JP2002189983A - Character recognizing device and delivery processor - Google Patents

Abstract

PROBLEM TO BE SOLVED: To shorten the entire adjustment period of the dictionary or the like of a recognizing part by analyzing more delivery data with high accuracy in a short time. SOLUTION: Misreading adjustment that has been conducted manually by a site coordinator so far is electronized by distributing the same VCS image to a plurality of VCD operators and combining results obtained by majority decision with the image by utilizing a VCS system.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a character recognition device for recognizing location information described in a delivery, and a delivery processing device for executing a sorting process based on a result recognized by the character recognition device. .

[0002]

2. Description of the Related Art In general, a delivery processing apparatus has been put to practical use that performs sorting processing based on the recognition result of location information (address information, address) described in a delivery or the like. The recognition result is output based on the similarity obtained by comparing a character pattern of each character obtained from a captured image of a delivery with a reference character pattern. Such a device is provided with a video coding system for displaying a delivery description screen and inputting location information or a code number by manual input of an operator when location information cannot be read by the recognition unit. .

In the delivery processing apparatus as described above, the dictionary of the recognition unit is adjusted when the apparatus is loaded and installed. For example, an adjustment is made so that recognition can be made for a place name specific to the installation area. That is,
One of the items of the installation adjustment is adjustment of the recognition level of the character dictionary. At present, these misreads are reduced by checking the contents of the deliveries manually sorted, extracting deliveries with many misreadings, and adjusting the character dictionary.

[0004] For example, one person in charge is 100
The non-recognized characters are discriminated from the location information of the rejected deliveries as a result of the classification by recognizing the zero deliveries, and the determination results are totaled. Based on the cumulative result, the weight for determining the recognition result for this character is changed. For example,
When the captured character is "dog" and the recognition result is "large" and the input character is "dog", the similarity between the captured character "dog" and the dictionary "large" is compared with the dictionary "dog" rather than the similarity. The weighting is changed so that the similarity increases.

[0005] As described above, since the erroneous classification check is performed manually, there is an oversight due to visual omission, and the like.
It is time-consuming and inefficient. In addition, the longer adjustment time delays the date and time of delivery of the sorting machine to the customer, which has the disadvantage of increasing costs for both the manufacturer and the user. As described above, misreading adjustments that were previously performed manually by the local coordinator can be shortened by analyzing the data of more deliveries in a short period of time with high accuracy, thereby shortening the overall adjustment period Things are desired.

[0006]

SUMMARY OF THE INVENTION According to the present invention, a misreading adjustment to be performed on a recognition device for recognizing location information of a delivery used in a delivery processing device can be performed with high accuracy by using more delivery data. The purpose is to make it possible to shorten the overall adjustment period by analyzing in a short time.

[0007]

A recognition device according to the present invention comprises:
An image pickup means for picking up an image of a delivery in which the location information is described; and recognition recognition by comparing each character of the location information from the image taken by the image pickup means with a dictionary having a reference character pattern. Means, and input means for inputting one character at a time for the location information of the delivery corresponding to the output of the image of the delivery taken by the imaging means,
Determining means for determining whether or not the input character of the location information by the input means and the result of recognizing the location information by the recognizing means coincide with each other; Changing means for changing the weight at the time of recognition by the recognizing means for the adjustment target character determined based on the result of the mismatch between the input character and the recognition result of the character.

[0008] The delivery processing apparatus of the present invention comprises a transporting means for transporting a delivery in which location information is described, an imaging means for capturing an image of the delivery transported by the transporting means, and an image capturing means. Recognition means for recognizing each character of the location information one by one from a captured image with a dictionary having a reference character pattern, and transported by the transport means based on the location information recognized by the recognition means. First processing means for sorting the deliverables, and when the location information cannot be recognized by the recognition means, the location information of the deliverables is stored in correspondence with the image of the deliverable imaged by the imaging means. Based on the input video coding system and the location information input by the video coding system, the delivery carried by the carrier is sorted and processed. Second processing means, input means for inputting, one by one, characters of the location information of the delivery corresponding to the output of the image of the delivery imaged by the imaging means, and the location information by the input means. Determining means for determining whether or not the input character of the location information and the result of recognizing the location information by the recognizing means match; and if the determination result by the determining means does not match, the recognition result for the location information and the input character Changing means for changing the weight at the time of recognition by the recognizing means on the adjustment target character determined based on the content of the mismatch.

