JP2018171579A - Address recognition device, parcel automatic sorting system, address recognition method and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an address recognition device which reduces erroneous recognition and reads a sorting number efficiently from a label or the like attached to a parcel.SOLUTION: An address recognition device includes a code reading part which reads code information displayed on an image with address information of a parcel included, and reads a sorting number indicated by the code information. The code reading part detects information as a reference of a display position of the sorting number on the image with the address information included when the sorting number cannot be read.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a destination recognition device, a parcel automatic sorting system, a destination recognition method, and a program.

  Systems that automate the sorting of parcels such as courier services have been provided. For example, Patent Document 1 describes a sorting system in which a collected parcel is placed on a conveyor and conveyed, and a sorting code printed on a parcel label is read by a barcode reader and distributed to an appropriate collection box. Has been. Further, according to the sorting system described in Patent Document 1, even if the sorting code cannot be read from the bar code, the delivery classification information such as the sorting number, the postal code, or the address is read by a scanner, and OCR (Optical Character Recognition) is performed. ), It is possible to sort the packages transported on the conveyor online. In addition, when the delivery classification information cannot be recognized by OCR, the clerk can perform sorting by inputting the sorting number of the package from a VCD (Video Coding Disk) device.

JP 2013-220362 A

As described in Patent Document 1, for automatic sorting of parcels, it is necessary to read the sorting number and address by OCR. However, the destination recognition for parcels is compared with the destination recognition method for letter mail called postal OCR. There are the following difficulties.
1. When the size of the parcel is generally large, it has a width of 80 cm and a length of 150 cm, and the recognition target range is wider than that of letter mail (maximum B4 size). As a result, when searching for an address area from within the recognition target range, more address area candidates are raised, and the difficulty of obtaining a correct address area is high.
2. Parcels are packed in cardboard boxes, wrapping paper, and paper bags, and address labels are often affixed to them, and the background pattern of the address area is more complicated than envelopes of letter mail. Easy to detect.
3. The letter mail is sandwiched between belts and conveyed with almost no inclination. For this reason, an image captured by a scanner or the like has a small inclination of the area where the destination is displayed, and is almost in the normal position. However, in the case of a parcel, it is placed on the conveyor at a free angle with the display of the destination as the upper surface and is conveyed. Then, the upper surface of the parcel must be imaged with a scanner or the like from above, and the parcel position and the destination display area must be found and recognized in the image obtained by imaging a wide range, but the rotation angle in the plane direction is 360 °. There is a degree of freedom, making recognition difficult.
In this way, it is generally difficult to specify an area where a sorting number or address is displayed in an image obtained by picking up a parcel. Patent Document 1 does not describe a solution to such a problem.

  Accordingly, an object of the present invention is to provide a destination recognition device, a parcel automatic sorting system, a destination recognition method, and a program that solve the above-described problems.

  The present invention includes a code reading unit that reads code information displayed on an image including parcel destination information and reads a sorting number indicated by the code information, and the code reading unit reads the sorting number. When it cannot, it is a destination recognition device that detects information serving as a reference for the display position of the sorting number in the image.

  According to the present invention, the destination recognition device described above, a camera that captures an image including the destination information of the parcel, and outputs the captured image to the destination recognition device, and a sorting number read by the destination recognition device. A parcel automatic sorting system comprising: a sorting destination determining machine that determines a sorting destination based on the sorting destination.

  The present invention also includes a step of reading code information displayed on an image including parcel destination information and reading a sorting number indicated by the code information, and if the sorting number cannot be read, the sorting number in the image Detecting information serving as a reference for the display position of the destination.

  In the present invention, the computer of the destination recognition device reads the code information displayed on the image including the destination information of the parcel, and reads the sorting number indicated by the code information, when the sorting number cannot be read, It is a program for functioning as means for detecting information serving as a reference for the display position of the sorting number in the image.

  According to the present invention, it is possible to reliably recognize the destination information of a parcel displayed in various modes.

It is a figure which shows an example of the destination recognition apparatus by one Embodiment of this invention. It is a figure which shows an example of the package automatic sorting system by one Embodiment of this invention. It is a figure which shows the hardware structural example of the sorting destination recognition apparatus for parcels by one Embodiment of this invention. It is a figure which shows an example of the sorting destination recognition apparatus for parcels by one Embodiment of this invention. It is a flowchart which shows an example of the sorting destination recognition process for parcels by one Embodiment of this invention. It is a flowchart which shows an example of the barcode detection and reading process by one Embodiment of this invention. It is a flowchart which shows an example of the classification number reading process by one Embodiment of this invention. It is a 1st flowchart which shows an example of the large label identification process by one Embodiment of this invention. It is a 2nd flowchart which shows an example of the large label identification process by one Embodiment of this invention. It is a flowchart which shows an example of the tracking barcode utilization process by one Embodiment of this invention. It is a 1st figure which shows an example of the parcel image by one Embodiment of this invention. It is a 2nd figure which shows an example of the parcel image by one Embodiment of this invention. It is a block diagram which shows the minimum structure of the destination recognition apparatus by one Embodiment of this invention. It is the 1st figure explaining the destination information in a general parcel. It is a 2nd figure explaining the destination information in a general parcel. It is a 2nd figure explaining the destination information in a general parcel.

<Embodiment>
Hereinafter, parcel destination information recognition processing according to an embodiment of the present invention will be described with reference to FIGS.
A number with a predetermined number of digits called a sorting number for sorting the packages is printed on the package. The address and address are also written on the front of the parcel, but it is possible to derive the sort number from the address by referring to the data table (table) that associates the address with the sort number. Displays (prints) the barcode of the sorting number that encodes the number. In the automatic package sorting process, generally, the barcode of the sorting number displayed on the package is read by a barcode reader provided on the transport route of the sorter that sorts the package, and the package is sorted. If the barcode cannot be read, it is classified manually, or the image of the parcel is visually observed by a person and the sorting number is keyed (video coding) to classify the parcel. The destination recognition apparatus for parcels according to the present embodiment automates the sorting operation in this sorter by recognizing information on the sorting destination mainly by a recognition technique that makes full use of OCR technology or image processing.

First, an example of a typical mode for displaying destination information in a parcel will be described. FIG. 14 is a first diagram for explaining destination information in a general parcel. FIG. 15 is a second diagram for explaining destination information in a general parcel.
Parcels generally have address information, segment information, and tracking information. The situation can be broadly divided into parcels issued by large traders and parcels issued by individuals. An example of a parcel issued by a large trader is shown in FIG. As an example, a parcel issued by a large trader is affixed with a printed label as shown in FIGS. In this specification, a label attached to a parcel issued by a large trader is referred to as a large label. The large label always displays a sorting number 1A, a sorting bar code 1B in which the sorting number is encoded, and tracking bar codes 1C and 1D. The tracking barcodes 1C and 1D usually include a tracking number of about 12 digits for making a parcel tracking inquiry, and the sorting bar code normally includes a sorting number of about 6 digits. On the large label, the address and name of the sender and the so-called address of the address and name of the recipient are printed with stamp characters. Since the sorting number can be specified by referring to the address database from this address, the address address printed on the large label is also recognized in this embodiment.

