CN105868816A - Label image generation method and apparatus - Google Patents

Abstract

The invention provides a label image generation method and apparatus. The method comprises the following steps: if the quantity of effective labels in an effective label image set is smaller than a first preset quantity, randomly generating a first label image; calculating a distance between the first label image and the effective label image set; and if the distance between the first label image and the effective label image set is greater than a preset distance threshold, determining that the first label image is an effective label image, and adding the first label image to the effective label image set. Therefore, there are certain distances between label images in the effective label image set, and there are quite large differences between the label images.

Description

Label image generation method and device

Technical Field

The invention belongs to the technical field of image processing, and particularly relates to a label image generation method and device.

Background

In recent years, various types of label images, such as two-dimensional code label images, have been widely used in people's lives, such as for identifying objects, paying, and the like. The label images are generated by a plurality of coding units according to a certain coding rule, and generally carry data information, which refers to relevant information of an object used for identification of the label images, such as the name of the object, a production lot number and the like.

However, in practical applications, there is also a need for using label images-for positioning. For example, when the robot moves in a room, for example, different tag images are respectively arranged on specific positions of the room, such as four walls, it can be known which wall the robot has moved to, and which direction the robot needs to turn to walk, based on the detection of the tag images. At this time, the label image may not carry data information, but only a different label image needs to be generated.

In the practical application of the above positioning type, how to generate a label image with robustness to ensure that a large number of generated label images have large differences, so as to facilitate accurate identification of the label image in practical application, is a problem to be solved urgently.

Disclosure of Invention

In view of the above problems, the present invention provides a label image generating method and apparatus for generating a plurality of label images with greater differences.

The invention provides a label image generation method, which comprises the following steps:

if the number of the effective labels in the effective label image set is smaller than a first preset number, randomly generating a first label image;

calculating a distance between the first label image and the set of active label images;

if the distance between the first label image and the effective label image set is larger than a preset distance threshold value, determining that the first label image is an effective label image, and adding the first label image to the effective label image set.

The present invention provides a label image generating apparatus, including:

the system comprises a processor and a memory, wherein the processor and the memory are connected through a bus;

wherein, a group of program codes are stored in the memory, and the processor calls the program codes stored in the memory to execute the following processing procedures:

if the number of the effective labels in the effective label image set is smaller than a first preset number, randomly generating a first label image;

calculating a distance between the first label image and the set of active label images;

if the distance between the first label image and the effective label image set is larger than a preset distance threshold value, determining that the first label image is an effective label image, and adding the first label image to the effective label image set.

According to the method and the device for generating the label images, provided by the invention, if the number of the effective labels in the current effective label image set is smaller than a first preset number required to be generated, one label image is randomly generated, the distance between the label image and the effective label image set is calculated, when the distance between the label image and the effective label image set is larger than a certain distance threshold value, the generated label image is proved to have obvious difference with other existing effective label images in the effective label image set, the generated label image is determined to be the effective label image at the moment, and the effective label image set is added into the effective label image set, so that a certain distance exists between the label images in the effective label image set, and the difference exists between the label images.

Drawings

FIG. 1 is a flowchart of a first embodiment of a label image generation method according to the present invention;

FIG. 2 is a flowchart of a second embodiment of a label image generation method according to the present invention;

fig. 3 is a schematic structural diagram of a first embodiment of a label image generating apparatus according to the present invention.

Detailed Description

Fig. 1 is a flowchart of a first embodiment of a tag image generation method according to the present invention, as shown in fig. 1, the method includes the following steps:

step 101, if the number of the effective labels in the effective label image set is smaller than a first preset number, randomly generating a first label image.

When a user needs to generate a first preset number of label images, an effective label image set D may be defined first for storing the generated effective label images. Initially, the active label image set D is empty, and when a first label image is randomly generated, the first generated label image is directly added to the active label image set. The first label image is not limited to the first label image to be generated.

