JPH02224083A - Method for detecting display part for article discrimination - Google Patents

Abstract

PURPOSE:To extract in a short period a position, a direction and a size of a character and a symbol to be recognized in image data by providing in advance a mark consisting of a roughly concentric circle-shaped pattern on an article, deriving a position of the mark in the image data brought to image pickup, and detecting an article discrimination use display part, based on the position as a reference. CONSTITUTION:Together with an article discrimination use display pattern 1D, marks Ma, Mb consisting of a roughly circular or roughly concentric circle-shaped pattern in a prescribed position relation against the article discrimination use display pattern 1D are provided on an article A. In this state, together with the discrimination use display pattern 1D, a range containing the marks Ma, Mb is brought to image pickup, and from in image data, positions of the marks Ma, Mb are derived. Based on these derived positions of the marks Ma, Mb as a reference, an article discrimination use display part is detected. In such a way, even when a position relation of the article discrimination use display pattern to a television camera for an image pickup, etc., is not unified, the image data of the article discrimination use display pattern can be extracted and recognized in a short period from the image data.

Description

Detailed Description of the Invention (a) Industrial Application Field The present invention is used for a device for recognizing characters and symbols for identifying an article displayed on the surface of an article, and is used for recognizing characters and symbols for identifying an article displayed on the surface of the article. The present invention relates to a method of detecting the position, direction, size, etc. of an article identification display section from image data.

(1)) Conventional technology In general, as a method for automatically identifying items, each item is marked with identification characters or symbols (t i =) in advance, and the item is imaged with a tele-hi camera or the like and ζ characters are placed on each item. - A method is used to identify objects by recognizing their shapes.

There are various methods for character/figure recognition technology, but in any case, when recognizing characters and symbols from captured image data, character/figure recognition is performed based on image data within a specific range on the screen. Therefore, an item identification display section consisting of characters or symbols is provided at a specific position of an item in a specific direction, and the item is imaged while being placed in a specific positional relationship with respect to a television camera.

(c1) The problem that the invention attempts to solve is, for example, when packages to which destination code number lists have been affixed in advance are automatically sorted by a conveyor, the destination code number list must be placed in a specific direction at a specific location of something. If the destination code number table is not always affixed, the baggage is not always transported in a specific direction on the conveyor, and the shape of the baggage itself is not constant, The positional relationship is completely random.

In such a case, it is necessary to capture an image so that almost the entire baggage is on the screen (52 persons), extract the image data of the destination code number table from the obtained image data, and then recognize it. the position of characters and symbols to be recognized on the screen;
It is extremely difficult to extract characters and symbols to be recognized in a short time from a state where neither direction nor size is specified.The purpose of this invention is to extract characters and symbols to be recognized from image data. An object of the present invention is to provide a method for detecting a display part for identifying an article, which can extract the position, direction, and size of a product in a short time, and can be applied to applications requiring real-time processing. be.

Means for Solving the Problems fd1 The method for detecting an article identification display part according to claim 1 of the present invention provides a method for detecting an article identification display part according to claim 1 of the present invention. A mark consisting of a concentric circular pattern is provided on the article, an image is taken of an area including the small identification table pattern and the mark, and the image data is stored in a first frame memory;
In this first frame, a pattern in a specific axis direction is extracted from the memory and stored in a second frame memory, and a pattern arrangement based on the marks is extracted from the second frame memory in an axial direction perpendicular to the specific axis direction. The present invention is characterized in that the position of the mark in the captured image data is determined by this method, and the article identification display section is detected based on the position of the mark.

The method for detecting an article identification display part according to claim 2 of the present invention comprises an article identification display pattern and an approximately circular or approximately concentric circular pattern in a predetermined positional relationship with respect to the article identification display pattern. At least one main mark and a sub mark having a certain positional relationship with respect to the main mark are provided on the article, and an image is taken of an area including the identification display pattern and the main mark and the sub mark. Data is stored in a first frame memory, a pattern in a specific axis direction is extracted from the first frame memory and stored in a second frame memory, and the main mark is extracted from the second frame memory in an axial direction perpendicular to the specific axis direction. The position of the main mark in the captured image data is determined by extracting the pattern array according to It is characterized in that the article identification display section is detected by

The article identification display part detection device according to claim 3 of the present invention includes a display pattern for article identification as well as a different pattern consisting of substantially circular or substantially concentric circular patterns in a fixed positional relationship with respect to the display pattern for article identification. Two marks are provided on the article, the identification display pattern and the second mark.
A range including two marks is imaged, the image data is stored in the first frame memory, a pattern in a specific axis direction is extracted from the first frame memories 1 and 2, and stored in the second frame 1 norm memory. 1. Find the positions of the two marks in the captured image data by extracting the pattern arrangement of the two marks from the two-frame memory in the axis direction perpendicular to the specific axis direction; The article identification display section is detected based on the positions of the two marks.

