CN110443306B - Authenticity identification method for wine cork - Google Patents

Authenticity identification method for wine cork Download PDF

Info

Publication number
CN110443306B
CN110443306B CN201910723327.4A CN201910723327A CN110443306B CN 110443306 B CN110443306 B CN 110443306B CN 201910723327 A CN201910723327 A CN 201910723327A CN 110443306 B CN110443306 B CN 110443306B
Authority
CN
China
Prior art keywords
image
wine
cork
wine cork
images
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201910723327.4A
Other languages
Chinese (zh)
Other versions
CN110443306A (en
Inventor
陈昌盛
吴梓贤
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shenzhen University
Original Assignee
Shenzhen University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Shenzhen University filed Critical Shenzhen University
Priority to CN201910723327.4A priority Critical patent/CN110443306B/en
Publication of CN110443306A publication Critical patent/CN110443306A/en
Application granted granted Critical
Publication of CN110443306B publication Critical patent/CN110443306B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/50—Information retrieval; Database structures therefor; File system structures therefor of still image data
    • G06F16/51—Indexing; Data structures therefor; Storage structures
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/40—Extraction of image or video features
    • G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/018—Certifying business or products
    • G06Q30/0185—Product, service or business identity fraud

Abstract

The present disclosure describes a method of classifying wine corks, comprising: collecting a plurality of wine cork images and labels of cork types associated with the wine cork images; carrying out image preprocessing on the wine cork image, obtaining a plurality of local texture images from the wine cork, and establishing a picture database comprising the local texture images and labels of cork types related to the local texture images; constructing a feature extraction module with a deep learning network structure and an artificial neural network with an output module of a multi-classification network, training the artificial neural network based on a picture database, and obtaining a feature extraction model and a classification model of the wine cork; and acquiring a wine cork image to be classified, performing image preprocessing to obtain a target image, classifying the target image based on an artificial neural network, and acquiring a classification result of the target image. By the method, the wine cork to be classified can be conveniently and accurately classified, and the authenticity of the wine cork can be judged.