In this embodiment, taking the generation of any first label image as an example, when the first label image needs to be generated, it is first determined whether the number of valid label images in the valid label image set D has reached a first preset number, and if not, one first label image is randomly generated.

Before specifically describing the random generation method of the tag image, first, the definition of the tag image is described:

suppose with m_gRepresents any one of the first label images, then m_g＝(w₀,w₁,...,w_n)，w_iFor the first label image m_gI is taken from 0 to n. Wherein, w_i＝(b₀,b₁，，，b_n|b_j∈{0,1})，b_jThe pixel value of the jth coding unit in the vector is coded for the ith column.

That is, any one of the label images is an image matrix composed of n rows and n columns of coding units, each of which is an element in the image matrix. This image matrix is composed of n column vectors, i.e., n columns of encoded vectors, each column of encoded vectors is composed of n encoding units, and each encoding unit corresponds to a pixel value of 0 or 1, i.e., a black or white binary pixel. Therefore, each coding unit corresponds to a coding cell, and is filled with black or white. Therefore, generating the label image corresponds to generating the image matrix, and then filling each coding unit in the generated image matrix with a corresponding color according to the pixel value, thereby forming the label image.

Therefore, when generating the tag image, in this embodiment, the tag image is generated in a manner of randomly generating each column of coding vectors in the tag image, and it is understood that if the tag image is supposed to be represented as being composed of n rows of coding vectors, each row of coding vectors is composed of n coding units, and the principle of randomly generating each row of coding vectors is similar to that, and is not repeated herein.

Specifically, randomly generating a first label image includes:

determining a first label image m according to equation (1)_gThe generation probability P of each column of code vectors(w＝w_i) To generate a probability P (w ═ w) according to_i) Generating a first label image m_g：

$p(w=w_i)=\frac{T\left(w_i\right)O(w_i,D)}{{\mathrm{\Σ}}_{w_{\mathrm{\γ}}\∈Z}T\left(w_{\mathrm{\γ}}\right)O(w_{\mathrm{\γ}},D)}---\left(1\right)$

Wherein, w_γFor any column of coded vectors, Z represents 2ⁿAnd (4) a coding mode. It is worth mentioning that w_γFor any column, vector is encoded, without limitation, by w_γFor the first label image m_gAny column in (2) encodes a vector, but refers to: if any column of code vectors consists of n code units, and the total number of the n code units is Z, Z code vectors are corresponding to the n code units, namely w in the column_γIs any one of the Z code vectors. The above generation probability refers to the code vector w_iFor the first label image m_gOne column encodes the probability of the vector.

Wherein, T (w)_i) Is w_iIs determined according to equation (2); o (w)_iD) is w_iThe probability of having occurred in the set of active label images D is determined according to equation (3):

$T\left(w_i\right)=1-\frac{{\mathrm{\Σ}}_{q=0}^{n-2}\mathrm{\δ}(w_i^{q+1},w_i^q)}{n-1}---\left(2\right)$

$O(w_i,D)=\left\{\begin{array}{cc}1-\frac{{\mathrm{\Σ}}_{m_g\∈D}{\mathrm{\Σ}}_{w_{\mathrm{\γ}}\∈m_g}\mathrm{\δ}(w_{\mathrm{\γ}},w_i)}{n\left|D\right|}& \left|D\right|\≠0\\ 1& \left|D\right|=0\end{array}\right.---\left(3\right)$

wherein,is w_iThe q-th edition in (1)Code unit whenWhen (w)_γ,w_i) Is equal to 0, whenWhen (w)_γ,w_i) Equal to 1, | D | is the number of valid label images in D.