[0009]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A delivery processing apparatus according to an embodiment of the present invention will be described below with reference to the drawings. 1 and 2, the delivery processing apparatus includes a reading sorting machine 1, a video coding system (VCS) 2, and an information processing terminal 3 that manages the reading sorting machine 1.

The information processing terminal 3 is composed of a personal computer (PC) or the like, and has an external P with address reading dictionary adjustment function.
Used as C (delivery comprehensive cooperation PC). The information processing terminal 3 includes, as shown in FIG. 3, a recognition result table 3a for storing a recognition result by an after-mentioned recognition unit 15 in the reading and sorting machine 1 and an ID number, and a video coding system as shown in FIG. If the destination information input by the (VCS) 2 and the recognition result by the recognition unit 15 described later do not match, there is a mismatch table 3b for storing the destination information, the recognition result, and the ID number.

[0011] On a delivery (postal matter) P handled by the delivery processing apparatus, location information such as an address for delivery is described. A barcode (ID barcode) assigned to specify the delivery P is described (printed) on the delivery P.

The address is composed of all or a part of the name of a prefecture, a name of a city, a town, a village, or the like.

The reading / sorting machine 1 reads the address written on the delivery P, recognizes the address from the read content,
The sorting machine sorts the deliveries P based on the address recognition result or predetermined sorting information (sort box number) corresponding to the code (address information) from the video coding system 2. The corresponding barcode is printed.

In the reading sorting machine 1, for example, a plurality of deliveries are set in batches (lots) or in deliveries rejected for the batches, and the set deliveries P,. A supply unit 10 for supplying, a scanner unit (imaging means) 11 for reading an entire image on one delivery P supplied by the supply unit 10, and a delivery P from the delivery P passing through the scanner 11 ID barcode reading unit 12 serving as a photoelectric conversion unit that reads a barcode (ID barcode) assigned to specify the ID, and corresponds to the image number on the delivery P that has passed through the ID barcode reading unit 12. A bar code printer unit 13 for printing the second ID bar code, and a delivery result that has passed through the bar code printer unit 13 is recognized as an address recognition result or a video codec. Or classified on the basis of the address information from the ring system 2, division section 1 divides the deliveries P that could not be recognized in Athena reject box into (not shown)
4, a recognition unit 15 for recognizing a destination from an image from the scanner unit 11, and a control unit (judgment means, processing means) 16 for controlling the above-mentioned units.

The supply section 10, the scanner section 11, the ID bar code reading section 12, the bar code printer section 13, and the sorting section 14 are provided in the transport section main body 1a.

The scanner unit 11 has an image sensor (not shown) as a photoelectric conversion element, and this image sensor senses the delivery P through an image window (not shown).

The recognizing section 15 has a dictionary 15a in which reference character patterns are stored. The character pattern of each character obtained from the picked-up image of the delivery P and the dictionary 1a are stored in the dictionary 15a.
A character having a high degree of similarity is output as a recognition result based on the degree of similarity obtained by comparison with the reference character pattern of 5a.

The video coding system 2 includes an image storage and distribution device 21, a video coding desk (VCD)
,... Are constituted by supervisor terminals 23. The image storage / distribution apparatus 21, video coding desk (VCD) 22,..., Supervisor terminal 23 are connected via a communication line such as Ethernet (registered trademark).

The image storage / distribution device 21 comprises a large-capacity storage device 24 and a distribution unit 25, and the supervisor terminal 2
3 is controlled. The supervisor terminal 23
The image of the delivery P for which the address cannot be recognized from the recognition unit 15 is temporarily stored in the large-capacity storage device 24.

Further, the supervisor terminal 23 distributes the image of the delivery P, whose address of the mass storage device 24 could not be recognized, to the video coding desk (VC
D) distribution to 22,.

The large-capacity storage device 24 stores an image of the delivery P supplied from the recognizing unit 15 whose address cannot be recognized. The mass storage device 24 is the same as that shown in FIG.
As shown in the video coding desk (VCD) 2
Result table 24a that stores the picked-up image and ID number of the delivery P corresponding to the address information input in Step 2
have. The distribution unit 25 distributes the image of the delivery P, whose address has not been recognized, read from the mass storage device 24 to the video coding desks (VCD) 22,.