On the other hand, FIG. 16 shows an example of a parcel delivered by an individual. In this specification, a label to be attached to a parcel issued by an individual is referred to as a small label. FIG. 16 is a third diagram for explaining destination information in a general parcel. A copy-type label as shown in FIG. 16 as an example is affixed to the personal parcel, and the tracking bar code 2C, address and postal code are written in handwritten or stamped characters on the label. In many cases, however, the address, address and postal code are handwritten. Further, a small label as shown in FIG. 16 is pasted on the parcel to which the copy form slip is attached, and a sorting number 2A and a sort barcode 2B corresponding to the address of the copy form slip are printed on the small label. Yes. In general, this small label is often attached in the vicinity of the copy-type label or overlapping the copy-type label.
The destination recognition apparatus of this embodiment efficiently performs automatic recognition of these parcel destination information.

FIG. 1 is a diagram illustrating an example of a destination recognition apparatus according to an embodiment of the present invention.
As illustrated, the destination recognition apparatus 10 includes an image input unit 11, a barcode reading detection unit 12, a sorting number reading unit 13, a large label identification unit 14, and a tracking barcode usage unit 15.
The image input unit 11 captures a parcel image (parcel image) with a one-dimensional scanner or the like.
The barcode reading detection unit 12 detects a barcode from the parcel image, reads the barcode, and obtains a sorting number. In addition, when the barcode cannot be read, the barcode reading detection unit 12 obtains the coordinate position and the inclination (rotation angle) of an image (rectangular portion) that looks like a barcode.
The sorting number reading unit 13 rotates the parcel image based on the coordinate position and inclination of the rectangular portion, reads the upper and right regions of the rectangular portion with the OCR with reference to the rectangular portion, and then further up and down the parcel image. , And then again read the upper and right regions of the rectangular portion with OCR.
The large label identifying unit 14 performs a pattern matching process of the label image on the parcel image input by the image input unit 11 based on a plurality of large label images of typical large customers registered in advance and frequently registered. Do. Then, the large label identifying unit 14 acquires the position and angle when the matching can be performed, rotates the original parcel image by the angle, and sorts together with the large label image registered in advance. With reference to the position coordinates of the number area and the address area, OCR processing is performed on the area. Accordingly, the large label identifying unit 14 reads the sorting number and the address.
The tracking barcode utilization unit 15 assumes that, when the barcode reading detection unit 12 succeeds in reading the tracking barcode, the input parcel image always includes a large label or a copy slip, Address information is read based on the relative positional relationship between the tracking barcode and the large label or the address area in the copy form. More specifically, when the tracking barcode utilization unit 15 successfully reads the tracking barcode, the tracking barcode utilization unit 15 rotates the input original parcel image by the inclination angle of the tracking barcode. The tracking barcode utilization unit 15 estimates a typical position where the address area is displayed on the large label or the copy slip from the position of the tracking barcode assumed to be on the large label or the copy slip, The postal code and address are read by performing OCR on the estimated position.

Next, a rough flow of destination recognition processing by these functional units will be described.
(Process 1)
First, the image input unit 11 acquires a parcel image obtained by imaging a parcel.

(Process 2)
Next, the barcode reading detection unit 12 reads some barcode from the parcel image, or obtains the coordinate position and the inclination of the rectangular portion where the barcode is displayed.

(Process 3)
When the coordinate position or the like of the rectangular part is read, the sorting number reading unit 13 rotates the parcel image based on the coordinate position and the inclination of the rectangular part, and the upper and right regions of the rectangular part with reference to the rectangular part. Read with OCR. Further, the sorting number reading unit 13 rotates the top and bottom of the parcel image, and then reads the upper and right regions of the rectangular portion again by OCR.

(Process 4)
Next, the large label identifying unit 14 performs a pattern matching process on the label image, estimates a position where the classification number area and the address area are displayed, and performs an OCR process on the estimated position, thereby obtaining a classification number or Read the address.

(Process 5)
When the barcode reading detection unit 12 succeeds in reading the tracking barcode, the tracking barcode utilization unit 15 rotates the input parcel image by the angle of inclination of the tracking barcode, estimates the address area, The postal code and the address are read by performing OCR processing on the estimated address area.

Next, the meaning and merit of recognizing the destination information by performing the processes in the order of (Process 1) to (Process 5) will be described.
The destination information exists in a freely rotated state in a wide range of unspecified places in the parcel image. It is generally difficult for image processing to directly detect a character string of a sorting number that is not a fixed value or a character string of an address. On the other hand, with regard to barcode detection even under the same conditions, barcodes are characterized by the fact that barcodes are lined up continuously within a certain range and the overall barcode shape is an elongated rectangle. The position and rotation angle are easy to detect. Even in an actual parcel sorting machine, the reading rate of the sorting bar code or the tracking bar code of the tunnel bar code reader reaches 70% to 90%, and for that reason, a one-dimensional bar code is attached to the parcel. Therefore, the barcode reading process with the highest reading rate is first performed (processing 2).

  In the present embodiment, even when barcode reading fails, the barcode position and inclination are detected (processing 2). Then, the identification number is recognized using the information (processing 3). The failure to read the barcode is that the barcode bar row cannot be found in the first place, and the start and stop bars cannot be found when the barcode row is detected and the barcode reading process is performed. Decoding processing may fail because the check digit does not match after reading the column. In the latter case, at least the position of the rectangular portion that seems to be a bar row and the 180 ° up / down inversion determination are unknown, but the rotation angle of the rectangular portion in the range of ± 90 ° can be obtained. ). In the present embodiment, barcode reading has failed, but the case where the coordinates and rotation angle of the rectangular portion of the bar row can be specified is referred to as having detected the barcode.

  In (Process 3), when a barcode is detected in (Process 2), based on the coordinates of the rectangular portion of the barcode and the rotation angle, the rotation angle is the center of the rectangular portion where the barcode is supposed to be printed ( Rotate the parcel image (in the direction of 0 ° rotation in the ± 90 ° range). Then, OCR reading of the barcode is performed on the rotated rectangular portion. At this time, the barcode of the rectangular portion is, for example, the tracking barcode 2C on the copy slip shown in FIG. 16 or the sorting barcode 2B on the small label shown in FIG. It cannot be determined whether the classification barcode 1B or the tracking barcode 1C, 1D on the large label is used.