Description

Authenticity identification method for wine cork
Technical Field
The disclosure particularly relates to a method for identifying wine corks.
Background
At present, the brands of wine in the core production areas of the world are pursued in the market, the price is higher, counterfeit wine is caused to appear in the market, and along with the improvement of counterfeit technology, the difficulty in identifying the authenticity of the wine is increased. The existing wine identification technology comprises the methods of identifying wine based on the package appearance, identifying by combining an anti-counterfeiting label with a wine label, identifying by using a radio frequency identification technology, identifying by using an anti-counterfeiting two-dimensional code and the like.
However, in the existing wine identification technology, the quality of wine is easily affected by the bottle opening inspection; in addition, the process of identifying by using the anti-counterfeiting code on the wine label needs to pass through a plurality of steps and needs to use some special scanning equipment, so that the identification time is too long, the anti-counterfeiting cost is higher or the anti-counterfeiting method is complicated to query; moreover, for example, although the radio frequency identification technology is high in safety, a set of compatibility standards is lacked, the anti-counterfeiting two-dimensional code is not essentially different from the traditional code anti-counterfeiting technology in the technical aspect, and the two-dimensional code has reproducibility but no anti-counterfeiting property, so that the two-dimensional code is easy to crack and is used for counterfeiting. Therefore, the wine identification technology with convenient query and high accuracy is urgently needed at present.
Disclosure of Invention
The present disclosure has been made in view of the above circumstances, and the present inventors have actively searched for a general, robust and efficient identification means, and have found that a wine cork is destroyed and cannot be restored once opened, and that it is influenced by various factors such as wood materials used for the cork, a production process technique of a producer, and a symbol mark unique to wines. The perfect imitation of a real wine cork is at a huge cost, and the human eye cannot easily find the slight texture difference between the wine corks. Based on the above, the invention provides a new identification method, which is used for identifying the authenticity of the corresponding wine by classifying the wine cork based on the artificial neural network and the complex texture characteristics of the wine cork.
To this end, a first aspect of the present disclosure provides a method for classifying wine corks, characterized in that it comprises: collecting a plurality of wine cork images and annotations of cork types associated with the wine cork images; performing image preprocessing on the wine cork image, obtaining a plurality of local texture images representing wine cork textures from the wine cork image, and establishing a picture database comprising the local texture images and labels of cork types associated with the local texture images; constructing an artificial neural network with a feature extraction module of a deep learning network structure and an output module of a multi-classification network, training the artificial neural network based on the image database, and obtaining a feature extraction model and a classification model of the wine cork; and acquiring a wine cork image to be classified, performing image preprocessing to serve as a target image, classifying the target image based on the artificial neural network, and acquiring a classification result of the target image.
In the method, the characteristic extraction model and the classification model of the wine cork are obtained based on the local texture image of the wine cork texture by constructing the artificial neural network with the characteristic extraction module of the deep learning network structure and the output module of the multi-classification network, so that the wine cork to be classified can be conveniently and accurately classified, and the authenticity of the wine cork is judged.
In the classification method according to the first aspect of the present disclosure, optionally, the image preprocessing on the wine cork image includes removing background information and identifying contour information on the wine cork image, and performing image cropping according to the contour information to form a plurality of local texture images, the local texture images including an image reflecting the wine cork texture. In this case, it is possible to obtain a wine cork having maximized texture characteristics and remove invalid information, thereby enhancing a local texture image of the wine cork.
In the classification method according to the first aspect of the present disclosure, optionally, the method further includes acquiring a local texture image of a wine cork containing a known cork type as a reference image; obtaining a reference feature vector of the reference image based on the feature extraction model, and obtaining a target feature vector of the target image based on the feature extraction model; calculating the characteristic distance between the reference characteristic vector and the target characteristic vector; and judging whether the wine cork corresponding to the target image and the wine cork corresponding to the reference image belong to the same type or not based on the characteristic distance. Thereby, it can be easily verified whether the wine corks corresponding to the target image to be classified and the reference image of the known class belong to the same type.
In the classification method related to the first aspect of the present disclosure, optionally, the characteristic distance is an euclidean distance, and when the euclidean distance is greater than a preset threshold, it is determined that a wine cork corresponding to the target image and a wine cork corresponding to the reference image are of different types; and when the Euclidean distance is smaller than or equal to the preset threshold value, judging that the wine cork corresponding to the target image and the wine cork corresponding to the reference image belong to the same type. Thereby, it can be easily verified whether the wine corks corresponding to the target image to be classified and the reference image of the known class belong to the same type.
In the classification method according to the first aspect of the present disclosure, optionally, in the image preprocessing on the wine cork image, the wine cork image is also subjected to a graying process and a binarization process to form a binarized image. In this case, the contour of the wine cork can be obtained by binarizing the image, and thereby, the contour information of the wine cork can be provided when the partial texture image is clipped.
In the classification method according to the first aspect of the present disclosure, optionally, in the image preprocessing on the wine cork image, after the background information is removed, the pattern of the wine cork occupies 90% or more of the target image size. In this case, it can be ensured that the pattern of the wine cork has sufficient textural feature information, whereby a better training effect can be obtained.