The following explains the meaning of the conversion rate expressed by the above formula (2) by way of example, assuming that w_iIs a coded vector consisting of 5 coding units, the coding mode of the 6 coding units is 010110, then T (010110) ═ 4/5, wherein the conversion rate represents the w_iIn this example, the first coding unit is 0, the second coding unit is 1, the number of changes from the first coding unit to the second coding unit is 1; correspondingly, the change from the second coding unit to the third coding unit is 1 to 0, the change is changed, and the change quantity is increased by 1 to 2; changing from 0 to 1 from the third coding unit to the fourth coding unit, and changing the change quantity plus 1 to 3; changing from 1 to 1 from the fourth coding unit to the fifth coding unit without changing, and keeping the change number unchanged at 3; from the fifth coding unit to the sixth coding unit, 1 is changed to 0, and the change amount plus 1 is changed to 4. Theoretically, the 6 coding units can have a variation number of 5 at most, so T (010110) ═ 4/5. The higher the conversion rate of any column of coded vectors, the higher the identifiability of the label images, and the more accurate the identification can be ensured.

Based on the meaning of the above-mentioned formula (2) and formula (3), it should be understood for formula (1): if w is_iThe higher the conversion rate of (c), it is in the label image m_gThe higher the probability generated in (1), and conversely, the lower the probability generated; if w is_iThe higher the probability of occurrence in other label images already generated in the set of active label images, the higher it is in label image m_gShould the probability generated inThe lower and vice versa the higher the probability of generation should be.

In summary, the generation probability expressed by the formula (1) is such that the label image m is_gThe label images are different from other label images already generated in the effective label image set as much as possible.

Step 102, calculating the distance between the first label image and the effective label image set.

In the present embodiment, the first label image m_gThe distance from the effective label image set D is the minimum distance between the first label image and each other label image in the effective label image set D, and specifically, the first label image m can be calculated by the following formula (4)_gDistance L (m) from active label image set D_g,D)：

$L(m_g,D)=\underset{m_h\∈D}{\min }\left\{L(m_g,m_h)\right\}---\left(4\right)$

Wherein m is_hFor any valid label image in D, L (m)_g,m_h) Is m_gAnd m_hThe distance between the two is determined according to equation (5):

$L(m_g,m_h)=\underset{k=\{0,1,2,3\}}{\min }\left\{H(m_g,R_k(m_h\left)\right)\right\}---\left(5\right)$

wherein H () represents the Hamming distance, R_k(m_h) Represents m_hRotated k × 90 degrees.

Step 103, if the distance between the first label image and the effective label image set is greater than a preset distance threshold, determining that the first label image is an effective label image, and adding the first label image to the effective label image set.

In the present embodiment, if the first label image m_gThe distance from the effective label image set D is larger than a preset distance threshold value tau₀Then the first label image m is determined_gFor a valid label image, the first label image m_gAdded to the active label image set D.

It will be appreciated that when a label image m_gAfter adding the effective label image set D, step 101 may be continuously executed, that is, it is determined whether the number of effective labels in the effective label image set D reaches a first preset number, if yes, the label image generation process is ended, and if not, a new label image is continuously generated.

In this embodiment, the minimum difference distance τ (D) of the effective label image set D generated in the above manner can be represented as:

$\mathrm{\τ}\left(D\right)=\min \left\{\underset{m_h\∈D}{\min }\right\{S\left(m_h\right)\},\underset{m_h\≠m_l\∈D}{\min }\{D(m_h,m_l)\left\}\right\}$

wherein,

wherein, S (m)_h) For any effective label image m in D_hSelf-rotation distance of m_lIs any one of D and m_hA different active label image.

The minimum difference distance τ (D) means that the generated valid label image set D can have error detection and correction capabilities of floor [ (τ (D) -1)/2], that is, when there are code means of floor [ (τ (D) -1)/2] errors in the generated valid label image, the generated valid label image can be corrected to correspond to a correct label image. Where floor () is a floor function. In addition, the generated label image has rotation, so that the placement direction of the label image can be detected, and the robustness and the accuracy are high in practical application.

Optionally, after step 103, the following step 104 may be further included:

and 104, generating an outer frame image corresponding to the first label image, wherein the first label image is filled in the outer frame represented by the outer frame image.