  There are various positions where the address and address of the large label are displayed, and in order to determine the address and address area using only the barcode rectangle detected on the large label, it is not necessary to take a wide area range. However, if the OCR application area is wide relative to the size of the target area where the address is displayed, it is possible to misread a non-address pattern or background included in such a wide area as the address. This is not preferable in recognition processing. In addition, since the copy-type slip shown in FIG. 16 generally has very few types of slips compared to many types of large labels, the relative position between the barcode position and the address or postal code is almost fixed. However, in the case of a copy-type slip, the address is often written by hand, and handwritten OCR is applied. However, since the OCR for handwriting has a lower recognition rate than the OCR using stamps and the possibility of misreading is high, the OCR for handwriting is applied to a certain area in a state where it is not determined that the barcode is a copy-type slip. This increases the possibility of misreading.

  However, paying attention to the fact that the sorting number of the small label is in the vicinity of the upper side of the sorting bar code and the sorting number of the large label is in the vicinity of the right side of the sorting bar code, in these two cases, scanning is performed. Since the area can be narrowed down, an excessive background does not enter the recognition area, and relatively stable OCR recognition is possible. Further, since the classification number is a stamp number and is an object that is easily recognized in OCR recognition, the OCR for the classification number can reduce the misreading rate.

Therefore, in the present embodiment, as a countermeasure against misreading, first, OCR is performed on the classification number. That is, in order to confirm the validity of the OCR reading area, first, labeling processing is performed in the captured image area, and the height and width sizes of the individual objects with labels are measured, and the character size of the classification number and Check if they are approximate. If they are approximate, OCR sorting number reading processing is performed. If there is no approximation, there is a low possibility that the classification number exists, and therefore the reading process of the classification number by OCR is not performed in order to avoid misreading of the OCR.
More specifically, in (Processing 3), the classification number reading unit 13 has a high probability that the classification number exists by using the feature that the classification number to be read by the OCR exists in the vicinity of the barcode. Identify the area. That is, the sorting number reading unit 13 pays attention to the area above the barcode area as the sorting number display area in consideration of the possibility that the barcode is a sorting barcode with a small label. Further, the sorting number reading unit 13 pays attention to the area on the right side of the barcode area as the sorting number display area in consideration of the possibility that the barcode is a sorting barcode of a large label. The classification number reading unit 13 confirms whether or not there is a density image having a density within a certain range in the focused area, and if the presence is confirmed, performs a reading process by OCR on the focused area.

  In this way, in (Process 3), when character recognition OCR is applied to an image with a complicated background such as a parcel, misrecognition is likely to occur. Do. In addition, when the sorting number is read on the right side of the detected barcode (where the sorting number of the large label is estimated to be displayed), the address information is obtained by OCR from the area below the sorting number image. read. This is based on the result of inspection that there are many address and address descriptions directly under the part where the classification number is printed as an aspect of the large label. By recognizing the address in this area and collating it with the read sorting number, it is possible to reduce misreading of the sorting number. As described above, (Process 3) is a stable OCR recognition process with few erroneous readings after the barcode reading process of (Process 2). Therefore, in the present embodiment, (Process 3) is performed after (Process 2).

  Further, in (Process 3), the parcel image is rotated at the rotation angle of the detected barcode in order to match the rotation angle of the read target area and the parcel image, but it is assumed that there is a classification number on the right side of the barcode detection area. Thus, when it is assumed that the label is a large label, if the tracking barcode is read, the image is first rotated at the inclination angle of the tracking barcode. In the case of large labels, the classification number and the tracking barcode are printed on the same label, and the two are in a parallel relationship. Therefore, if the image is rotated according to the rotation angle of the tracking barcode, This is because when reading the sorting number is intended, there is an advantage that the inclination shift of the sorting number area is reduced, and the recognition performance can be improved. The tracking bar code is used because the recognition process can be completed in (Process 2) when the sorting bar code can be read.

  In the next (Process 4), it is assumed that some pattern matching method including free rotation of the input parcel image is applied to the detection of the large label. This pattern matching method generally registers a plurality of matching target images in advance, that is, in the case of the present invention, a plurality of large label images, but the large label is created individually by the large contractor who is the sender. Therefore, it is not possible to register all kinds of labels in advance. Therefore, even if it is assumed that the performance of the pattern matching process is ideal, it is difficult to detect a large label as compared with a sorting barcode or a tracking barcode that is always displayed on the surface of the parcel. Therefore, (Process 4) is considered to be inferior in detection performance compared with (Process 3), and is performed after (Process 3) as the priority order of recognition.

  By the way, the pattern matching process is a large object such as a parcel, and the large label that is the search target is also large in area size, so the processing time is very large due to the synergy of both processes, Processing may not be in time for parcel classification. Therefore, in this embodiment, when a tracking barcode is detected, a device for performing pattern matching is performed from the vicinity of the position of the tracking barcode (device 1). In addition, in pattern matching with a free rotation angle range, it is necessary to perform matching while rotating the reference pattern image registered for matching by 360 °. If a tracking barcode is detected, it corresponds to the tracking barcode. It can be estimated that there is a large label in a direction parallel to the rotation angle of the rectangular part to be searched, so the search is performed by narrowing down near the rotation angle of the rectangular part, and the rotation angle range of the image at the time of pattern matching is narrowed down Perform (device 2). Thereby, the position scanning range and the rotational scanning range required for pattern matching can be narrowed down, and as a result, the processing time can be greatly reduced.

  In addition, as shown in FIG. 14, the large label has a tracking bar code, an address and address of the addressee, and two description columns of senders, and the same information is often described in each of them. This is because the large label can be divided into two labels, and one label is used as a record. Observe how large labels are attached to parcels. When large labels are attached to parcels, they are usually placed on the top of the package or on a flat surface. However, if the size of the parcel is small, the large label is folded in two. It may be pasted on two sides across the ridgeline. When pasting across two sides in this way, large labels are not randomly folded, but are folded so as to be separated by two labels as described above, and at least one addressing area is on one plane. It is often applied to be displayed on the screen. In this way, when folded in half across the ridgeline of the parcel packaging box, the original appearance of the large label is not reflected in the image on the image, so the size of the large label at the time of registration is also in shape Are not similar, and pattern matching does not match. Therefore, in the present embodiment, based on the above observation result, when registering an image for collation of a large label, an image in which only half of the large label is intentionally specified is registered together with the registration of the entire label image. Do some ingenuity. Thereby, a pattern matching process is applicable also to the large label affixed straddling the above ridgelines in half.

