KR100404306B1 - Coded pattern and method for the extraction of code from the same - Google Patents

Abstract

The present invention relates to a configuration of a pattern having code information and a coded pattern for acquiring a pattern image thereof, recognizing it, and then extracting a code.

The pattern has a circular structure in which each of the symbols constituting the pattern has a code value, each symbol is located in a two-dimensional lattice structure, and the code value is determined according to the symbol size. Is a combination of a grid having a plurality of angles of at least a triangle, wherein the center point of each symbol is located at each vertex of the grid structure, and the center points of each symbol are equidistant from the center points of neighboring symbols.

As a result, 1) the distortion of the image is weak, 2) easy to separate from the background, 3) it is possible to form a variety of patterns, it is possible to design aesthetic patterns can be applied to the field having a rejection bar code 4) the structure of the pattern has a structure that can be analyzed by a low-resolution camera, and 5) the pattern has a resistance to noise.

Description

Coded pattern and its code extraction method {CODED PATTERN AND METHOD FOR THE EXTRACTION OF CODE FROM THE SAME}

The present invention relates to a coded pattern and a method of extracting a code of a pattern having code information, a pattern image obtained by recognizing a pattern image thereof by a camera, and then extracting a code.

In general, code patterns that allow an image to be acquired by a camera or scanning means and then extract code from the image include maxicode, data matrix, code one, and discard code. Two-dimensional barcodes such as Vericode, codablock, array tag, Phillips dot code, softstrip code, QR code, etc. may be representative examples.

These two-dimensional barcodes are currently used in a variety of applications throughout the industry and real life, and as a representative example thereof, are used to update inventory records at the same time as sales in order to increase the management efficiency in merchandise stores such as supermarkets.

In addition, it is used for labeling various documents, and is actively used for checking the mail moving status.

However, since the two-dimensional bar code has a bar or a connected square shape, errors in the code extracted therefrom are prone to distortion of shapes for movement, tilt, and defocus.

In addition, in order to obtain a barcode image by the camera and extract a code therefrom, a camera having a relatively high resolution must be used, and there is a problem in that a barcode may be rejected due to a religious position.

Accordingly, in view of the above problems, the present invention provides a coded code such that each code symbol has a circular basic figure, has a different code value according to the size of the symbol, and also allows each code symbol to be located in a certain lattice structure. Its purpose is to provide a customized pattern.

Another object of the present invention is to provide a method of extracting a code by pattern recognition by acquiring an image of a coded pattern generated by the present invention.

1 shows an example of a coded pattern according to the present invention.

2 shows a state in which a code symbol according to the invention is in a lattice structure with a certain rule;

3 shows an example of code symbols of different sizes.

4 is a diagram showing an example of an order in which codes are recorded based on reference symbols.

5 shows an example of a code value according to each size of a code symbol.

6 illustrates an apparatus for extracting code from a coded pattern of the present invention.

7 illustrates a method for extracting code from a coded pattern of the present invention.

8 is a view for explaining a circular cluster detection process.

9 through 11 illustrate various examples of the inventive coded pattern.

<Explanation of symbols for the main parts of the drawings>

100: reference symbol 110: symbol

120 grid structure 130 center point

200: image acquisition unit 210: processor

220: memory

The present invention coded pattern for achieving the above object,

In the coded pattern,

Each of the symbols constituting the pattern has a circular structure, each having a code value,

Each symbol is characterized in being located in a two-dimensional lattice structure.

The code value is determined according to the size of the symbol.

The grating structure is characterized by consisting of a grating having a plurality of angles of at least triangular or more.

The center point of each symbol is located at each vertex of the lattice structure.

The center points of the respective symbols may be equidistant from the center points of neighboring symbols.

The coded pattern of the present invention configured as described above will be described in detail with reference to FIGS. 1 to 5.

1 is a diagram illustrating a coded pattern according to the present invention. Each symbol 110 having each code value is provided with at least one or more numbers, and each symbol 110 has a circular structure.

The reason why each of the symbols 110 has a circular structure is that motion distortion due to hand shake or a two-dimensional pattern is obtained when an image of a pattern formed by each symbol 110 is acquired by an image acquisition means such as a camera or a scanner. This is because the image has relatively strong characteristics such as distortion due to the tilt angle and distortion of the shape relative to the defocus when the image is acquired from the top, bottom, left, and right sides, and is particularly easy to correct the tilt.