In the classification method according to the first aspect of the present disclosure, optionally, the method further includes acquiring a local texture image of a wine cork containing a known cork type as a reference image; performing the image preprocessing on the reference image; obtaining a reference feature vector of the reference image based on the feature extraction model, and obtaining a target feature vector of the target image based on the feature extraction model; and fusing the reference characteristic vector and the target characteristic vector to form a fused characteristic vector, inputting the fused characteristic vector into an auxiliary artificial neural network with a two-classification function, and judging whether the reference image and the target image belong to the same type. Thereby, it can be easily verified whether the wine corks corresponding to the target image to be classified and the reference image of the known class belong to the same type.
In the classification method according to the first aspect of the present disclosure, optionally, the label of the cork type includes at least one of a manufacturer, a place of production, wine, and a year of production. Thus, more comprehensive wine information can be obtained through the wine cork.
A second aspect of the present disclosure provides an electronic apparatus, comprising: one or more processors; storage means for storing one or more programs; the one or more programs, when executed by the one or more processors, cause the one or more processors to implement the classification method as described in any of the first aspects above.
In the electronic device according to the second aspect of the present disclosure, the classification method according to any one of the first aspects is employed, and therefore, the advantageous technical effects of the classification method according to any one of the first aspects can be obtained.
A third aspect of the present disclosure provides a computer-readable medium in which a computer program is stored, which computer program, when executed by a processor, implements the method as described in any of the implementations of the first aspect.
The computer-readable medium according to the third aspect of the present disclosure stores a computer program for executing the classification method according to any one of the first aspects, so that the advantageous effects of the classification method according to any one of the first aspects can be obtained.
According to the method, the wine cork texture features are extracted, and the wine cork texture features are used for wine counterfeit identification. The method for identifying the fake of the wine has the technical effects of convenience in query and high accuracy.
Drawings
The disclosure will now be explained in further detail by way of example only with reference to the accompanying drawings, in which:
fig. 1 is a flowchart illustrating a method of classifying wine corks according to an embodiment of the present disclosure.
Fig. 2 is a schematic diagram illustrating a method of classifying a wine cork according to an embodiment of the present disclosure.
Fig. 3 is an image processing process showing a wine cork image according to an embodiment of the present disclosure.
Fig. 4 is a schematic diagram illustrating a classification framework of a wine cork classification method according to an example of the present disclosure.
Fig. 5 is a schematic diagram showing an example of the texture of a wine cork in a wine cork sorting method according to an example of the present disclosure.
Fig. 6 is a schematic diagram illustrating one example of a method of classifying a wine cork according to an example of the present disclosure.
Fig. 7 is a schematic diagram illustrating another example of a wine cork sorting method according to an example of the present disclosure.
Detailed Description
Hereinafter, preferred embodiments of the present disclosure will be described in detail with reference to the accompanying drawings. In the following description, the same components are denoted by the same reference numerals, and redundant description thereof is omitted. The drawings are schematic and the ratio of the dimensions of the components and the shapes of the components may be different from the actual ones.
It is noted that the terms "comprises," "comprising," and "having," and any variations thereof, in this disclosure, for example, a process, method, system, article, or apparatus that comprises or has a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include or have other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
In addition, the headings and the like referred to in the following description of the present disclosure are not intended to limit the content or scope of the present disclosure, but merely serve as a reminder for reading. Such a subtitle should neither be understood as a content for segmenting an article, nor should the content under the subtitle be limited to only the scope of the subtitle.
Hereinafter, embodiments of the present disclosure will be described in detail with reference to the drawings and examples.
Fig. 1 is a flow chart illustrating a method of classifying wine corks according to an example of the present disclosure. Fig. 2 is a schematic diagram illustrating a classification principle of a wine cork classification method according to an example of the present disclosure.
As shown in fig. 1, the method for classifying a wine cork according to the present embodiment includes: collecting a plurality of wine cork images and labels of cork types associated with the wine cork images (step S10); performing image preprocessing on the wine cork image, obtaining a plurality of local texture images representing wine cork textures from the wine cork image, and establishing a picture database comprising the local texture images and labels of cork types associated with the local texture images (step S20); constructing an artificial neural network with a feature extraction module of a deep learning network structure and an output module of a multi-classification network, training the artificial neural network based on a picture database, and obtaining a feature extraction model and a classification model of the wine cork (step S30); and acquires a cork image of wine to be classified and performs image preprocessing as a target image, and classifies the target image based on an artificial neural network to obtain a classification result of the target image (step S40).
In the method, the characteristic extraction model and the classification model of the wine cork are obtained based on the local texture image of the wine cork texture by constructing the artificial neural network with the characteristic extraction module of the deep learning network structure and the output module of the multi-classification network, so that the wine cork to be classified can be conveniently and accurately classified, and the authenticity of the wine cork is judged.
In step S10, a plurality of wine cork images and annotations of cork type associated with the wine cork images may be collected. The wine cork image may be from a plurality of classes of wine corks, such as the wine cork from a well-known place of production (hereinafter sometimes referred to as "cork") that is currently relatively popular on the market.
Additionally, in some examples, images of the wine stopper may be captured by a camera, wherein the camera may include a cell phone terminal with camera functionality. In some examples, the handset terminal may be a commercially available smart phone such as an apple (iPhone) handset or a hua cell phone. In some examples, the acquisition is maintained under the same lighting, environment, location, etc. conditions during the image acquisition. Each cork may take a plurality of color images, such as RGB images, with a camera.