In practical applications, the generated label image generally needs to be attached to an object, and the generated label image is a binary image having two colors, i.e., black and white, and if the background pattern of the attached object is not easily distinguished from the label image, it may be inconvenient for an identification device, such as an identification device provided on a robot, to be accurately positioned to the label image.

Therefore, in this embodiment, optionally, an outer frame image may also be generated on the periphery of the generated label image, where the outer frame image is formed by, for example, black coding units occupying a certain number of coding units, that is, a circle of black frame image is generated around the label image, so as to facilitate accurate positioning to the label image.

In this embodiment, if the number of active labels in the current active label image set is less than a first preset number that needs to be generated, one label image is randomly generated, and the distance between the label image and the active label image set is calculated, and when the distance between the label image and the active label image set is greater than a certain distance threshold, it indicates that there is significant difference between the generated label image and other active label images already existing in the active label image set, and at this time, it is determined that the generated label image is an active label image, and the active label image set is added to the active label image set, so that there is a certain distance between label images in the active label image set, and there is a large difference between label images.

Fig. 2 is a flowchart of a second embodiment of a label image generation method according to the present invention, as shown in fig. 2, on the basis of the embodiment shown in fig. 1, after the step 102, the following step may be further included:

step 201, if the distance between the first label image and the set of valid label images is not greater than the preset distance threshold, discarding the first label image, and adding one to the number of invalid label images.

On the basis of the embodiment shown in fig. 1, if the first label image m calculated by step 102 is the first label image m_gThe distance from the effective label image set D is not more than a preset distance threshold value tau₀First label image m_gThe first label image m generated may be discarded when the difference from other effective label images in the effective label image set D is not large enough_gThe number of invalid label images ρ is incremented by one.

Step 202, determining whether the number of the invalid label images is less than a second preset number, if so, performing step 203 and 206, otherwise, performing step 207 and 212.

And step 203, randomly generating a second label image.

Step 204, judging whether the distance between the second label image and the effective label image set is greater than a preset distance threshold, if so, executing step 205, otherwise, executing step 206.

Step 205, determining that the second label image is an effective label image, adding the effective label image into the effective label image set, and setting the number of the ineffective label images to be 0.

Step 206, discarding the second label image, and adding one to the number of the invalid label images.

After step 206, the judgment processing of step 202 is executed.

In the present embodiment, when the first label image m is generated_gAnd for the invalid label images, after adding one to the number rho of the invalid label images, judging whether the number of the invalid label images reaches a certain threshold value, namely a second preset number. If the number of the invalid label images is less than the second preset number, the preset distance threshold value tau adopted currently is indicated₀Can also be used continuously, i.e. with the distance threshold τ₀The criterion is satisfied as to whether the tag image is valid. Wherein the establishment means employing the distance threshold τ₀It may still be sufficient for the user to generate a first preset number of active tags with a large difference in distance from each other.

At this time, a next label image, that is, a second label image, is randomly generated according to the formula (1), and further referring to the embodiment shown in fig. 1, the distance between the second label image and the effective label image set D is calculated, if the distance between the second label image and the effective label image set D is greater than the preset distance threshold τ₀If the number of the label images is not equal to the preset number, the second label image is a valid label image, the second label image is added into the valid label image set D, and the number of the invalid label images is increasedAnd setting the label number to be 0, further judging whether the number of the effective labels in the effective label image set D reaches a first preset number, and if not, continuing to generate a new label image.

Conversely, if the distance between the second label image and the valid label image set D is not greater than the preset distance threshold τ₀And if the second label image is an invalid label image, adding one to the number rho of the invalid label images. If the number p of invalid label images at this time still does not reach the second preset number, the execution continues from step 203 to step 206.

It should be noted that, in this embodiment, the number ρ of invalid label images is actually a statistical number of invalid label images generated continuously, as long as one valid label image is generated before the number ρ of invalid label images does not reach the second preset number, the number ρ of invalid label images is set to 0, and at this time, the preset distance threshold τ is described₀The method is also effective, that is, the method can be used as a basis for judging whether the label image is effective or not.