Each symbol 110 having such a circular structure is arranged with a certain rule by the lattice structure 120 as shown in FIG. 2. The reason for being located in the triangular lattice structure is that the center point of each circular symbol is located. This is because the 130 may always have a constant equidistant distance from the center of neighboring symbols so that the most symbols can be provided in the same area.

In this case, the lattice structure may be a combination of a lattice having a plurality of angles (that is, quadrilateral, pentagon, ...) of at least triangular shape.

In addition, there is a space between each of the symbols 110, and the space is easily separated from the background because it has a structure that is all connected with a predetermined rule between the symbols.

It is obvious that the number of codes included in the coded pattern of the present invention can be further extended by the expansion of the lattice structure and the increase in the number of symbols.

Unlike the other symbols 110, the reference symbol 100 has a circular structure but has an empty ring structure inside, which is a reference position for extracting a code.

The position of the reference symbol 100 can be changed by the appointment as much as possible, the shape may also be different.

FIG. 3 is a diagram for explaining a state in which the above-described symbols 110 vary code values according to the size of the radius, and describe three code values by way of example.

Each symbol has a unique value with the largest symbol with a radius of '9', the middle symbol with a radius of '7', and the smallest symbol with a radius of '5'. .

The values 9, 7, and 5 are not code values themselves but merely values representing sizes.

4 is a view illustrating an example in which each symbol 110 of a coded pattern according to the present invention is arranged with reference to the reference symbol 100 as an example. The reference symbol 100 is used as a starting position. A symbol 110 having a series of code values in a counterclockwise direction is listed, and a symbol having a spiral shape toward the center symbol 18 is listed.

This is just one example, and the order in which they can be changed can vary as promised.

5 shows a state in which each symbol 110 having a code value is extracted based on this order of listing.

That is, as described above, the largest symbol having a radius of '9' has a code value of '2', and the middle symbol having a radius of '7' has a code value of '1'. For example, suppose that the smallest symbol whose radius is '5' has a code value of '0'.

In this case, if the size of each symbol is more diverse, the code value may be more diverse.

The code value extracted according to the order of the above-described symbol 110 is extracted as a value of '1 2 2 2 2 2 2 2 2 2 1 2 2 1 2 0 2 1', which is a hexadecimal value. When expressed as a value of '001 010 010 010 010 010 010 010 010 010 001 010 010 001 010 000 010 001' will be represented as a data bit string.

Thus, this binary bit string will then have meaningful data.

As one example to which the present invention is applied, a coded symbol of the present invention is printed on a movie poster, and when a user thereof acquires a pattern image using a camera (web camera, a camera of an IMT terminal, etc.) or a scanner, If the data in the code refers to the web site's domain address, you can access the site immediately and get information about the film (ie title, starring actor, show time, short story of the movie, etc.). Allows you to search or download and display.

In addition, as another application example, if the coded pattern of the present invention is applied to the device to be controlled, the pattern image is obtained, the code is extracted, and the control command list of the code is downloaded from the remote server.

In other words, when a coded pattern is applied to a television, the control command list of the television is downloaded to the terminal.

Accordingly, when the user selects a specific command, the remote server provides its control command to the home server, and the home server can apply the control command to the device.

As a result, the coded pattern of the present invention can be used to automatically download desired data from a remote location, or to control a specific control target.

Its scope of application is not limited to this and can be varied or extended as much.

On the other hand, the code extraction method by the recognition of the coded pattern according to the present invention,

Performing a preprocessing process for correcting a pattern image acquired through the image capturing means, normal segmentation, and normal cluster merging process,

A circular detection process of detecting circular figures by analyzing scale invariant momentum of each merged cluster;

A circular cluster detection process of searching and grouping bundles of neighboring circular figures connected to each other within a radius of a predetermined size with respect to each circular figure when the circular figures are detected by the circular detection process;

Determining a size and position of a symbol for determining whether the detected circular clusters are located in a lattice structure and estimating the size of each of the circular clusters;

A class separation process of normalizing the size of the symbol to a most desirable size and separating the symbol into n classes according to the size;

It is characterized in that it is performed by a code extraction process for extracting a code value according to the size of the symbol by a predetermined rule from the symbols separated into the above class.

