WO2024051427A1

WO2024051427A1 - Coin identification method and system, and storage medium

Info

Publication number: WO2024051427A1
Application number: PCT/CN2023/111654
Authority: WO
Inventors: 徐青松; 李青
Original assignee: 杭州睿胜软件有限公司
Priority date: 2022-09-09
Filing date: 2023-08-08
Publication date: 2024-03-14
Also published as: CN115359302A

Abstract

Provided in the present invention are a coin identification method and system, and a storage medium. The method comprises: acquiring a coin picture, which is provided by a user; performing identification on the coin picture to acquire a coin region, and performing segmentation to acquire coin region pictures; performing identification on the coin region pictures to acquire a plurality of feature regions; performing identification on each of the plurality of feature regions to acquire a plurality of pieces of coin feature information; and synthesizing the plurality of pieces of coin feature information to acquire classification information of a coin. The coin identification method provided in the present invention can improve the accuracy of coin identification.

Description

Coin identification method, system and storage medium

Technical field

The invention relates to the field of computer technology, and in particular to a coin identification method, system and storage medium.

Background technique

Coins can be divided into different types such as coins and commemorative coins. Whether they are circulating currencies, collected coins or commemorative coins, there is a certain need for identification. However, due to reasons such as less circulation or natural rarity of some coins, samples are scarce. At the same time, Due to the unique materials of coins, different textures between old and new, problems with light reflection, uncertain user shooting environment, and complex backgrounds, even the same coins have various characteristics, and the difference between user coin images and standard coin images is relatively large. Large, different currencies have similar shapes, and the key distinguishing features are textures, text and other features. Therefore, the shooting quality has a great impact on recognition. A slight blur will lead to the loss of key information. At the same time, there are different situations where the shooting is too far or the resolution is low. Therefore, Coin recognition is technically difficult, resulting in poor recognition accuracy of existing coins.

Contents of the invention

One of the purposes of this disclosure is to provide a coin identification method, including:

Get the coin image provided by the user;

Identify the coin image to obtain the coin area, and segment the coin area image to obtain the coin area image;

Identify the coin area pictures and obtain multiple feature areas;

Respectively identify the multiple feature areas to obtain multiple coin feature information;

Combining the feature information of the multiple coins, the classification information of the coins is obtained.

In some embodiments, identifying the coin image to obtain the coin area includes: applying UNet image segmentation processing to the coin image to obtain the coin area.

In some embodiments, identifying the coin image to obtain the coin area, and segmenting the coin area image to obtain the coin area include:

Identify the coin area according to the area detection model, obtain the coin mask area, and classify Cut to get the coin area picture.

In some embodiments, the method further includes: performing ellipse detection and perfect circle correction on the coin area image.

In some embodiments, the method further includes: performing image preprocessing on the coin area image obtained by the segmentation to obtain a coin area image with background and noise removed;

Identifying the coin area picture and obtaining multiple characteristic areas includes: identifying the coin area picture that has undergone the preprocessing and obtaining multiple characteristic areas.

In some embodiments, the method further includes: performing image preprocessing on the coin image provided by the user to obtain the coin image with background and noise removed;

Recognizing the coin image to obtain the coin area, and segmenting the coin area image to obtain the coin area image includes: identifying the pre-processed coin image to obtain the coin area, and segmenting the coin area image to obtain the coin area image.

In some embodiments, after the coin area image is obtained by segmentation, the method further includes: performing post-processing on the coin area image with a CV algorithm to remove excess noise.

In some embodiments, obtaining the coin picture provided by the user includes: providing at least two windows for the user to select on the interactive interface, so that the user can respectively choose to provide pictures of different sides of the coin currently to be identified.

In some embodiments, it is determined whether the coin image provided by the user belongs to the front or the back according to a pre-trained front and back recognition model.

In some embodiments, identifying the coin area picture and obtaining multiple feature areas includes: identifying the coin area picture according to a pre-trained feature area recognition model to obtain multiple feature areas.

In some embodiments, a self-supervised pre-training backbone model based on contrastive learning is used to identify the plurality of feature areas respectively and obtain feature information of multiple coins.

In some embodiments, the combination of multiple coin characteristic information and obtaining the classification information of the coin includes: using a similarity comparison method to obtain the comprehensive similarity of the coin, and classifying the coins according to the comprehensive similarity. Classification information is sorted.

According to another aspect of the present disclosure, a coin identification system is proposed, including a processor and storage A program is stored in the memory, and when the program is executed by the processor, the coin identification method as described above is implemented.