  Next, (Process 5) will be described. In the barcode recognition process in (Process 2), both the sorting barcode and the tracking barcode are recognized. If the sorting barcode can be read here, the sorting barcode is read regardless of whether the tracking barcode is read or not. Classification is possible by the classification number stored in the code. On the other hand, if only the tracking barcode can be read, the tracking barcode is likely to be either a tracking barcode on a copy form or a tracking barcode on a large label. As a premise, the destination label affixed on the parcel is either a copy-type slip or a large label, but the copy-type slip always contains one tracking barcode, while the large label has the same two tracking Suppose a barcode is included. The reason for including two tracking barcodes in large labels is that many large labels are composed of two square labels that are connected vertically and horizontally as described above. This is because it is peeled off and used as a record. The label shown in FIG. 14 is an example of a large label in which an address and a tracking bar code are connected vertically. The label shown in FIG. 15 is an example of a large label in which an address and a tracking barcode are arranged on the left and right. As described above, in the case of a large label, there are a mode in which two identical tracking bar codes are arranged vertically on the image, and a mode in which they are horizontally arranged side by side, and there are very many large labels in these modes. Using this property, if one tracking bar code can be read in an image, it is assumed to be a duplicate slip, and if two identical tracking bar codes can be read within a relatively close distance, it is a large label tracking. It can be estimated that it is a barcode.

  When a copy-type slip is estimated, the format of the copy-type slip is constant, so the address and postal code area are estimated from the tracking bar code position, and the address and postal code are read by OCR to obtain information on the sorting destination. Can do. Also. The copy-type slip has a 12-digit tracking number printed at a fixed position on the slip. Therefore, if one tracking bar code can be read, whether it is a tracking bar code on a copy form or not, a 12-digit tracking number in OCR for a predetermined tracking number area estimated from the tracking bar code position. By reading the above, if the number read by the OCR is completely or a certain number of digits matches the tracking number included in the tracking barcode, it can be trusted that it is a copy-type slip.

  On the other hand, when it is determined that the label is a large label, there are characteristics in the position of the address description area depending on whether the two tracking barcodes are aligned vertically on the image or horizontally aligned horizontally. When observing the state of large labels, in the former case, the address is often printed above the upper tracking barcode, and in the latter case, the address is printed above the left tracking barcode. There are many. The tracking bar code may include a service type determined by the parcel delivery company in addition to the tracking information. An example is a time service that delivers a parcel to a customer at a specified time. When using this service, a label for a dedicated time service is used. In the case of the dedicated label, the tracking bar code includes a type code indicating the service. When such a service type exists by reading a tracking barcode, it is possible to specify an area including a sorting number and an address by specifying a dedicated label used for the service from the service type. .

  Sorting destination information can be obtained by estimating the address area by various judgments based on the state of the copy form slip and the large label and reading the print address by OCR. As described above, (Process 5) is a means for obtaining an address area by estimation only with tracking barcode reading information under a situation where the sorting barcode cannot be read, the sorting number cannot be read, and the large label cannot be specified. However, since there is no remarkable feature in the relative position of the tracking bar code and the sorting number, and the region of the sorting number cannot be estimated, in (Process 5), the OCR reading of the address is applied and not the OCR reading of the sorting number. Sort by address. Alternatively, a sorting number is obtained from the read address with reference to the data table, and the sorting is performed based on the sorting number.

  As described above, the order of processing 2 to processing 5 is processing from the processing with the fewest estimation elements to the processing with the largest estimation elements. In the destination recognition method of this embodiment, various reading processes may be performed using the same OCR. The waterfall method is adopted in order from the method with the lowest possibility of misreading. Thereby, a classification number etc. can be read efficiently.

<Example>
The destination recognition method of this embodiment will be described in more detail below using FIGS. 2 to 15 and an example of a system that automatically sorts parcels with a sorter having an optical image reader.
FIG. 2 is a diagram illustrating an example of an automatic parcel sorting system according to an embodiment of the present invention.
As shown in FIG. 2, the parcel automatic sorting system 1 includes a transport control device 200, an image input camera 300, an image input device 400, a parcel sorting destination recognition device 500, and a sorting destination determination device 600. .

The conveyance control device 200 conveys the parcel 100. The parcel 100 is attached with a label or slip printed with a sorting number, a mailing address, a sorting barcode, and a tracking barcode.
The image input camera 300 captures the parcel 100 being conveyed and captures a grayscale image.
The image input device 400 acquires a grayscale image obtained by capturing the parcel 100 from the image input camera 300. The grayscale image acquired by the image input camera 300 is image data indicating an image composed of a plurality of pixels indicating, for example, 256 gradations of shades of 0 to 255. The gradation indicates that 0 indicates black, and the image becomes brighter (closer to white) as the numerical value approaches 255.
The parcel sorter recognition apparatus 500 recognizes the parcel sort number by image processing.
The sorting destination determining device 600 accommodates the parcel 100 in an appropriate container 700 from among a plurality of containers 700 prepared for each destination based on the sorting number recognized by the sorting destination recognition device 500 for parcels.

FIG. 3 is a diagram illustrating a hardware configuration example of a parcel sorting destination recognition apparatus according to an embodiment of the present invention.
As shown in FIG. 3, the parcel sorter recognition apparatus 500 includes a CPU (Central Processing Unit) 51, a ROM (Read Only Memory) 52, a RAM (Random Access Memory) 53, an HDD (Hard Disk Drive) 54, and an interface 55. Each hardware configuration may be provided as an example. The interface 55 is, for example, a liquid crystal touch panel or a button. The parcel sorter recognition apparatus 500 may be configured to further include other hardware.

  In the parcel automatic sorting system 1 in FIG. 3, the image input camera 300 captures an image of the parcel 100 conveyed by the conveyance control device 200, and outputs the grayscale image (parcel image) to the image input device 400. The image input device 400 inputs the acquired grayscale image to the parcel sorter recognition device 500, and the sort destination determination device 600 selects the sort destination based on the sort number that the parcel sortee recognition device 500 outputs within a predetermined time. Then, the conveyance control device 200 stores the package in one of the plurality of containers 700, thereby realizing automatic sorting of the package. That is, in the parcel automatic sorting system 1 of FIG. 3, the parcel sorting destination recognition apparatus 500 corresponds to the destination recognition apparatus 10 described in FIG. 1 and the like, and recognizes the parcel sorting number by the above-described series of processing.