A method of extracting a code by recognizing the coded pattern of the present invention made as described above will be described in detail with reference to FIGS. 6 to 9.

First, when the image of the pattern coded by the present invention is acquired by the image acquisition unit 200, such as a camera or a scanner, the processor 210 receives the obtained image and performs preprocessing (ST300, ST310). ).

In this case, the processor 210 may be provided in a computer when the image acquisition unit 200 is a web camera, and may be provided in the case of an IMT terminal.

In addition, the processor 210 is to embed the program for pattern recognition in itself to perform the pattern recognition process of the present invention.

The preprocessing step ST310 is performed for defocusing, motion distortion, and background noise removal.

In general, image restoration is a method of obtaining information that cannot be obtained directly by improving image quality of an image that is blurred or contains a lot of noise.

This blurring of the image mainly occurs when the camera or / and subject is in motion or when the focus is not correct, and noise is caused by electrical causes, homogeneity of the CCD cell, finite size of the CCD cell and finite size when digitizing. Occurs as a cause.

Thus, such images can be effectively restored by the known Richardson-Lucy method.

In addition, image distortion due to a motion such as a hand shake can be effectively corrected by a known Kalman filter.

The median filter is used to remove the noise of the background. In order to extract the circular shape from the image, the median filter has the weakest blurring at the edge of the image, and each code of the coded pattern of the present invention is used. It is preferable to use this median filter because the symbol has a circular structure.

The median filter removes noise generated in portions other than the circular symbol by calculating a center value of surrounding pixel values at a given point and then replacing the pixel value of the current position with the center value.

The median filter can most effectively process an image that appears in the form of a spike.

The processor 210 performs the preprocessing process ST310 and then performs the segmentation process ST320. This process analyzes geometric information such as size, shape, and position in each region of the image. The known connected component labeling method is used.

This method removes parts of the image that do not require attention and analyzes them by labeling key areas sequentially.

That is, the threshold of the similar color is determined, and if the similar color value of each pixel is within the threshold, the same label is given, and if the threshold is larger than the threshold, a new label is assigned to cluster the pixels having the same label to perform segmentation.

Thereafter, the cluster merging process ST330 is performed by the processor 210, which uses a well-known Pedro's method.

This method merges these pixels into one cluster if the measure of the color between pixels assigned meaning by the segmentation process ST320 is smaller than the threshold value, and otherwise does not merge.

This process will form a cluster between a plurality of pixels.

After performing the cluster merging process (ST330), the scale invariant momentum of each cluster is analyzed. The process of invertive momentum irrespective of the rotation, movement, and scale of each cluster is performed first. Calculate

The definition of this momentum is defined by Equation 1 below.

In addition, the normalization of the momentum, that is, the normalized momentum (centralized momentum) is made by the following equation (2).

here,

to be.

Afterwards, a circular detection process ST350 is performed. A circular figure is detected by calculating a momentum invariant from the definitions of Equations 1 and 2 in which all points are applied on the same cluster, and the momentum invariant is obtained. Is made by the following equation (3).

Where circular phi_1 = , phi_2 or more is 0. When the circular figure is projected as an ellipse due to the tilt of the screen, the value of phi_2 is about 2, and phi_3 or more is 0.

When n circular figures are detected by the circular detection process ST350, the circular figures (which do not have the meaning of a code symbol until now) are connected to neighboring circular figures within a radius of a predetermined size. A circular cluster detection process (ST36) for searching a bundle of s is performed.

In a given circular figure, the radius is gradually scaled from the minimum allowable size to the maximum size, and the connection state is judged by comparing the center point with the neighboring circular figure.

That is, as shown in FIG. 8, if the center point of neighboring circles within the allowed radius (values of 5, 7, and 9) is present, the neighboring circles are formed into one cluster.

Eventually, this cluster will be a code symbol, and this plurality of circular clusters will be a coded pattern according to the present invention.

When a plurality of circular clusters are detected by the circular cluster detection process (ST360), the reference symbols are detected in the circular clusters (ST370). A circular cluster having no digital value, that is, a digital value of '0' is determined as the reference symbol.

When the distribution of the circle and the position of the reference symbol are determined by the above process, the symbol having a meaning as a code is found by determining whether the reference cluster is located in the lattice structure adjacent to the reference symbol from the center (ST380).