In contrast, if the number ρ of invalid label images has reached the second preset number, the following processing procedure of step 207 and subsequent steps is performed:

and step 207, updating the preset distance threshold, wherein the updated distance threshold is smaller than the preset distance threshold.

And step 208, judging whether the updated distance threshold is larger than a preset cut-off distance threshold, if so, executing step 209, otherwise, ending.

In this embodiment, if the number ρ of invalid label images has reached the second preset number, it indicates that the preset distance threshold τ is set₀The second threshold is used as an invalid threshold, so that the generated label images can be guaranteed to have larger difference, but the first preset number of label images required by the user cannot be generated. Therefore, the threshold needs to be updated, e.g. to τ₁＝τ₀-1, with updated distance threshold τ₁As a subsequent birthAnd whether the label image is valid or not.

However, it should be noted that the distance threshold is not updated without limitation, because if the distance threshold is updated without limitation, although the requirement of generating a large number of label images by a user may be met, the difference between the label images cannot be better guaranteed, and therefore, in this embodiment, the distance threshold is updated with a certain limitation, if the current updated distance threshold τ is updated, the distance threshold τ is updated at a certain time₁The limited cutoff distance threshold has been reached, the tag image generation process is ended.

Conversely, if the limited cutoff distance threshold is not reached, then generation of a new label image, i.e., a third label image, may continue and an updated distance threshold τ may be employed₁As a basis for determining whether the third label image is valid.

And step 209, randomly generating a third label image.

Step 210, determining whether the distance between the third label image and the effective label image set is greater than the updated distance threshold, if so, executing step 211, otherwise, executing step 212.

And step 211, determining that the third label image is an effective label image, adding the effective label image into the effective label image set, and setting the number of the ineffective label images to be 0.

Step 212, discarding the third label image, and adding one to the number of invalid label images.

After step 212, the determination process of step 202 is continued.

The processing logic of step 209 to step 212 may refer to the foregoing description, and is not described again.

In this embodiment, by counting the number of invalid labels generated continuously and by using a distance threshold updating mechanism, a larger difference between the generated valid label images is ensured, and a reasonable tradeoff is made between a certain number of label images required by enough users.

Fig. 3 is a schematic structural diagram of a first embodiment of a label image generating apparatus according to the present invention, as shown in fig. 3, the label image generating apparatus includes:

the device comprises a processor 11 and a memory 12, wherein the processor 11 and the memory 12 are connected through a bus 13. Wherein, the memory 12 stores a group of program codes, and the processor 11 calls the program codes stored in the memory 12 to execute the following processing procedures:

if the number of the effective labels in the effective label image set is smaller than a first preset number, randomly generating a first label image;

calculating a distance between the first label image and the set of active label images;

if the distance between the first label image and the effective label image set is larger than a preset distance threshold value, determining that the first label image is an effective label image, and adding the first label image to the effective label image set.

Optionally, the processor 11 is further configured to:

if the distance between the first label image and the effective label image set is not larger than the preset distance threshold, discarding the first label image, and adding one to the number of the ineffective label images;

if the number of the invalid label images is smaller than a second preset number, randomly generating a second label image;

if the distance between the second label image and the effective label image set is larger than the preset distance threshold, determining that the second label image is an effective label image, adding the effective label image into the effective label image set, and setting the number of invalid label images to be 0;

and if the distance between the second label image and the effective label image set is not larger than the preset distance threshold, discarding the second label image, and adding one to the number of the ineffective label images.