According to another aspect of the present disclosure, a storage medium is proposed on which a program is stored, which when executed implements the coin identification method as described above.

Other features and advantages of the present disclosure will become more apparent from the following detailed description of exemplary embodiments of the present disclosure with reference to the accompanying drawings.

Description of the drawings

The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments of the disclosure and, together with the description, serve to explain principles of the disclosure.

The present disclosure may be more clearly understood from the following detailed description with reference to the accompanying drawings, in which:

Figure 1 shows a schematic flowchart of a coin identification method according to an embodiment of the present invention.

Figure 2 shows a schematic structural diagram of a coin identification system provided by an embodiment of the present invention.

Note that in the embodiments described below, the same reference numerals are sometimes commonly used between different drawings to represent the same parts or parts having the same functions, and repeated description thereof is omitted. In some instances, similar numbers and letters are used to identify similar items so that, once an item is defined in one figure, it does not require further discussion in subsequent figures.

In order to facilitate understanding, the positions, dimensions, ranges, etc. of each structure shown in the drawings and the like may not represent the actual positions, dimensions, ranges, etc. Therefore, the present disclosure is not limited to the positions, dimensions, ranges, etc. disclosed in the drawings and the like.

Detailed ways

Various exemplary embodiments of the present disclosure will be described in detail below with reference to the accompanying drawings. It should be noted that the relative arrangement of components and steps, numerical expressions, and numerical values set forth in these examples do not limit the scope of the disclosure unless otherwise specifically stated.

The following description of at least one exemplary embodiment is merely illustrative in nature and is in no way intended to limit the disclosure, its application or uses. That is, the structures and methods herein are shown in an exemplary manner to illustrate different embodiments of the structures and methods in the present disclosure. However, this Those skilled in the art will understand that they are merely illustrative of exemplary ways in which the disclosure may be practiced, and are not exhaustive. Furthermore, the drawings are not necessarily to scale and some features may be exaggerated to illustrate details of specific components.

Techniques, methods and devices known to those of ordinary skill in the relevant art may not be discussed in detail, but where appropriate, such techniques, methods and devices should be considered part of the authorized specification.

In all examples shown and discussed herein, any specific values are to be construed as illustrative only and not as limiting. Accordingly, other examples of the exemplary embodiments may have different values.

Figure 1 shows a schematic flowchart of a coin identification method according to an embodiment of the present invention. This method can be implemented in an application program (app) installed on a smart terminal such as a mobile phone or tablet computer. As shown in Figure 1, the method includes:

Step S100: Obtain the coin image provided by the user;

Step S200: Identify the coin image to obtain the coin area, and segment the coin area image to obtain the coin area image;

Step S300: Identify the coin area image and obtain multiple feature areas;

Step S400: Identify the multiple feature areas respectively and obtain multiple coin feature information;

Step S500: Combine the feature information of the multiple coins to obtain the classification information of the coins.

In some embodiments, identifying the coin image to obtain the coin area includes: applying UNet image segmentation processing to the coin image to obtain the coin area. The UNet image segmentation adopts a pre-trained neural network model, which can identify coin images to obtain coin areas, and perform foreground segmentation on coin images to obtain coin areas. In addition, other image segmentation processing methods can also be used, which is not the case in this disclosure. Make restrictions.

The coin area is identified according to the area detection model, the coin mask area is obtained, and the coin area image is segmented and obtained. You can use the coin area obtained by the above UNet image segmentation process and directly crop it to obtain the coin area picture. You can also identify the coin area based on the pre-trained coin area detection model to obtain a more accurate coin area and crop it. Get a picture of the coin area.

In some embodiments, the method further includes: performing ellipse detection and perfect circle correction on the coin area image. The actual processing is to perform ellipse detection on the coin area picture, obtain the equation of the ellipse, and then calculate the pixel mapping parameters to restore the circle according to the ellipse equation, perform affine transformation and interpolation processing, and convert the coin ellipse in the user's coin picture into A perfect circle shape, thereby completing the perfect circle correction process, which facilitates later feature recognition processing and corresponding content recognition.

In some embodiments, the method further includes: performing image preprocessing on the segmented coin area image, obtaining a coin area image with background and noise removed, and then identifying the preprocessed coin area image, Get multiple feature areas. Process the segmented coin area image into a clean, background-free, noise-free image, minimizing the interference of redundant information, such as tables or other coins in the background of the coin image and other non-foreground background factors of the coin area to be identified. It will cause certain interference to the recognition. Removing such background interference factors can accurately obtain the image of the coin area to be identified, so that when the characteristic area of the coin area image is subsequently identified, the background area will not be recognized and cause interference.