Furthermore, the method for detecting an article identification display part according to claim 4 of the present invention is such that the article identification display pattern and the article identification display pattern are arranged in a two-dimensional direction in a predetermined positional relationship with respect to the article identification display pattern. A plurality of marks consisting of circular or substantially concentric circular patterns are provided, an image is taken of a range including the identification display pattern and the plurality of marks, and the image data is stored in a first frame memory. An image captured by extracting a pattern in a specific axis direction from a frame memory, storing it in a second frame memory, and extracting a pattern arrangement of the plurality of marks from the second frame memory in an axial direction orthogonal to the specific axis direction. The present invention is characterized in that the positions of the plurality of marks in the data are determined, and the article identification display section is detected based on the positions of the plurality of marks.

(e) Function The horn method for detecting a display part for product identification of the present invention includes a mark consisting of a pattern of substantially circular or substantially concentric circular patterns in a fixed positional relationship with respect to the display pattern for product identification, as well as a display pattern for product identification. is attached to the article, a range including the mark is imaged together with an identification display pattern, and the position of the mark is determined from the image data.

The article identification display section is detected based on the position of this determined mark.

A configuration example according to claim 3 of the present invention is shown in FIG. In the same figure, A is an article to be identified, and on its specific surface there is an article identification display pattern ID and, with respect to this article identification display pattern ID45, a substantially circular or substantially concentric circular pattern with a fixed positional relationship. A television camera 61 is provided with two marks Ma and Mb having different patterns, and a television camera 61 captures an image of an area including the two marks Ma and Mb together with the product identification display pattern ID. and store the data obtained. 3 in the diagram
extracts a unique pattern appearing in the X-axis direction of the two marks Ma and Mb from the first frame memory 2 and writes it into the second frame memory 4. Furthermore, at 5 in the figure, a specific pattern arrangement of the two marks Ma and Mb is extracted from the second frame memory 4, and its position is determined as the position data of the two marks Ma and Mb.

The two marks consist of circular or concentric circular patterns and are different from each other, so that they can be easily detected independently regardless of the imaging direction. Since these two marks Ma and Mb are in a fixed positional relationship with the product identification display pattern ID, the image read into the first frame memory 2 is obtained by determining the positions of the marks Ma and Mb on the screen. The position and direction of the article identification display pattern ID in the data can be determined by the positional relationship between the two marks Ma and Mb, and the size can be determined by the distance between the two marks Ma and Mb.

In the method according to the first aspect, the position of at least one mark is determined in the same manner as described above, and the article identification display section is detected based on the position of the mark.

In the method according to claim 2, the position of the main mark consisting of a substantially circular or substantially concentric circular pattern is determined in the same manner as described above, and the position of the main mark and the separately determined position of the sub-mark are used as a reference to determine the position of the article. The identification display section is detected.

In the method according to claim 4, the positions of a plurality of marks arranged in a two-dimensional direction are determined in the same manner as described above, and the position, direction, and inclination of the article identification display section are determined from the positional relationship of each mark.

(f) Embodiment FIG. 2 shows an example in which the article identification display detection method of the present invention is applied to an automatic article sorting device.

In FIG. 2, the upper part is a belt conveyor 103, which carries a plurality of articles At, A2. A top view showing a state in which A3 is placed and cut, the lower part is a front view thereof, and a block diagram of a control unit of the automatic article sorting device. As shown in the figure, a television camera 1 is installed at a fixed position above the bell l-conveyor 103, and sequentially captures images of the top surfaces of the articles being conveyed. The image processing device 100 processes the video signal of the television camera 1 to create two 7...
− Find the position of the mark. The host computer 101 receives the position data of the two marks from the image processing device 100, recognizes the article identification display pattern from the image data obtained by imaging, and identifies the article according to the identification display. A control signal is output to the peripheral control device 102 in order to distribute the information. Further, if the display pattern for identifying the article is dirty and cannot be recognized, the belt conveyor is stopped and an alarm is sounded. In that case, the staff member reads the display pattern and presses the keyboard 10.
You can directly input the destination code number from 4.

Next, the configuration of the image processing device 100 will be explained.

FIG. 3 is a block diagram of the image processing apparatus 100, and FIGS. 4 and 5 are diagrams showing a part of the block diagram in detail.