Video coding desk (VCD) 2
Are composed of an image display unit 6a such as a CRT and an input unit 22b such as a keyboard.
4 is displayed on the image display unit 22a, and the address information of the displayed image is input by the operator to the input unit 22a.
"b" is input. This input section 22b
Is input to the control unit 16 via the supervisor terminal 23 together with the image number. Further, the supervisor terminal 23 is connected to the video coding desk (V
CD) 22,... Are returned to the control unit 16.

Next, in the above-described configuration, when the delivery processing apparatus is installed, the adjustment processing of the dictionary 15a of the recognition unit 15 of the reading / sorting machine 1 will be described with reference to flowcharts shown in FIGS. It will be described with reference to FIG.

That is, when the delivery processing device is installed, the coordinator sets the character dictionary recognition level adjustment mode, which is an item of installation adjustment, on the information processing terminal 3 (ST1). Based on this setting, the information processing terminal 3 notifies each unit of the recognition level adjustment mode of the character dictionary (ST
2). Further, the coordinator sets the actual delivery P divided at the installation location in the supply unit 10 and instructs the start of the process (ST3).

According to this instruction, the delivery P from the supply unit 10 is conveyed, so that one delivery P is imaged (read) by the scanner unit 11 (ST4). Thereby, the control unit 16 assigns an ID number to each of the captured images of the delivery P by the scanner unit 11, and combines the captured image of the delivery P and the ID number with a video coding system (VCS). 2 (ST5). Further, the picked-up image of the delivery P is recognized by the recognition unit 15 (ST6). That is, the recognizing unit 15 recognizes a character having a high similarity based on the similarity obtained by comparing the character pattern of each character of the address portion obtained from the captured image of the delivery P with the reference character pattern of the dictionary 15a. Is output to the control unit 16. As a result, the control unit 16 adds the ID number to the recognition result for the address part obtained from the captured image of the delivery P and outputs the result to the information processing terminal 3 (ST7).

The information processing terminal 3 registers the supplied recognition result and ID number in the recognition result table 3a (S
T8). The above video coding system (VCS) 2
The supervisor terminal 23 stores the picked-up image of the delivery P distributed from the reading / sorting machine 1 in the large-capacity storage device 24 in the image storage / distribution device 21 in association with the ID number (ST).
9).

Next, under the control of the supervisor terminal 23, the distribution unit 25 converts the picked-up images (same images) of the delivery P stored in the large-capacity storage device 24 in the image storage and distribution device 21 into two images. Video coding desk (VC
D) Distribution (distribution) to 22,... (ST10).

As a result, the two video coding desks (VCD) 22, 22 are connected to the image display unit 6 such as a CRT.
a to display the picked-up image of the delivery P (ST1).
1). Based on this display, the operator of each video coding desk (VCD) 22, 22 inputs address information through the input section 22b (ST12).

Based on this input, the supervisor terminal 2
3 is two video coding desks (VCD) 2
It is determined whether the answers from 2, 22 match (ST
13). As a result of this judgment, if the answers from the two video coding desks (VCD) 22 match,
The supervisor terminal 23 registers the captured image of the delivery P and the ID number corresponding to the input address information in the input result table 24a of the mass storage device 24 (ST1).
4).

If the answers from the two video coding desks (VCDs) 22 do not match as a result of the determination in step 13, the supervisor terminal 23
The picked-up image of the same delivery P is delivered to the second VCD 22 (ST15). That is, the supervisor terminal 23 transfers the picked-up image (the same image) of the delivery P stored in the large-capacity storage device 24 in the image storage and distribution device 21 to the third video coding desk (VCD) by the distribution unit 25. 22
Distributed (distributed) to

Thus, the third video coding desk (VCD) 22 displays the picked-up image of the delivery P on the image display unit 6a such as a CRT (ST16).
Based on this display, the operator of the third video coding desk (VCD) 22 inputs address information through the input section 22b (ST17).