In some embodiments, the method further includes: performing image preprocessing on the coin image provided by the user, obtaining the coin image with the background and noise removed, and then identifying the preprocessed coin image to obtain the coin area, and segmenting Get a picture of the coin area. In this embodiment, the preprocessing step is advanced to after obtaining the coin image provided by the user, and the user's coin image is directly preprocessed into a clean, background-free, and noise-free image, so as to minimize the Interference from redundant information. Afterwards, the pre-processed coin image is identified to obtain the coin area according to the pre-trained area recognition model (such as YOLO and other models), and the coin area image is segmented to obtain the coin area image. Advancing the preprocessing steps can also remove the interference of background factors and improve the accuracy of recognition.

In some embodiments, after the coin area image is obtained through the segmentation, the coin area image is post-processed with a CV algorithm to remove excess noise. Segmentation to obtain the coin area image may include interference factors such as redundant image blocks. The edges of the coin area can be obtained through edge detection models or algorithms, such as Hough circle detection algorithms, pre-trained coin edge detection models, etc. to obtain the coins. Rounded edges can also be assisted by morphological processing methods to remove excess noise interference in the coin area and obtain a more stable coin mask area. At the same time, coin pictures with more interference factors can be re-invested into the training model as supplementary samples, so as to obtain a more accurate recognition model for identifying coin areas.

In some embodiments, obtaining the coin picture provided by the user includes: providing at least two windows for the user to select on the interactive interface, so that the user can respectively choose to provide pictures of different sides of the coin currently to be identified. The user-side interactive interface guides the user to provide pictures of the front and back of the coin, which can facilitate the acquisition of more coin features and make subsequent identification processing more accurate. In addition, when the user only provides a picture of one side of the coin or multiple identical or different pictures, Identification can be performed, and this disclosure does not limit this.

In some embodiments, it is determined whether the coin image provided by the user belongs to the front or the back according to a pre-trained front and back recognition model. The front and back sides of coins are classified through the image binary classification method, and the coin front and back sides classification and recognition model is obtained through the pre-established sample training model. By identifying and confirming the classification of the front and back sides of the coin pictures provided by the user, the characteristic areas of the coins on the pictures can be classified and identified more accurately, thereby improving the accuracy of coin recognition.

In some embodiments, identifying the coin area picture and obtaining multiple feature areas includes: identifying the coin area picture according to a pre-trained feature area recognition model to obtain multiple feature areas. Different coins have similar basic image structures and can be roughly classified, such as portraits or patterns on the front, patterns on the back, year, text (letters or numbers), etc., which can be disassembled and classified. Due to the various problems described in the background art, the overall recognition accuracy of coins in the prior art is low. However, it is possible to split the coin features into multiple pre-classified feature areas for separate identification, and to obtain the classification information of the coins through comprehensive judgment. Significantly improve the accuracy of coin recognition. The pre-trained feature area recognition model performs area recognition, detection and cropping processing on the key common areas of the coins, and processes the overall features of the coins into multiple different feature areas to facilitate subsequent feature recognition.

In some embodiments, a self-supervised pre-training backbone model based on contrastive learning is used to separately identify multiple feature areas and obtain feature information of multiple coins. By extracting features from multiple different feature areas of the coin, and then identifying or retrieving them, the corresponding feature identification information is obtained. The unique distinguishing features between coins are textures, text, etc. The general pre-training model is less effective. The core difficulty lies in the small number of training images. Pre-training is done through self-supervision based on contrastive learning, similarity comparison is performed, and the correspondence is obtained. The self-supervised pre-training backbone model based on contrastive learning can significantly improve the recognition accuracy. Conduct fine-tuning and regression testing on the self-supervised pre-training backbone model based on contrastive learning, adjust parameters and test sample sets, and obtain bones with higher recognition accuracy. dry model.

In some embodiments, the combination of multiple coin characteristic information and obtaining the classification information of the coin includes using a similarity comparison method to identify the characteristic information of multiple different characteristic areas of the coin, and through the combination with the standard The characteristic information is compared with the obtained similarities, and then the different similarities of the multiple characteristic areas are combined to obtain the comprehensive similarity of the coin, and the classification information of the coins is sorted according to the comprehensive similarity. The recognition results of one or more coins with the highest similarity matching degree can be output and displayed to the user, or all coin classification information can be output and displayed to the user in order of similarity, so that the user can browse and select. The similarity between each feature area and standard feature information can be obtained by searching in a preset standard feature information database or using pre-trained model recognition. Calculating the comprehensive similarity of coins can also be done by setting different weights according to the information of different feature areas.