In FIG. 3, the television camera 1 is fixed at a fixed position above the belt conveyor as shown in FIG. For example, one screen worth of image data consisting of one pixel and three bits is created and written into the first frame memory 2. The binarization circuit 11 binarizes 3-bit data sequentially output from the first frame memory 2 using a predetermined threshold value and outputs the binarized data to the first shift register 12 as 1-bit data per pixel. Note that the differential value of the image data may be binarized.

In the figure, 14 is a clock generator that generates a clock signal, and the address counter 15 sequentially updates the read address of the first frame memory 2 using the clock signal. If this reading direction is the X-axis direction, the first shift register 12 stores a specific bit of the image data in the X-axis direction.
The contents of the number will be read. Note that each time the address counter 15 selects a read address of the first frame memory 2, a shift pulse (not shown) is applied to the first shift register 12, and data overflowing from the right side in the figure is lost.

13 in the figure is PAL (Programmable
This is a logic gate circuit consisting of an array logic (Array Logic), and the contents of the first shift register 12 are
When a specific pattern of two marks is obtained, data representing that pattern is output and written into the second frame memory 4. In the second frame memory 4, the same address as the read address of the first frame memory 2 is selected as a write address by the address counter 15, similarly to the first frame memory 2. In this way, data representing a specific pattern in the X-axis direction of the first frame memory 2 is obtained in the second frame memory 4.

Thereafter, the address counter 18 is set to the second frame memory 4.
The contents of the second frame memory 4 are sequentially output to the second shift register 16. At this time, the address counter 18 reads Y from the second frame memory 4.
Address selection is performed so that the contents in the axial direction are read. Therefore, a pattern is obtained in the second shift register 16 in which data representing a specific pattern in the X-axis direction written in the second frame memory 4 is arranged in the Y-axis direction.

In the figure, 17 is a logic gate circuit consisting of a PAT, similar to 13 above, and when the contents of the second shift register 16 become a specific pattern arrangement based on marks consisting of a substantially circular or substantially concentric circular pattern, signal 3a,
Outputs 3b. When Sa is output, the latch circuit 19
latches the contents of address counter 18 at that time, and s
When b is output, the latch circuit 20 latches the contents of the address counter 18 at that time. In this way, the positions of the two marks are determined as address data of the first frame memory (-address data of the second frame memory), and this is output to the host computer J01 (see FIG. 2).

As described above, detection of the position of a mark made of a substantially circular or substantially concentric circular pattern is performed in real time. Therefore, the image data in the first frame memory 2 is refreshed to captured image data at each predetermined mark position detection timing. Once the positions of the two marks have been detected, refreshing of the first frame memory is stopped and the two mark position data are output to the host computer.

Thereby, the host computer reads and recognizes the article identification pattern from the first frame memory based on the two mark position data.

In this way, detection of the mark and reading of the identification display pattern of the article are performed in different modes.

Next, the specific configuration and operation of each part shown in FIG. 3 will be explained.

FIG. 7 shows the PA from the first shift register 12 shown in FIG.
An example of signal processing up to L17 is schematically shown. 7th
In the figure, X indicates the pattern of marks Ma written in the first frame memory, and when this is read out in the X-axis direction and set in the first shift Hz register,
Shows the contents of the shift register. In addition, D is PA when the contents of the first shift register become the contents shown in X.
Each shows 3-bit data output from L13. For example, the content of the first shift 1 register is 0000.
When 01111111100000, PAL 1
00000001 is output from 3 and 1. When the content of the first shift register is 000111111111111000, the PAL 13 outputs 0ooooioio. By reading the contents of the first frame memory in the X-axis direction and inputting them continuously to the first shift register in this way, the second frame memory is read out in the X-axis direction, and the second frame memory is read out in the X-axis direction and inputted continuously to the first shift register. /- The data shown in FIG. 7 is written to the address of the memory.

The data written in the second frame memory is sequentially read out in the Y-axis direction and sequentially input into eight second shift registers 16 having a predetermined number 1) i. The PAL 17 outputs a mark detection signal Sa when the contents of the second shift register 16 become a specific pattern arrangement of marks Ma.

Figure 4 shows the first shift register 12 and PA shown in Figure 3.
It is a figure which shows the specific example of L13. The upper part of the figure shows an example of the Ma pattern written in the first frame memory 2 and the contents read out in the X-axis direction and set in the first shift register 12. The lower part of the figure shows the configuration of the first shift register 12 and the PAL 13 connected to its output.