Next, the parcel sorter recognition apparatus 500 will be described.
FIG. 4 is a diagram illustrating an example of a parcel sorter recognition apparatus according to an embodiment of the present invention.
As shown in FIG. 4, the parcel sorter recognition apparatus 500 includes an image buffer unit 501, a barcode reading detection unit 502, a reading result determination unit 503, a sorting number reading unit 504, a large label identifying unit 505, A tracking barcode usage unit 506, a result output unit 507, a large label registration unit 508, and a large label storage unit 509 are provided.
The image buffer unit 501 acquires an image from the image input device 400 and temporarily stores it.
A barcode reading detection unit 502 reads a barcode from an image stored in the image buffer unit 501. The barcode reading detection unit 502 detects the position coordinates of the rectangular portion of the barcode and the rotation angle on the image.
The reading result determination unit 503 controls the sorting destination recognition process based on the reading result of the sorting number.
The sorting number reading unit 504 reads the sorting number by OCR based on the image stored in the image buffer unit 501 and the position coordinates and rotation angle information of the rectangular portion detected by the barcode reading detection unit 502.
The large label identifying unit 505 reads the classification number and address of the large label by OCR by pattern matching.
The tracking barcode utilization unit 506 is configured to generate an address area for a large label or a copy-type label based on the image stored in the image buffer unit 501 and the information on the position coordinates and the rotation angle of the rectangular portion detected by the barcode reading detection unit 502. The address and address are read by OCR.
The result output unit 507 outputs the sorting number of the read result to the sorting destination determination device 600.
The large label registration unit 508 registers the reference image for the large label used for pattern matching in the large label storage unit 509.
The large label storage unit 509 stores a large label image.

FIG. 5 is a flowchart showing an example of the parcel sorting destination recognition process according to the embodiment of the present invention.
First, the image buffer unit 501 acquires an image from the image input device 400 (step S11). The image buffer unit 501 temporarily stores the acquired parcel image until the classification destination is determined by the classification destination determination unit 600.

  Next, the barcode reading detection unit 502 acquires an image from the image buffer unit 501, and detects and reads the barcode (step S12). When the barcode reading detection unit 502 detects a barcode (tracking barcode or sorting barcode), the barcode reading detection unit 502 outputs the position coordinates of the rectangular portion of the barcode and the rotation angle on the image to the reading result determination unit 503. Further, when reading the barcode, the barcode reading detection unit 502 outputs the barcode reading result to the reading result determination unit 503. Further, the barcode reading detection unit 502 outputs a reject signal indicating that detection is impossible to the reading result determination unit 503 when the barcode detection also fails. Step S12 corresponds to the above (Process 2).

  Next, the reading result determination unit 503 determines whether the sorting number has been read based on the information acquired from the barcode reading detection unit 502 (step S13). If the sorting number has been read (step S13; Yes), that is, if the sorting bar code has been successfully read in step S12, the process proceeds to step S19 described later. When the sorting number cannot be read (step S13; No), the sorting number reading unit 504 acquires the parcel image from the image buffer unit 501, acquires the position coordinates and the rotation angle of the rectangular portion from the barcode reading detection unit 502, The sorting number is read by OCR (step S14). The sorting number reading unit 504 outputs the reading result (sorting number) or a rejection signal indicating reading failure to the reading result determination unit 503. Step S14 corresponds to the above (Process 3).

  Next, the reading result determination unit 503 determines whether the sorting number has been read based on the information acquired from the sorting number reading unit 504 (step S15). When the classification number is read (step S15; Yes), the process proceeds to step S19 described later. When the classification number cannot be read (step S15; No), the large label identifying unit 505 extracts the image from the image buffer unit 501 and the position coordinates and the rotation angle of the rectangular portion of the tracking barcode detected by the barcode reading detection unit 502. And the sorting number and the like are read by pattern matching (step S16). The barcode reading detection unit 502 outputs a reject signal to the reading result determination unit 503 when the read sorting number or the sorting number cannot be read. Step S16 corresponds to the above (Process 4).

Next, the reading result determination unit 503 determines whether the sorting number has been read based on the information acquired from the large label identification unit 505 (step S17). When the classification number can be read (step S17; Yes), the process proceeds to step S19 described later. When the sorting number cannot be read (step S17; No), the tracking barcode usage unit 506 reads the sorting number using the tracking barcode (step S18). Specifically, the tracking barcode using unit 506 acquires an image from the image buffer unit 501, acquires the rectangular position and rotation angle information of the tracking barcode from the barcode reading detection unit 502, and outputs a large label or a copy type Read the address on the label. The tracking barcode usage unit 506 outputs the reading result to the result output unit 507. Step S18 corresponds to the above (Process 5).
Next, the result output unit 507 outputs the reading result (step S19). The result output unit 507 outputs the sorting result of the reading result or the reject signal to the sorting destination determining device 600. The sorting destination determination device 600 determines one of the sorting destination containers 700 based on the sorting number acquired from the package sorting destination recognition apparatus 500. The transport control device 200 sorts the parcel 100 according to the decided sorting destination.

Next, the process of step S12 will be described in detail with reference to FIG.
FIG. 6 is a flowchart illustrating an example of barcode detection and reading processing according to an embodiment of the present invention.
First, the barcode reading detection unit 502 acquires a parcel image from the image buffer unit 501 and detects and reads a barcode (step S21). A barcode reading detection unit 502 detects and reads a sorting barcode and a tracking barcode on a parcel image. Next, the barcode reading detection unit 502 determines whether or not the sorting number has been read in the process of step S21 (step S22). When the classification number is read (step S22; Yes), the barcode reading detection unit 502 sets the classification number as output information (step S23). Then, the process proceeds to step S29. In this case, since the sorting number has been read, the recognition process is completed.

  When the classification number cannot be read (step S22; No), the barcode reading detection unit 502 determines whether or not the tracking barcode can be read (step S24). When one or more tracking barcodes have been read (step S24; Yes), the barcode reading detection unit 502 outputs the position coordinates and rotation angles (360 °) of the rectangular portions of all the read tracking barcodes. Information is set (step S25). Then, the process proceeds to step S26. If the tracking barcode has not been read (step S24; No), the process proceeds to step S26.

  Next, the barcode reading detection unit 502 determines whether or not the barcode has been detected (step S26). When the barcode can be detected (step S26; Yes), the barcode reading detection unit 502 sets the position coordinates and the rotation angle (360 °) of the rectangular portions of all the detected barcodes as the output information ( Step S27). Then, the process proceeds to step S29. If reading is not possible in step S24, the barcode is not read, so the 360 ° rotation direction is unknown, and the output rotation angle range is 180 ° (± 90 °). When the tracking barcode cannot be read (step S26; No), the barcode reading detection unit 502 sets the output information to undetectable (step S28), and proceeds to the process of step S29. In step S29, the barcode reading detection unit 502 outputs the value set in the output information.