In some embodiments, the recognition results of each of the multiple feature areas and the confidence of the recognition results are obtained according to the self-supervised pre-training backbone model based on contrastive learning (the confidence is the current value obtained by model recognition). The credibility of the identification result), thereby comprehensively obtaining the overall confidence of the identification result of the coin to be identified, and at the same time, combined with the comprehensive similarity of the coin to be identified that has been obtained above, the overall confidence and comprehensive similarity The probability score of the current recognition result of the coin to be recognized is obtained through weighted calculation, and finally the coin recognition results are sorted and output and displayed according to the probability score. Calculating the overall confidence of the coin can also be done by setting different weights according to the information in different feature areas.

In some embodiments, after the above-mentioned coin recognition results are sorted according to similarity or likelihood score, the year recognition results of the coins to be recognized are obtained, filtered according to the closest coin year number, and then the recognition results are output, for example, the same coins The recognition results have multiple year versions, and the recognition results of the same year or the closest year version as the current coin to be recognized are filtered and output are displayed.

Based on the same inventive concept, the present invention also proposes a coin identification system, which includes a processor and a memory. A program is stored on the memory. When the program is executed by the processor, the coin identification method as described above is implemented. Please refer to Figure 2, which is a schematic structural diagram of a coin identification system provided by an embodiment of the present invention. As shown in Figure 2, the coin identification system includes a processor 301, a communication interface 302, memory 303 and communication bus 304.

The processor 301, the communication interface 302, and the memory 303 complete communication with each other through the communication bus 304.

The memory 303 is used to store computer programs.

When the processor 301 is used to execute the program stored in the memory 303, it implements the following steps:

Get the coin image provided by the user;

Identify the coin area pictures and obtain multiple feature areas;

For the specific implementation and related explanations of each step of the method, please refer to the method implementation shown in Figure 1 above, and will not be described again here.

In addition, other implementations of the coin identification method implemented by the processor 301 executing the program stored on the memory 303 are the same as the implementations mentioned in the foregoing method implementation section, and will not be described again here.

The communication bus 304 mentioned in the above-mentioned electronic equipment may be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The communication bus 304 can be divided into an address bus, a data bus, a control bus, etc. For ease of presentation, only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus.

The communication interface 302 is used for communication between the above-mentioned electronic device and other devices.

The processor 301 may be a central processing unit (CPU), or other general-purpose processor, a digital signal processor (Digital Signal Processor, DSP), an application specific integrated circuit (Application Specific Integrated Circuit, ASIC), Field-Programmable Gate Array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general processor may be a microprocessor or the processor may be any conventional processor, etc. The processor 301 is the control center of the electronic device and uses various interfaces and lines to connect various parts of the entire electronic device.

The memory 303 can be used to store the computer program. The processor 301 implements various functions of the electronic device by running or executing the computer program stored in the memory 303 and calling the data stored in the memory 303. Function.

The memory 303 may include non-volatile and/or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous chain Synchlink DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

According to another aspect of the present disclosure, the present invention also proposes a storage medium on which a program is stored. When the program is executed, the following steps are implemented:

Get the coin image provided by the user;

Identify the coin area pictures and obtain multiple feature areas;

The computer-readable storage medium in the embodiment of the present invention may be any combination of one or more computer-readable media. The computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared or semiconductor system, device or device, or any combination thereof. More specific examples (non-exhaustive list) of computer readable storage media include: an electrical connection having one or more conductors, a portable computer hard drive, a hard drive, random access memory (RAM), read only memory (ROM), Erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination of the above. As used herein, a computer-readable storage medium may be any tangible medium that contains or stores a program for use by or in combination with an instruction execution system, apparatus, or device.

A computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave carrying computer-readable program code therein. Such propagated data signals may take many forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium that can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device .

Computer program code for performing the operations of the present invention may be written in one or more programming languages, or a combination thereof, including object-oriented programming languages such as Java, Smalltalk, C++, and conventional Procedural programming language - such as "C" or similar programming language. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In situations involving remote computers, the remote computer can be connected to the user's computer through any kind of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computer (such as an Internet service provider) through the Internet. ).

