition means or the correct word name, and specifying the first correction parameter and the second correction parameter;
The word recognition device according to claim 2, comprising: The simulation means collates the recognition result by the simulation means with the recognition result by the recognition means or the correct word name, and outputs a collation result,
The parameter selection means specifies the first correction parameter and the second correction parameter based on an operation input.
The word recognition device according to claim 4 or 5.   The simulation means outputs the word image when the recognition result by the simulation means does not match the recognition result by the recognition means or the correct word name, and is stored by the storage means based on an operation input. The word recognition apparatus according to claim 4, wherein the recognition result by the recognition means is rewritten.   The word recognition device according to claim 4, wherein the parameter selection unit outputs the specified first correction parameter and the second correction parameter via a network. A word recognition program executed in a word recognition device comprising a word dictionary for storing a plurality of words and an image receiving means for receiving an image containing the word, wherein the word recognition program
Word image extraction means for extracting a word image for each word from the image;
Character candidate extraction means for extracting character candidates from the word image;
An analytical matching means for recognizing the character candidates and calculating a first evaluation value for each word stored in the word dictionary;
An overall matching means for calculating a feature probability that a feature appears, and calculating a second evaluation value based on the word image, a word model for each word stored in the word dictionary, and the feature probability;
Correction means for correcting the first evaluation value with a first correction parameter set in advance, and correcting the second evaluation value with a second correction parameter set in advance;
Integrated evaluation value calculating means for calculating a third evaluation value by integrating the corrected first evaluation value and the corrected second evaluation value;
Output means for outputting the third evaluation value calculated by the integrated evaluation value calculating means;
And make it work. A capturing means for capturing paper sheets;
Conveying means for conveying the paper sheet;
Image reading means for reading an image including a word on the paper sheet;
A word dictionary for storing multiple words,
An image receiving means for receiving an image including a word;
Word image extraction means for extracting a word image for each word from the image;
Character candidate extraction means for extracting character candidates from the word image;
An analytical matching means for recognizing the character candidates and calculating a first evaluation value for each word stored in the word dictionary;
An overall matching means for calculating a feature probability that a feature appears, and calculating a second evaluation value based on the word image, a word model for each word stored in the word dictionary, and the feature probability;
Correction means for correcting the first evaluation value with a first correction parameter set in advance, and correcting the second evaluation value with a second correction parameter set in advance;
Integrated evaluation value calculating means for calculating a third evaluation value by integrating the corrected first evaluation value and the corrected second evaluation value;
Recognition means for recognizing a word based on the third evaluation value;
Sorting processing means for sorting the paper sheets based on the words recognized by the recognition means;
A paper sheet processing apparatus comprising:

Images