The value set in the output information is one of the following:
1. Sorting number 2. Position coordinates and 360 ° rotation angle of the rectangular part of the tracking barcode. Since there are cases where the same tracking barcode is printed on a large label, a plurality of tracking barcodes may be output.
3. Position coordinates and 180 ° rotation angle of the rectangular part of the detected barcode. Multiple outputs may occur.
4). Both “2” and “3” When the detection is impossible “1.” or “5.”, the barcode reading detection unit 502 outputs the set output information (sorting number) to the result output unit 507 and the reading result determination unit 503. In the case of “1.”, the result output unit 507 outputs the sorting number set in the output information to the sorting destination determining device 600. In the case of “5.”, deal with it by sorting manually.
In the case of “2.” to “4.”, the barcode reading detection unit 502 outputs the output information to the sorting number reading unit 504 and the reading result determination unit 503 (step S29). Information on the position coordinates and the rotation angle of the rectangular portion of the barcode output to the sorting number reading unit 504 is referred to as barcode detection information.

Next, the process of step S14 will be described in detail with reference to FIG.
FIG. 7 is a flowchart showing an example of a sorting number reading process according to an embodiment of the present invention.
First, the sorting number reading unit 504 acquires barcode detection information (step S31). Also, the sorting number reading unit 504 acquires a parcel image from the image buffer unit 501.
Next, the sorting number reading unit 504 determines whether there is barcode detection information (step S32). When there is no barcode detection information (step S32; No), the sorting number reading unit 504 sets no barcode detection information as output information. In this case, the process ends when performing the recognition process. When there is barcode detection information (step S32; Yes), the sorting number reading unit 504 performs image rotation processing on the parcel image (step S33). Specifically, the classification number reading unit 504 rotates the neighborhood image including the detected rectangular portion of the barcode using the rotation angle of the detected rectangular portion of the barcode included in the barcode detection information. If the detected barcode is a tracking barcode, the sorting number is arranged in parallel to the tracking barcode, and therefore the direction of the sorting number is aligned horizontally by this rotation process. The sorting number reading unit 504 uses the rotation angle of the tracking barcode when the barcode reading detection unit 502 reads the tracking barcode, and if the barcode reading detection unit 502 only detects the barcode, Using the rotation angle of the detection barcode, the parcel image is rotated by the rotation angle.

Next, based on the position coordinates of the detected rectangular portion of the barcode included in the barcode detection information, the sorting number reading unit 504 reads the sorting number by OCR for the area above the barcode (step S34). . If the detected barcode is a small label sorting barcode, the sorting number is displayed in the upper area of the sorting barcode, and thus the sorting number can be read by this processing.
Next, the sorting number reading unit 504 determines whether or not the sorting number has been read in the process of step S34 (step S35). When the sorting number is read (step S35; Yes), the sorting number reading unit 504 sets the read sorting number as output information, and proceeds to the process of step S395. If the sorting number could not be read (step S35; No), the sorting number reading unit 504 reads the sorting number by OCR for the area on the right side of the barcode based on the position coordinates of the rectangular portion of the barcode. (Step S36). If the detected bar code is a large label sorting number bar code, the sorting number is often displayed in the right area of the large label sorting bar code. it can.

  Next, the sorting number reading unit 504 determines whether the sorting number has been read in the process of step S36 (step S37). When the sorting number is read (step S37; Yes), the sorting number reading unit 504 sets the read sorting number as output information, and proceeds to the process of step S395. If the sorting number could not be read (step S37; No), the sorting number reading unit 504 performs image rotation processing (step S38). Specifically, the sorting number reading unit 504 further rotates the image rotated in step S33 by 180 °. In this rotation process, since the rotation angle of the rectangular portion of the barcode included in the barcode detection information may be in the range of ± 90 °, the image is inverted 180 ° and turned upside down and again the classification number This is done to perform the recognition process.

The subsequent processing procedure is the same as that in steps S34 to S37. That is, the sorting number reading unit 504 reads the sorting number by OCR for the area above the barcode rectangular portion (step S39). If the sorting number has been read (step S391; Yes), the sorting number reading unit 504 sets the read sorting number as output information, and the process proceeds to step S395. If the sorting number could not be read (step S391; No), the sorting number reading unit 504 reads the sorting number by OCR for the right area of the rectangular portion of the barcode (step S392). If the sorting number can be read (step S393; Yes), the sorting number reading unit 504 sets the read sorting number as output information, and the process proceeds to step S395. If the sorting number cannot be read (step S393; No), the sorting number reading unit 504 determines whether there is any other barcode detection information (step S394). If there is other barcode detection information (step S394; Yes), the processing from step S33 is repeated. If there is no other barcode detection information (step S394; No), the process proceeds to step S395.
In step S395, if the classification number is obtained, the classification number reading unit 504 outputs the classification number set in the output information to the reading result determination unit 503 and the result output unit 507 (step S395). In this case, the read result determination unit 503 determines Yes in step S15 of FIG. 5, and the result output unit 507 outputs the sorting number set in the output information to the sorting destination determining device 600. The sorting destination determination device 600 sorts the parcel 100. The sorting number reading unit 504 outputs a reject signal to the reading result determining unit 503 and the large label identifying unit 505 if the sorting number is not obtained. In this case, in step S15 of FIG. 5, the reading result determination unit 503 determines No, and the process of the next step S16 is performed.

Next, the processing in step S16 will be described in detail with reference to FIGS.
FIG. 8 is a first flowchart showing an example of a large label identification process according to an embodiment of the present invention.
First, the large label identifying unit 505 acquires a parcel image from the image buffer unit 501 and barcode information from the barcode reading detection unit 502 (step S41). Next, the large label identifying unit 505 rotates the parcel image based on the acquired barcode detection information (step S42). For example, if the barcode detection information includes the position coordinates and the rotation angle of the tracking barcode, the large label identifying unit 505 rotates the image only in the vicinity region around the position of the tracking barcode. Also in the image matching process in the subsequent step S44, the plane search range is limited to the vicinity region centered on the position of the tracking barcode, and the angle search range is also in the vicinity of the horizontal angle 0 ° with respect to the rotated image Limited to a range of angles. This is because large labels and duplicating labels always have one or more tracking barcodes. If the tracking barcode is readable, it is within the plane range that includes the tracking barcode and within the inclination angle range of the tracking barcode. This is because a large label is considered to exist, so that the search range is limited both planarly and angularly, thereby shortening the free rotation matching processing time, which takes time for processing. Next, the large label identifying unit 505 refers to the large label images one by one from a plurality of large label image groups registered in the large label storage unit 509 (step S43).
Next, the large label identifying unit 505 performs image matching processing for searching for registered images one by one and outputting a matching result to the parcel image (step S44). This image matching process is a free rotation image matching process performed not only in the plane direction but also in the rotation direction. When performing the matching process, the large label identifying unit 505 calculates the degree of coincidence with the registered image.