It should be noted that the devices and methods disclosed in the embodiments of this article can also be implemented in other ways. The device embodiments described above are only illustrative. For example, the flowcharts and block diagrams in the accompanying drawings show the possible implementation architecture, functions and operations of the devices, methods and computer program products according to various embodiments of this document. . In this regard, each block in the flowchart or block diagrams may represent a module, program, or portion of code that contains one or more operable functions for implementing the specified logical functions. Execution instructions, the module, program segment or part of the code contains one or more executable instructions for implementing the specified logical function. It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two consecutive blocks may actually execute substantially in parallel, or they may sometimes execute in the reverse order, depending on the functionality involved. It will also be noted that each block in the block diagram and/or flowchart illustration, and combinations of blocks in the block diagram and/or flowchart illustration, can be designed into specialized hardware-based systems that perform the specified functions or acts. Implemented, or may be implemented using a combination of dedicated hardware and computer instructions.

In addition, each functional module in each embodiment of this article can be integrated together to form an independent part, each module can exist alone, or two or more modules can be integrated to form an independent part.

The above description is only a description of the preferred embodiments of the present invention and does not limit the scope of the present invention in any way. Any changes or modifications made by those of ordinary skill in the field of the present invention based on the above disclosure shall fall within the scope of the claims.

Claims

A coin identification method, characterized by including:

Get the coin image provided by the user;

Identify the coin image to obtain the coin area, and segment the coin area image to obtain the coin area image;

Identify the coin area pictures and obtain multiple feature areas;

Respectively identify the multiple feature areas to obtain multiple coin feature information;

Combining the characteristic information of the multiple coins, the classification information of the coins is obtained.
The coin identification method according to claim 1, wherein identifying the coin image to obtain the coin area includes: applying UNet image segmentation processing to the coin image to obtain the coin area.
The coin identification method according to claim 1, wherein identifying the coin image to obtain the coin area and segmenting the coin area image includes:

The coin area is identified according to the area detection model, the coin mask area is obtained, and the coin area image is segmented and obtained.
The coin identification method according to claim 1, characterized in that the method further includes: performing ellipse detection and perfect circle correction processing on the coin area image.
The coin identification method according to claim 1, characterized in that the method further includes: performing image preprocessing on the coin area pictures obtained by the segmentation to obtain the coin area pictures with background and noise removed;

Identifying the coin area picture and obtaining multiple characteristic areas includes: identifying the coin area picture that has undergone the preprocessing and obtaining multiple characteristic areas.
The coin identification method according to claim 1, characterized in that the method further includes: performing image preprocessing on the coin pictures provided by the user to obtain the coin pictures with background and noise removed;

Recognizing the coin image to obtain the coin area, and segmenting the coin area image to obtain the coin area image includes: identifying the pre-processed coin image to obtain the coin area, and segmenting the coin area image to obtain the coin area image.
The coin identification method according to claim 1, characterized in that the segmentation acquisition hard After collecting the coin area image, the method also includes: performing post-processing on the coin area image with a CV algorithm to remove excess noise.
The coin identification method according to claim 1, wherein the obtaining the coin picture provided by the user includes: providing at least two windows for the user to select on the interactive interface, so that the user can respectively choose to provide different images of the current coins to be identified. Pictures above.
The coin recognition method according to claim 8, characterized in that it is judged whether the coin picture provided by the user belongs to the front or the back according to the pre-trained front and back recognition model.
The coin identification method according to claim 1, characterized in that identifying the coin area picture and obtaining a plurality of characteristic areas includes: identifying the coin area picture according to a pre-trained characteristic area recognition model, and obtaining multiple characteristic areas. characteristic area.
The coin identification method according to claim 1, characterized in that a self-supervised pre-training backbone model based on contrastive learning is used to identify the plurality of characteristic areas respectively to obtain the plurality of coin characteristic information.
The coin identification method according to claim 1, wherein said integrating multiple coin characteristic information and obtaining the classification information of the coin includes: using a similarity comparison method to obtain the comprehensive similarity of the coin, and according to the required The comprehensive similarity ranks the classification information of the coins.
A coin identification system, characterized in that it includes a processor and a memory, and a program is stored on the memory. When the program is executed by the processor, the coin according to any one of claims 1 to 12 is realized. recognition methods.
A storage medium on which a program is stored, characterized in that when the program is executed, the coin identification method according to any one of claims 1 to 12 is implemented.