As shown in the figure, an input gate 1-A is connected to each bit output of the first shift register 12, and an inverted signal and a non-inverted signal are obtained for each bit. 7 on this signal line
For example, Li,
Lk and Lo are provided, and each line forms an AND gate with a maximum of 28 people depending on whether or not the matrix intersections are connected. In the figure, * represents the intersection of 7 trisox; in this example, when the pattern in the row Xi at the center of the mark Ma is set in the shift register 12, the output of the line Li becomes 1, and the pattern in the Xk row is the shift register] ,
When set to 2, the output of line Lk becomes 1, and when the pattern of row XO is set to the shift register 12, the output of line Lo becomes 1.
Gate input is selected. However, since the size of the mark written to the first frame memory is not constant, the black area and white N area of each AND gate input must have a width so that the same AND gate output can be obtained within a certain range. There is. (bi of shift register 12 without a fist in the figure)
In addition, when dealing with marks of different sizes, several other types of A are used to extract a specific pattern according to the size.
The ND gate circuit (not shown) may be configured in advance by a matrix.

The output states of the plurality of AND gates configured in this way are output as 3-bit data, and this is written into the second frame memory 4 shown in FIG. 3.

FIG. 6 shows two pattern arrays written in the second frame memory 4 and two pattern arrays written in the first frame memory 4.
It is shown overlapping the two mark patterns. This example shows an example in which each mark is coded for eight lines in the X-axis direction. To be precise, the write address of the pattern written in the second frame memory 4 is written at a position that is shifted (offset in the X direction) from the address of the mark pattern written in the first frame memory. However, here, in order to clarify the correspondence with the mark, the code representing the pattern is superimposed on the center of the mark.

FIG. 5 shows the second shift register 16 and PA shown in FIG.
The configuration of L 17 is shown. In FIG. 5, the second shift register 16 is eight shift registers each consisting of several tens of bits, and the D
Input the 8-bit data from O to D7 bit by bit. P.A.
In L17, vertical lines indicate output lines of predetermined bits of each shift register, and horizontal lines indicate input lines of an AND gate which constitutes an AND gate by a matrix with each output line. The outputs of the plurality of AND gates are logically summed by ORa to output a signal Sa, and are logically summed by ORb to output a signal sb.

Now, for example, the code pattern array CO, C2C5, C4, of the mark Ma shown in FIG. 6 is stored in the shift register 16.
C5, C5, C4, C3, C2° When CO is set, the output line of the shift register 16 and, for example, line L
The output of an AND gate consisting of a matrix with a becomes 1, and the signal Sa becomes 1. Since such a code pattern arrangement differs depending on the size of the mark written in the first frame memory and the read position in the X-axis direction, multiple types of these code arrangement are determined in advance, and when a match is made with any of them, A, N so that if the code pattern arrangement is based on marks Ma, the signal Sa will be 1, and if the code pattern arrangement is based on marks Mb, the signal sb will be 1.
The connection state of the intersections of the matrix that constitutes the D gate is determined.

2 in the image data written to the first frame memory by the image processing apparatus 100 shown in FIG.
Find the position of the two marks as the address of the first frame,
This is output to the host computer 101. However, the center address of the mark pattern written in the first frame memory 2 and the center address of the code pattern array written in the second frame memory 4 are
There is a constant offset in the axial direction, and the contents of the address counter 18 when the signals Sa and Sb are output from the logic gate circuit 17 are constant in the Y-axis direction from the center address of the code pattern array in the second frame memory 4. There is an offset of Therefore, host computer 101
After correcting this offset, the image data of the display pattern for article identification to be recognized is read out from the first frame memory in the image processing apparatus 100 and recognized. that time,
The direction of the article identification display pattern is normalized from the positional relationship of the two marks, and the size of the article identification display pattern is normalized from the distance between the two marks in the screen.

In this example, the article identification display pattern is placed between two mar... 7, but in addition, for example, as shown in FIG. It is also possible to separate them into two rows and arrange them together with the pattern.

In addition, the present invention is not limited to two marks that serve as a reference for detecting the position of the product identification display section.If a single mark is provided, it is determined in advance from 1, its φ, and - marks. It is possible to extract display patterns for article identification from locations separated by a certain distance. At that time, for example, as shown in FIG.
If the main mark is set as the main mark and the sub mark Ms is provided in a fixed positional relationship with respect to the main mark M, then the article identification display part can be detected based on the positions of the main mark M and the sub mark Ms. be able to.