That is, as shown in FIG. 2, when the center point of the circular cluster neighboring the grid of equilateral triangles is positioned around the reference symbol, the circular cluster at the position is determined as a symbol, and the circular cluster determined by the symbol is defined. The size of the symbol is determined by the radius from the center point to the outer boundary.

However, since the size of the symbol determined by the above process does not have a radius of 9, 7, 5 (for example, 9.3, 7.6, 5.2, etc.), an error occurs when the symbol is applied as it is. Possession.

Accordingly, the process of separating the data into n classes (that is, three classes since the symbols having three radii of 9, 7, and 3 are taken as an example) is performed (ST390).

In other words, each symbol has a radius value of 9, 7, and 5 in an ideal case, but if it is applied to an actual image, it may not be an accurate ratio but may have a medium value.

Therefore, using a known statistical maximization (EM) method, the most probable boundary value is obtained and normalized into three classifications.

When the coded pattern as shown in FIG. 1 is recognized by the above processes ST300 to ST380, as shown in FIG. A code having meaning is extracted (ST390).

The data of the order of code extraction will be stored in the memory 220 in advance, and the order of its extraction can be changed by appointment.

That is, as shown in FIG. 5, the processor 210 extracts a value of 1 2 2 2 2 2 2 2 2 2 1 2 2 1 2 0 2 1 ', which is a ternary value, and thus a binary value' 001 010 010 010 010 010 010 010 010 010 001 010 010 001 010 000 010 001 'is converted into a data bit string and output.

9 to 11 are views showing another form of the coded pattern of the present invention. In the case of FIG. 9, the coded pattern is implemented by shaping the shape of the letter 'A'.

Again, each symbol is located in the grid structure.

In the case of FIG. 10, the pattern is embodied in the shape of a telephone. In FIG. 11, the pattern is embodied in the shape of a television.

In this way, it is possible to freely form a pattern having various shapes by determining only the positions of symbols in the lattice structure.

As described above, the coded pattern of the present invention and the method of extracting code thereof have the following effects as the information of the symbols constituting the pattern is determined by the size of a circle, and each of these symbols is formed by a grid structure having a constant distance from each other. Has

1) The distortion of the image is weak because each symbol has a circular basic figure.

2) It is easy to separate from the background because all the spaces between symbols are connected.

3) The shape of the pattern can be variously formed into a character type, a telephone type, a TV type, etc., and the aesthetic pattern can be designed, and thus it can be applied to a field having a rejection on a barcode.

4) Since the structure of the pattern has a structure that can be analyzed by a low resolution camera, it is possible to lower the cost of obtaining the object image.

5) Since the pattern is resistant to noise, the information acquisition efficiency can be improved when the pattern is used for moving images such as TV and movies.

Claims (8)

In the coded pattern, Each of the symbols constituting the pattern has a circular structure, each having a code value, Wherein each symbol is located in a two-dimensional lattice structure, wherein the code value is determined according to the size of the symbol. delete The method of claim 1, And the grating structure comprises a combination of gratings having a plurality of angles of at least triangular or more. The method of claim 1, And a center point of each symbol is located at each vertex of the lattice structure. The method of claim 1, Wherein the center points of each symbol maintain equidistant distances from the center points of neighboring symbols. In the code extraction method of the coded pattern, Performing a preprocessing process, a normal segmentation process, and a normal merging process for correcting a pattern image obtained through the image obtaining unit; A circular detection process of detecting circular figures by analyzing scale invariant momentum of each merged cluster; A circular cluster detection process of searching and grouping bundles of neighboring circular figures connected to each other within a radius of a predetermined size with respect to each circular figure when the circular figures are detected by the circular detection process; Determining a size and position of a symbol for determining whether the detected circular clusters are located in a lattice structure and estimating the size of each of the circular clusters; A class separation process of normalizing the size of the symbol to a most desirable size and separating the symbol into n classes according to the size; The code extraction method of the coded pattern, characterized in that for performing the code extraction process for extracting the code value according to the size of the symbol by a predetermined rule from the symbols separated into the class. The method of claim 6, And the lattice structure is a combination of a lattice having a plurality of angles of at least triangular or more. The method of claim 6, The center point of each circular cluster is located at each vertex of the grid structure code extraction method of the coded pattern.