At this time, it is determined whether or not the address information by the third video coding desk (VCD) 22 matches any of the address information by the two video coding desks (VCD) 22 (ST1).
8). As a result of this determination, when the address information by the third video coding desk (VCD) 22 matches any of the address information by the two video coding desks (VCD) 22, the supervisor terminal 23
The captured image of the delivery P and the ID number corresponding to the address information are registered in the input result table 24a of the mass storage device 24 (ST19).

If the result of the determination in step 18 is that there is a mismatch, the supervisor terminal 23 sets another V
The captured image of the same delivery P is distributed to the CD 22 (ST2).
0). That is, the supervisor terminal 23 transfers the captured image (the same image) of the delivery P stored in the large-capacity storage device 24 in the image storage and distribution device 21 to another video coding desk (VCD) 22 by the distribution unit 25. Distribute (distribute). At this time, the same video coding desk (V
CD) 22 can receive the same image.

Thus, the video coding desk (VCD) 22 to which the same image has been distributed displays the picked-up image of the delivery P by the image display unit 6a such as a CRT (ST21). Based on this display, the operator of each video coding desk (VCD) 22 inputs address information through the input section 22b (ST22).

At this time, each video coding desk (V
CD) 22,..., It is determined whether or not there is a majority of matching destination address information input (ST23).
In this manner, the same image is repeated until the address information is obtained by the majority decision, and the video coding desk (VC
D) Distribution to 22,. As a result of this judgment, if the address information is obtained by majority vote, the supervisor terminal 23
Proceeds to step 19, and stores the destination image information and the corresponding picked-up image of the delivery P and the ID number in the large-capacity storage device 2.
4 in the table 24a.

As a result of the determination in step 23, when the address information is not obtained by the majority decision, the supervisor terminal 23 determines whether or not a predetermined number of distributions have been performed (S
T24). As a result of this determination, the supervisor terminal 23
If the predetermined number of distributions have not been performed, the process returns to step 20. If the predetermined number of distributions have been performed, the captured image of the delivery P is excluded from the processing.

Further, under the control of the supervisor terminal 23, the picked-up image (same image) of the delivery P stored in the large-capacity storage device 24 in the image storage / distribution device 21 is converted into two video images by the distribution unit 25. Coding desk (VC
D) The processing from step 10 to distribution (distribution) to 22,... To the step 25 is performed in parallel.

After collecting various information as described above, at the stage of starting the adjustment, the adjuster operates the information processing terminal 3 to instruct the start of the adjustment (ST26). Then, based on this instruction, the information processing terminal 3 instructs the supervisor terminal 23 to call the registered contents of the input result table 24a of the mass storage device 24 of the image storage / distribution device 21 (ST27).

Based on this instruction, the mass storage device 24
When the registered contents of the input result table 24a are notified,
The information processing terminal 3 compares each piece of destination information with the recognition result of the same ID number registered in the recognition result table 3a to determine whether they match (S).
T28).

If the result of this determination is that they match, the information processing terminal 3 returns to the determination processing for the next destination information.
If the result of determination in step 28 is a mismatch, the information processing terminal 3 newly stores the read result and the input result for the ID number in the mismatch table 3b (ST2).
9).

When the comparison between the registered contents of the input result table 24a and the registered contents of the recognition result table 3a is completed (ST30), the information processing terminal 3 reads the data stored in the mismatch table 3b by the display unit. A situation where the result and the input result do not match is displayed on the display unit (ST31). With this display, the adjuster can grasp the recognition state of the recognition unit 15.

In this case, the read image or the character pattern image at the time of recognition can also be supplied from the large-capacity storage device 24 to the information processing terminal 3 and displayed in comparison with the read result and the input result.

As a result, the coordinator instructs the dictionary adjustment based on the display contents of the information processing terminal 3. For example, the weight for determining the recognition result for a predetermined character is changed. For example, when the imaged character is "dog" and the recognition result is "large" and the input character is "dog", the similarity between the imaged character "dog" and the dictionary "dog" is smaller than the similarity obtained by comparing the dictionary with "large". The weighting is changed so that the similarity by comparison increases.

Further, the image storage / distribution device 21 may delete the image corresponding to the ID number which is not in the mismatch table 3b at this stage. The coordinator can perform operations such as referencing and sorting information in the information processing terminal 3,
Further, the image corresponding to the ID number in the mismatch table 3b can be called from the image storage / distribution device 21 and referenced. The coordinator can operate the recognition unit 15 from the information processing terminal 3 with this information to make adjustment. Further, the registered contents of the mismatch table 3b and the image with the ID number can be stored in a removable storage medium.