Next, the large label identifying unit 505 determines whether or not there is another registered large label that has not been subjected to matching processing (step S45). If another large label image exists (step S45; Yes), The matching process is repeated with all registered large labels, and the degree of match is calculated for each matching process.
When matching with all the large label images is completed (step S45; No), the large label identifying unit 505 next selects the largest matching degree (step S46). Next, the large label identifying unit 505 determines whether or not the selected largest matching degree is equal to or greater than a predetermined threshold (step S47). The predetermined threshold is a minimum value at which it can be determined that the images are sufficiently matched. When the largest matching degree is less than the predetermined threshold (step S47; No), the large label identifying unit 505 sets the mismatch as output information, and proceeds to the process of step S491.

  On the other hand, when the largest matching degree is equal to or greater than the predetermined threshold (step S47; Yes), the large label identifying unit 505 refers to the registered large label image having the largest matching degree and displays the classification number in the registered large label image. The coordinate information of the display area of the area and the address is acquired (step S48). Next, the large label identifying unit 505 uses the coordinate information of the classification number area and the address address area acquired in step S48 to determine the classification number area and the address address area in the large label in the parcel image acquired from the image buffer unit 501. The target OCR reading is performed (step S49). If the OCR reading is successful, the sorting number can be acquired. Next, the large label identifying unit 505 sets the sorting number as output information when the sorting number is acquired, and sets a reject signal as output information when the sorting number cannot be read. The large label identifying unit 505 may check that the sorting number is correct by collating the read address with the sorting number.

  Next, if the classification number is obtained, the large label identifying unit 505 outputs the classification number set in the output information to the reading result determination unit 503 and the result output unit 507 (step S491). In this case, in step S <b> 17 of FIG. 5, the read result determination unit 503 determines Yes, and the result output unit 507 outputs the sorting number set in the output information to the sorting destination determining device 600. The sorting destination determination device 600 sorts the parcel 100. Further, the large label identifying unit 505 outputs a reject signal to the reading result determining unit 503 and the tracking barcode using unit 506 if the sorting number is not obtained. In this case, the reading result determination unit 503 determines No in step S <b> 17 of FIG. 5, and the process of the next step S <b> 18 is performed.

Next, the processing in step S16 in FIG. 8 will be described in detail with reference to FIG.
FIG. 9 is a second flowchart illustrating an example of a large label identification process according to an embodiment of the present invention.
FIG. 9 is a processing flow for registering large label images and creating large label registration information acquired in step S48. The registration of the large label is a process performed in advance for the parcel 100.
First, the large label registration unit 508 inputs an image of a parcel with a large label to be registered (step S51). Next, the large label registration unit 508 displays the parcel image to the registrant who performs the large label registration operation (step S52). Next, the large label registration unit 508 performs large label ID assignment processing for assigning the ID number designated by the registrant to the large label in order to make the large label unique (step S53). Next, the large label registration unit 508 registers the outline region of the large label designated by the registrant (step S54). At this time, the registrant looks at the image and designates the entire rectangle of the large label. Next, the large label registration unit 508 registers the sorting number area and the address address area in the large label designated by the registrant (step S55). At this time, the registrant visually checks the image and designates the sorting number description area and the address and address description area in the large label.
Next, the registrant visually determines whether the large label is in a state where two identical labels are lined up and down or left and right (step S56). If the same label is arranged (step S56; Yes), the registrant designates one rectangular label area of the two labels in the large label, and performs label half area designation processing to be cut out as the large label. (Step S57). If the same label is not arranged (step S56; No), the process proceeds to step S59.
Next, the registrant designates the sorting number description area coordinates and the address address description area coordinates in the area cut out in step S57 (step S58). Finally, the large label registration unit 508 corresponds to the large label image, the ID number thereof, the coordinates of the sorting number description area and the address address description area determined by the above processing, the rectangular label area for one sheet of a series of labels, and the like. At the same time, it is registered in the large label storage unit 509 (step S59). Through these processes, the registered large label is registered in the large label storage unit 509.

Next, the process of step S18 will be described in detail with reference to FIG.
FIG. 10 is a flowchart illustrating an example of tracking barcode usage processing according to an embodiment of the present invention.
First, the tracking barcode utilization unit 506 acquires a parcel image from the image buffer unit 501 and barcode information from the barcode reading detection unit 502 (step S61). Next, the tracking barcode utilization unit 506 determines whether only two tracking barcodes can be read from the parcel image and whether the rotation angles of the two tracking barcodes on the image are substantially the same ( Step S62). When two tracking barcodes in almost the same direction can be read (step S62; Yes), the tracking barcode utilization unit 506 has a neighboring image centered on a substantially middle point between the two tracking barcode rectangular coordinate values. Is rotated by the rotation angle of the two tracking barcodes (step S63). Next, the tracking barcode utilization unit 506 determines whether two tracking barcodes are arranged side by side (step S64). When two tracking barcodes are arranged side by side (step S64; Yes), the tracking barcode utilization unit 506 roughly displays a rectangular area above the rectangular coordinates of the barcode on the left side of the tracking barcode arranged side by side. And OCR reading of the address is performed on the area (step S65). The tracking barcode utilization unit 506 sets the reading result as output information, and proceeds to step S692.

  When the two tracking barcodes are not arranged side by side (step S64; No), the tracking barcode utilization unit 506 determines whether the two tracking barcodes described above are arranged vertically (step S66). . When two tracking barcodes are lined up and down (step S66; Yes), the tracking barcode utilization unit 506 is roughly rectangular above the rectangular coordinates of the tracking barcode above the tracking barcode lined up and down. An area is defined, and OCR reading of the address is performed within the area (step S67). The tracking barcode utilization unit 506 sets the reading result as output information, and proceeds to step S692. When two tracking barcodes are lined up and down (step S66; No), the tracking barcode utilization unit 506 sets a reject signal in the output information, and proceeds to step S692.

  If two tracking barcodes cannot be read (step S62; No), the tracking barcode usage unit 506 determines whether only one tracking barcode can be read from the parcel image (step S68). If no tracking barcode is readable (step S68; No), the tracking barcode utilization unit 506 sets a reject signal in the output information, and the process proceeds to step S692. When only one tracking barcode can be read (step S68; Yes), the tracking barcode utilization unit 506 rotates the neighboring image by the rotation angle around the rectangular portion of one tracking barcode (step S68). S69). Next, the tracking bar code utilization unit 506 assumes that the tracking bar code is a tracking bar code of a copy form slip, and uses a relative predetermined positional relationship between the tracking bar code position of the copy form slip and the address address description position. An address description area and an address description area are determined, and OCR reading of the address is performed within the area (step S691). The tracking barcode utilization unit 506 sets the reading result as output information, and proceeds to step S692. The tracking barcode utilization unit 506 outputs the classification number set in the output information to the result output unit 507 if the classification number is obtained, or the reject signal if it is obtained (step S692). The result output unit 507 outputs the sorting number set in the output information to the sorting destination determining device 600. The sorting destination determination device 600 sorts the parcel 100.