Optionally, the processor 11 is further configured to:

if the distance between the first label image and the effective label image set is not larger than the preset distance threshold, discarding the first label image, and adding one to the number of the ineffective label images;

if the number of the invalid label images is larger than a second preset number, updating the preset distance threshold, wherein the updated distance threshold is smaller than the preset distance threshold;

if the updated distance threshold is larger than the preset cut-off distance threshold, randomly generating a third label image;

and if the distance between the third label image and the effective label image set is greater than the updated distance threshold, determining that the third label image is an effective label image, adding the effective label image into the effective label image set, and setting the number of invalid label images to be 0.

The tag image generating apparatus of this embodiment may be used to execute the technical solutions of the method embodiments shown in fig. 1 and fig. 2, and the implementation principles and technical effects thereof are similar and will not be described herein again.

In the above embodiments of the tag identification apparatus, it should be understood that the Processor may be a Central Processing Unit (CPU), other general purpose processors, an image Processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), etc. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like. The steps of a method disclosed in connection with the embodiments of the present invention may be directly implemented by a hardware processor, or may be implemented by a combination of hardware and software modules in the processor.

Those of ordinary skill in the art will understand that: all or part of the steps for implementing the method embodiments may be implemented by hardware related to program instructions, and the program may be stored in a computer readable storage medium, and when executed, the program performs the steps including the method embodiments; and the aforementioned storage medium includes: various media that can store program codes, such as ROM, RAM, magnetic or optical disks.

Finally, it should be noted that: the above embodiments are only used to illustrate the technical solution of the present invention, and not to limit the same; while the invention has been described in detail and with reference to the foregoing embodiments, it will be understood by those skilled in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some or all of the technical features may be equivalently replaced; and the modifications or the substitutions do not make the essence of the corresponding technical solutions depart from the scope of the technical solutions of the embodiments of the present invention.

Claims (10)

1. A label image generation method, comprising:

if the number of the effective labels in the effective label image set is smaller than a first preset number, randomly generating a first label image;

calculating a distance between the first label image and the set of active label images;

if the distance between the first label image and the effective label image set is larger than a preset distance threshold value, determining that the first label image is an effective label image, and adding the first label image to the effective label image set.

2. The method of claim 1, further comprising:

and if the distance between the first label image and the effective label image set is not larger than the preset distance threshold, discarding the first label image, and adding one to the number of the ineffective label images.

3. The method of claim 2, wherein after said discarding said first label image, said method further comprises:

if the number of the invalid label images is smaller than a second preset number, randomly generating a second label image;

if the distance between the second label image and the effective label image set is larger than the preset distance threshold, determining that the second label image is an effective label image, adding the effective label image into the effective label image set, and setting the number of invalid label images to be 0;

and if the distance between the second label image and the effective label image set is not larger than the preset distance threshold, discarding the second label image, and adding one to the number of the ineffective label images.

4. A method according to claim 2 or 3, characterized in that the method further comprises:

if the number of the invalid label images is larger than a second preset number, updating the preset distance threshold, wherein the updated distance threshold is smaller than the preset distance threshold;

if the updated distance threshold is larger than the preset cut-off distance threshold, randomly generating a third label image;

and if the distance between the third label image and the effective label image set is greater than the updated distance threshold, determining that the third label image is an effective label image, adding the effective label image into the effective label image set, and setting the number of invalid label images to be 0.

5. The method of any of claims 1 to 3, wherein the randomly generating a first label image comprises:

determining a generation probability P (w ═ w) of each column of encoding vectors in the first label image according to formula (1)_i) To generate a probability P (w ═ w) from the said_i) Generating the first label image m_g：

$P(w=w_i)=\frac{T\left(w_i\right)O(w_i,D)}{{\mathrm{\Σ}}_{w_{\mathrm{\γ}}\∈Z}T\left(w_{\mathrm{\γ}}\right)O(w_{\mathrm{\γ}},D)}---\left(1\right)$