As described above, the delivery is supplied to the reading sorting machine, and all the images read at this time are transferred to the VCS. In VCS, the operator looks at the image of the delivery and taps the address information on the VCD. At this time, the same image is transmitted to two or more VCDs in order to prevent inconsistency between the image and the address information due to erroneous keying as much as possible. Use the keying result obtained by majority vote (V is required until majority is obtained)
Distribute to CD. The same operator may receive the same image more than once). The keying result is stored in the image storage / distribution device together with the image. Next, using an information processing terminal as an external PC with a destination name reading dictionary adjustment function,
The stored image is distributed to the recognition unit of the reading and sorting machine, and the recognition result of the recognition unit is compared with the input data of the corresponding image to adjust the readability of each character and the character dictionary is used. It is intended to be registered in.

Also, the ID number given to each delivery is used as a bridge, and the recognition result for the character to be recognized, the key input result,
A data collection in which images are combined is obtained. At this time, those having a mismatch between the recognition result and the keying input result are collected.

In this way, the character dictionary of the recognition unit obtains the characteristic similarity by comparing each character with the reference pattern, and uses this degree as a score parameter. By changing the predetermined score to be compared up or down,
When character recognition is adjusted, by displaying a collection of mismatched recognition results and keystroke input results,
It can help you identify trends in misrecognition.

Further, a collection of images having inconsistencies in the recognition result and the key input result and an image thereof are brought back to the factory and used for creating a character dictionary. By separating the image of each character one by one and manually inputting the correct recognition result of the image and registering it in the character dictionary, the number of recognizable character patterns can be increased.

[0049]

As described in detail above, according to the present invention, the misreading adjustment performed with respect to the recognition device for recognizing the location information of a delivery used in the delivery processing device can be performed with high accuracy and with more accuracy. By analyzing the delivery data in a short time, the entire adjustment period can be shortened.

[Brief description of the drawings]

FIG. 1 is a diagram showing a schematic configuration of a delivery processing apparatus for describing an embodiment of the present invention.

FIG. 2 is a diagram showing a schematic configuration of a delivery processing apparatus.

FIG. 3 is a diagram for explaining a storage example of a recognition result table.

FIG. 4 is a diagram for explaining a storage example of a mismatch table;

FIG. 5 is a diagram for explaining a storage example of an input result table.

FIG. 6 is a flowchart for explaining a dictionary adjustment process of the recognition unit of the reading / sorting machine.

FIG. 7 is a flowchart illustrating a dictionary adjustment process of a recognition unit of the reading / sorting machine.

FIG. 8 is a flowchart for explaining a dictionary adjustment process of the recognition unit of the reading / sorting machine.

[Explanation of symbols]

P: delivery, 1 ... read sorting machine, 2 ... video coding system, 3 ... information processing terminal, 3a ... recognition result table, 3b ... mismatch table, 10 ... supply unit, 11 ... scanner unit, 14 ... sorting unit, 15 ... Recognition unit, 15a ... Dictionary,
16 control unit, 21 image storage / distribution device, 22 video coding desk, 23 supervisor terminal, 24
Large-capacity storage device, 24a: input result table, 25: distribution unit.

Claims (7)