An example of a parcel image obtained by picking up a conveyed parcel with a camera from above is shown in FIGS. As shown in FIG. 11 (a), FIG. 11 (b), FIG. 12 (a), and FIG. 12 (b), there are parcels having labels in various modes. According to the present embodiment, it is possible to recognize the sorting number with high accuracy from a parcel having such various types of labels.
As described above, the parcel sorter recognition apparatus 500 uses the fact that there is a sort number display area, an address and an address display area in the vicinity of the barcode position assigned to the parcel or at a predetermined relative position. Thus, by performing the estimation of the OCR reading target area in a waterfall manner in a probable order, it is possible to efficiently perform the sorting number recognition process in a short time while reducing the possibility of misreading.

FIG. 13 is a block diagram showing a minimum configuration of a destination recognition apparatus according to an embodiment of the present invention.
The destination recognition device 10 includes at least a barcode reading detection unit 12.
The barcode reading detection unit 12 reads the code information displayed on the parcel image including the parcel destination information, and reads the sorting number indicated by the code information. If the sorting number cannot be read, the barcode reading detection unit 12 detects information serving as a reference for the display position of the sorting number in the parcel image.

  The destination recognition device 10 and the package sorting destination recognition device 500 described above have a computer inside. Each process of the destination recognition apparatus 10 and the package sorting destination recognition apparatus 500 described above is stored in a computer-readable recording medium in the form of a program, and the program is read and executed by the computer. Processing is performed. Here, the computer-readable recording medium means a magnetic disk, a magneto-optical disk, a CD-ROM, a DVD-ROM, a semiconductor memory, or the like. Further, the computer program may be transmitted to the computer via a communication line, and the computer that has received the transmission may execute the program. The program may be for realizing a part of the functions described above. Furthermore, what can implement | achieve the function mentioned above in combination with the program already recorded on the computer system, what is called a difference file (difference program) may be sufficient. In addition, all or a part of the functions of the destination recognition device 10 and the package sorting destination recognition device 500 described above may be ASIC (Application Specific Integrated Circuit), PLD (Programmable Logic Device), FPGA (Field-Programmable Gate Array), You may implement | achieve using hardware, such as PLC (Programmable Logic Controller).

  In addition, it is possible to appropriately replace the components in the above-described embodiments with known components without departing from the spirit of the present invention. The technical scope of the present invention is not limited to the above-described embodiment, and various modifications can be made without departing from the spirit of the present invention. The barcode reading detection unit 12 and the barcode reading detection unit 502 are examples of a code reading unit. The large label identifying unit 14 and the large label identifying unit 505 are examples of a label identifying unit. The tracking barcode usage unit 15 and the tracking barcode usage unit 506 are examples of the tracking code usage unit. The tracking bar codes 1C and 1D are examples of tracking code information.

DESCRIPTION OF SYMBOLS 1 ... Parcel automatic classification system 10 ... Destination recognition apparatus 11 ... Image input part
DESCRIPTION OF SYMBOLS 12 ... Barcode reading detection part 13 ... Sorting number reading part 14 ... Large label identification part 15 ... Tracking barcode utilization part 200 ... Conveyance control apparatus 300 ... Image input camera 400- ··· Image input device 500 ··· Sorting device recognition device for parcel 501 ··· Image buffer portion 502 ··· Barcode reading detection portion 503 ··· Reading result determination portion 504 · · · Sorting number reading portion 505 ···・ Large label identification unit 506... Tracking bar code using unit 507 .. result output unit 508... Large label registration unit 509... Large label storage unit 600.

Claims (10)

A code reading unit that reads the code information displayed on the image including the parcel destination information and reads the sorting number indicated by the code information;
With
The code reading unit detects information serving as a reference for the display position of the sorting number in the image when the sorting number cannot be read;
Destination recognition device. The coordinate information displayed as the reference information for the display position of the sorting number, the coordinates of the display of the code information, the inclination information of the display of the code information, and the image are acquired, and the image is classified by the inclination. The code information display area based on a predetermined positional relationship between the code information display area and the classification number display area and the coordinates of the code information, and the estimated display area is estimated. Sorting number reading unit that performs sorting number reading processing as a target,
The destination recognition apparatus according to claim 1, further comprising: The sorting number reading unit performs sorting number reading processing for an upper area of the code information display area or an area on the right side of the code information display area.
The destination recognition apparatus according to claim 2. Image matching is performed between a plurality of reference label images including destination information and an image including destination information of the parcel, and the reference label image having the highest matching degree is equal to or higher than a predetermined threshold value. Based on the coordinate information of the display area of the sorting number, a label identifying unit that performs a reading process of the sorting number for the display area of the sorting number in the image including the parcel destination information,
The destination recognition device according to any one of claims 2 to 3, further comprising: When the code reading unit reads the parcel tracking code information from the image including the parcel destination information, the display position of the parcel destination information is specified based on the display position of the tracking code information. , A tracking code using unit for reading destination information from the display position,
The destination recognition device according to claim 4, further comprising: The tracking code utilization unit changes the display position of the destination information of the parcel to be identified according to the number of the tracking code information read.
The destination recognition apparatus according to claim 5. First, the sorting number is read by the code reading unit,
Next, the sorting number reading unit performs the sorting number reading process,
Next, the label identification unit performs the sorting number reading process,
Finally, read processing of the destination information by the tracking code using unit,
The destination recognition apparatus according to claim 6. The destination recognition apparatus according to any one of claims 1 to 7,
A camera that captures an image including destination information of the parcel and outputs the captured image to the destination recognition device;
A package automatic sorting system comprising: a sorting destination determining machine that determines a sorting destination based on a sorting number read by the destination recognition device. Reading the code information displayed on the image including the destination information of the parcel, and reading the sorting number indicated by the code information;
If the sorting number cannot be read, detecting information serving as a reference for the display position of the sorting number in the image;
A destination recognition method. The computer of the destination recognition device
Means for reading the code information displayed on the image including the destination information of the parcel and reading the sorting number indicated by the code information;
Means for detecting information serving as a reference for the display position of the sorting number in the image when the sorting number cannot be read;
Program to function as.