Wherein m is_g＝(w₀,w₁,...,w_n)，w_iFor the first label image m_gThe ith column of code vectors, w_i＝(b₀,b₁,,,b_n|b_j∈{0,1})，b_jEncoding the pixel value of the j encoding unit in the i column of the encoding vector; w is a_γFor any column of coded vectors, Z represents 2ⁿA coding mode is planted;

wherein, T (w)_i) Is w_iIs determined according to equation (2); o (w)_iD) is w_iThe probability of having occurred in the set of active label images D is determined according to equation (3):

$T\left(w_i\right)=1-\frac{{\mathrm{\Σ}}_{q=0}^{n-2}\mathrm{\δ}(w_i^{q+1},w_i^q)}{n-1}---\left(2\right)$

$O(w_i,D)=\left\{\begin{array}{cc}1-\frac{{\mathrm{\Σ}}_{m_g\∈D}{\mathrm{\Σ}}_{w_{\mathrm{\γ}}\∈m_g}\mathrm{\δ}(w_{\mathrm{\γ}},w_i)}{n\left|D\right|}& \left|D\right|\≠0\\ 1& \left|D\right|=0\end{array}\right.---\left(3\right)$

wherein,is w_iOf the q-th coding unit whenWhen (w)_γ,w_i) Is equal to 0, whenWhen (w)_γ,w_i) Equal to 1, | D | is the number of valid label images in D.

6. The method of any of claims 1-3, wherein said calculating the distance of the first label image from the set of active label images comprises:

calculating the first label image m according to formula (4)_gDistance L (m) from the active label image set D_g,D)：

$L(m_g,D)=\underset{m_h\∈D}{min}\left\{L(m_g,m_h)\right\}---\left(4\right)$

Wherein m is_hFor any valid label image in D, L (m)_g,m_h) Is m_gAnd m_hThe distance between the two is determined according to equation (5):

$L(m_g,m_h)=\underset{k=\{0,1,2,3\}}{min}\left\{H(m_g,R_k(m_h\left)\right)\right\}---\left(5\right)$

wherein H () represents the Hamming distance, R_k(m_h) Represents m_hRotated k × 90 degrees.

7. The method according to any one of claims 1 to 3, further comprising:

and generating an outer frame image corresponding to the first label image, wherein the first label image is filled in an outer frame represented by the outer frame image.

8. A label image generation apparatus, characterized by comprising:

the system comprises a processor and a memory, wherein the processor and the memory are connected through a bus;

wherein, a group of program codes are stored in the memory, and the processor calls the program codes stored in the memory to execute the following processing procedures:

if the number of the effective labels in the effective label image set is smaller than a first preset number, randomly generating a first label image;

calculating a distance between the first label image and the set of active label images;

if the distance between the first label image and the effective label image set is larger than a preset distance threshold value, determining that the first label image is an effective label image, and adding the first label image to the effective label image set.

9. The tag generation apparatus of claim 8, wherein the processor is further configured to:

if the distance between the first label image and the effective label image set is not larger than the preset distance threshold, discarding the first label image, and adding one to the number of the ineffective label images;

if the number of the invalid label images is smaller than a second preset number, randomly generating a second label image;

if the distance between the second label image and the effective label image set is larger than the preset distance threshold, determining that the second label image is an effective label image, adding the effective label image into the effective label image set, and setting the number of invalid label images to be 0;

and if the distance between the second label image and the effective label image set is not larger than the preset distance threshold, discarding the second label image, and adding one to the number of the ineffective label images.

10. The tag generation apparatus according to claim 8 or 9, wherein the processor is further configured to:

if the distance between the first label image and the effective label image set is not larger than the preset distance threshold, discarding the first label image, and adding one to the number of the ineffective label images;

if the number of the invalid label images is larger than a second preset number, updating the preset distance threshold, wherein the updated distance threshold is smaller than the preset distance threshold;

if the updated distance threshold is larger than the preset cut-off distance threshold, randomly generating a third label image;

and if the distance between the third label image and the effective label image set is greater than the updated distance threshold, determining that the third label image is an effective label image, adding the effective label image into the effective label image set, and setting the number of invalid label images to be 0.