[Claims] 1. An image pickup means for picking up an image of a delivery in which location information is described, and comparing each character of the location information from an image picked up by the image pickup means with a dictionary having a reference character pattern. And input means for inputting, one by one, characters of the location information of the delivery corresponding to the output of the image of the delivery imaged by the imaging means, and the location information by the input means. Determining means for determining whether or not the input character of the location information and the result of recognizing the location information by the recognizing means match; and if the determination result by the determination means does not match, the recognition result for the location information and the input character Changing means for changing the weight at the time of recognition by the recognizing means with respect to the adjustment target character determined based on the content of mismatch of the character recognition device. . 2. An image pickup step of picking up an image of a delivery in which location information is described, and comparing each character of the location information from the image picked up by the image pickup step with a dictionary having a reference character pattern. A recognition step of recognizing, and an input step of inputting, one by one, characters of the location information of the delivery corresponding to the output of the image of the delivery imaged by the imaging step, and the location information by the input step. A determination step of determining whether or not the input character of the location information and the result of recognizing the location information in the recognition step match; and, if the determination result in the determination step does not match, the recognition result for the location information and the input character A changing step of changing a weight at the time of recognition in the recognition step for the character to be adjusted determined based on the content of the mismatch of the character. The method of adjustment. 3. A transporting means for transporting a delivery having location information described thereon, an imaging means for capturing an image of the delivery transported by the transporting means, and location information based on the image captured by the imaging means. A recognition unit that recognizes each character by comparing it with a dictionary having a reference character pattern, and a sorting process for sorting the deliveries conveyed by the conveyance unit based on the location information recognized by the recognition unit. When the location information cannot be recognized by the processing means of (1) and the recognition means,
A video coding system for inputting the location information of the delivery corresponding to the image of the delivery taken by the imaging means; and a video coding system for transporting the delivery based on the location information input by the video coding system. Second processing means for sorting the delivered goods, and input means for inputting, one by one, the location information of the delivered goods in correspondence with the output of the image of the delivered goods imaged by the imaging means. Judging means for judging whether or not the input character of the location information by the input means and the result of recognizing the location information by the recognizing means coincide with each other; A changer for changing the weighting of the adjustment target character determined by the recognition unit based on the content of the mismatch between the recognition result and the input character. A delivery processing apparatus comprising: a step; 4. An image pickup means for picking up an image of a delivery in which location information is described, an extraction means for extracting a character pattern for each character of the location information from an image picked up by the image pickup means, By comparing a character pattern of each character of the location information extracted by the means with a dictionary having a reference character pattern, a recognition means for recognizing, and an output of the image of the delivery imaged by the imaging means are provided. Input means for inputting the location information of the delivery one character at a time; and determining whether or not the input character by the input means and the result of recognition of the location information by the recognition means match in character units. A judging means; and a character pattern extracted by the extracting means in the dictionary of the recognizing means for the input character of the location information when the judgment results do not match. Character recognition apparatus characterized by comprising a processing means for adding. 5. An image pickup means for picking up an image of a delivery item on which location information is described, and comparing each character of the location information from the image picked up by the image pickup means with a dictionary having a reference character pattern. The first storage means stores the result of recognizing the location information by the recognition means together with the identification number given to the corresponding deliverable, and the image is taken by the imaging means. Input means for inputting, one by one, characters of the location information of the delivery corresponding to the output of the image of the delivery; input characters of the location information by the input means are given to the corresponding delivery; A second storage unit that stores the identification information together with the identification number; and a recognition result for the location information of the delivery to which the same identification number is assigned is read from the first storage unit. Information input characters are read from the second storage means, and a determination is made as to whether or not the input characters match. When the determination results by the determination means do not match, the recognition result for the location information and the input characters do not match. A character recognition device, comprising: changing means for changing the weight of the adjustment target character determined based on the content at the time of recognition by the recognition means. 6. An image pickup means for picking up an image of a delivery in which location information is described, and comparing each character of the location information from the image picked up by the image pickup means with a dictionary having a reference character pattern. , At least two input means for inputting, one by one, characters of the location information of the delivery corresponding to the output of the image of the delivery taken by the imaging means; Determining means for determining whether or not the input character of the location information whose input result matches the result of recognition of the location information by the recognition means; and determining whether the determination result by the determination means does not match. The weight of the character to be adjusted, which is determined based on the mismatch between the recognition result of the location information and the input character, is changed when the recognition unit recognizes the character to be adjusted. Character recognition apparatus characterized by comprising a stage, a. 7. An image pickup means for picking up an image of a delivery in which location information is described, and comparing each character of the location information from the image picked up by the image pickup means with a dictionary having a reference character pattern. , A plurality of input means for inputting one character at a time of the location information of the delivery corresponding to the output of the image of the delivery imaged by the imaging means, and the plurality of input means. Determining means for determining whether or not the input character of the location information determined by the majority decision of the input result according to the above and the result of recognizing the location information by the recognizing means coincide with each other; The weight of the character to be adjusted, which is determined based on the mismatch between the recognition result of the location information and the input character, is changed when the recognition unit recognizes the character to be adjusted. Character recognition apparatus characterized by comprising a stage, a.