In this embodiment, the article identification display pattern is arranged in a position parallel to the television camera I.6. However, for example, if the surface on which the product identification display pattern is attached is tilted with respect to the optical axis of the camera, the product identification display pattern may be The accuracy of character/figure recognition can also be improved by also detecting the inclination angle of the pattern and normalizing it.

For example, as shown in Figure 10G, 2.2 marks M1,
Ma is placed in the X-axis direction, marks Ml and M2 are placed in the Y-axis direction, and three marks Ml are placed.

M2. From the position Ma, it is possible to correct the distortion of the display pattern for product identification due to the inclination of the product. That is,
Image distortion occurs when the item is tilted, but
Since the parallel relationship between xi-xi and X2-X2, the parallel relationship between Y2-Y2 and Y3-Y3 with respect to yi-yi, and the interval ratio of each parallel line are constant, from these relationships each You can find the center of a number.

(Effects of the Phantom Invention As described above, according to the present invention, the position and direction of the item identification display pattern included in the imaged screen can be detected in a short time. Even if the positional relationship of the product identification display pattern to the product identification display pattern is not unified, the image data of the product identification display pattern can be extracted and recognized from the image data in a short time.
can be done

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing an example of the configuration of the present invention. FIG. 2 is a diagram showing the configuration of an automatic article sorting device according to an embodiment of the present invention. FIG. 3 is a block diagram of the image processing bag 2100 in FIG.
FIG. 2 is a diagram showing the configuration of a part of the diagram. FIG. 6 is a diagram showing a storage example of the first frame memory and the second frame memory. FIG. 7 is a schematic diagram showing an example of co-image processing of signal 3 in the image processing apparatus. Do Figures 8 to 10 relate to other embodiments?
- is a diagram showing an example. 1-Te1. /Vicamera, 13.17-Logic gate circuit (PAL), A1. A2.
A3-article, ID-item identification display baton, M, Ma, Mb-mark.

Claims (4)

[Claims] (1) Along with the product identification display pattern, a mark consisting of a substantially circular or substantially concentric circular pattern is provided on the product in a fixed positional relationship with respect to the product identification display pattern, and the identification display pattern and the mark are provided on the product. capturing an image of the included range and storing the image data in a first frame memory;
The image was captured by extracting a pattern in a specific axis direction from this first frame memory and storing it in a second frame memory, and extracting a pattern arrangement of the marks from the second frame memory in an axial direction perpendicular to the specific axis direction. A method for detecting an article identification display section, characterized in that the position of the mark in image data is determined, and the article identification display section is detected based on the mark position. (2) together with the product identification display pattern, at least one main mark consisting of a substantially circular or substantially concentric circular pattern in a fixed positional relationship with respect to the product identification display pattern; and with a fixed positional relationship with respect to the main mark; A sub mark located on the article is provided on the article, an image is taken of a range including the identification display pattern, the main mark and the sub mark, and the image data is stored in a first frame memory. The main mark in the captured image data is extracted by extracting a direction pattern and storing it in a second frame memory, and extracting a pattern arrangement of the main marks in an axial direction perpendicular to the specific axis direction from the second frame memory. , the position of the sub mark in the image data is determined based on the main mark position, and the article identification display section is detected based on the positions of the main mark and the sub mark. Identification display detection method. (3) Along with the product identification display pattern, two marks with different patterns consisting of substantially circular or substantially concentric circular patterns are provided on the product in a fixed positional relationship with respect to the product identification display pattern, and the identification display pattern is A range including the pattern and the two marks is imaged, the image data is stored in a first frame memory, a pattern in a specific axis direction is extracted from the first frame memory and stored in a second frame memory, and the image data is stored in a first frame memory. By extracting the pattern arrangement of the two marks from the frame memory in the axial direction perpendicular to the specific axis direction, the positions of the two marks in the captured image data are determined, and the positions of the two marks are used as a reference. 1. A method for detecting an article identification display section, characterized in that the article identification display section is detected by using the following method. (4) Along with the product identification display pattern, a plurality of marks consisting of substantially circular or substantially concentric circular patterns are provided in a two-dimensional direction in a fixed positional relationship with respect to the product identification display pattern, and A range including the display pattern and the plurality of marks is imaged, the image data is stored in a first frame memory, a pattern in a specific axis direction is extracted from the first frame memory and stored in a second frame memory, and the image data is stored in a first frame memory. The positions of the plurality of marks in the captured image data are determined by extracting the pattern arrangement of the plurality of marks in the axial direction orthogonal to the specific axis direction from the two-frame memory, and the positions of the plurality of marks are used as a reference. A method for detecting an article identification display section, characterized in that the article identification display section is detected by using the following steps.