In addition, the size of the collected wine cork image is not particularly limited, and may be, for example, JPEG images having pixel sizes of 1024 × 768, 1600 × 1200, 2048 × 1536, and 4000 × 3000.
In some examples, a single tone background may be used as the capture background during image acquisition for subsequent image background separation and processing. Specifically, the single-tone background may be white, blue, red, or the like. Therefore, the background information can be removed more accurately in the subsequent image preprocessing process.
In some examples, the label for the type of wine cork may include at least one of a manufacturer, a place of production, a wine, a year of production. Thus, more comprehensive wine information can be obtained through the wine cork.
In some examples, the same number of wine cork images may be collected for each type of wine cork. Therefore, the method is beneficial to the artificial neural network to obtain more balanced results on the wine cork. The present embodiments are not limited in this regard and in other examples, the number of wine cork images may be different for each type of wine cork.
In other examples, a wine cork type with a smaller number of wine cork images may be collected under a number of different collection conditions. This makes it possible to amplify a sample of the wine cork image of the wine cork.
In step S20, the wine cork image may be image preprocessed, a plurality of local texture images representing wine cork textures are obtained from the wine cork image, and a picture database including the local texture images and annotations of cork types associated with the local texture images is built. Here, this picture database is also referred to as a training data set for an artificial neural network.
Fig. 3 is an image processing process showing a wine cork image according to an embodiment of the present disclosure.
In some examples, image pre-processing the wine cork image may include removing background information and identifying contour information from the wine cork image, and performing image cropping based on the contour information to form a plurality of partial texture images, the partial texture images including images reflecting wine cork texture. Thereby, it is possible to maximize the input texture features, remove invalid information, and enhance data.
In some examples, in image preprocessing the wine cork image, the wine cork image may also be subjected to a graying process and a binarization process to form a binarized image. This enables the detection of the contour of the wine cork in the image and provides cork contour information when the partial texture image is cut out.
In some examples, as shown in fig. 3, the acquired wine cork image is a color image (e.g., an RGB image) that includes background information, and the multi-channel image may be converted to a single-channel image by graying the color image, and the binary image may be processed. Then, the contour of the binary image is detected to obtain the contour position information of the wine stopper image, and then a regular rectangular image of the contour is obtained by intercepting according to the contour position information (see fig. 3). In some examples, the pattern of the wine cork may account for more than 90% of the image size after background information removal in image pre-processing the wine cork image. In this case, it can be ensured that the input training data has sufficient texture feature information, and thus, a better training effect can be obtained.
In some examples, the wine cork image may be subjected to a compression operation in image pre-processing the wine cork image. Thereby, the input requirements of the artificial neural network can be matched. Additionally, in some examples, in image pre-processing of the wine cork image, the size of the wine cork image may be compressed, for example, to 299 x 299, to match the input requirements of the deep neural network architecture inclusion-v 4 described later.
In some examples, in step S30, an artificial neural network having a feature extraction module of a deep learning network structure and an output module of a multi-classification network is constructed, and the artificial neural network is trained based on a picture database, to obtain a feature extraction model and a classification model of the wine cork.
Fig. 4 is a schematic diagram illustrating the authentication principle of the classification method of the wine cork according to the example of the present disclosure. Fig. 5 is a schematic diagram showing an example of the texture of a wine cork in a wine cork sorting method according to an example of the present disclosure.
In some examples, in step S30, extracting texture features of the wine cork as the assist features using a Local Binary Pattern (LBP) operator may be included. Thereby, the wine corks can be classified in combination with the assist feature. Here, a Local Binary Pattern (LBP) operator is an efficient texture description operator, can measure and extract texture information of an image local, and has invariance to illumination.
In addition, an example of the texture of the wine cork can be seen in fig. 5, four texture images of a texture image P1, a texture image P2, a texture image P3, and a texture image P4 are listed by way of example, and as is apparent from fig. 5, the texture image P1, the texture image P2, the texture image P3, and the texture image P4 have clearly distinguished textures.
In some examples, the artificial neural network may include a deep convolutional neural network. The deep convolutional neural network simulates a biological visual perception mechanism and has the capability of learning image characteristics. Meanwhile, the calculation amount of the artificial neural network can be reduced through operations such as parameter sharing, pooling and the like. In addition, since the processing is performed on the image, here, the deep Convolutional Neural Network (CNN) is mainly used as the Convolutional Neural Network. Because the convolutional neural network has the advantages of local receptive field, weight sharing and the like, the training of parameters can be greatly reduced, the processing speed can be improved, and the hardware overhead can be saved. In addition, the convolutional neural network can more effectively process the identification of the image. .
In some examples, an artificial neural network may be constructed based on google's inclusion-v 4 neural network structure. The Incep-v 4 neural network node can simultaneously comprise a plurality of convolution and pooling operations of different types, feature maps of different sizes can be subjected to filling processing to obtain feature maps of the same size, and then channels of the feature maps are superposed. The inclusion-v 4 neural network can include a convolution with a 1 x 1 convolution kernel to reduce the number of channels of the feature map, thereby enabling a reduction in the amount of artificial neural network computations.
In some examples, in step S40, a cork image of wine to be classified is acquired and image pre-processed as a target image, and the target image is classified based on an artificial neural network, obtaining a classification result of the target image.
In some examples, the classifier of the artificial neural network described above may include a Softmax classifier. The Softmax classifier is a multi-class linear classifier, and the neural network score vector output by the Softmax classifier is used as the probability distribution of a plurality of normalized classes. Thereby, a classification result of the target image can be obtained.
The results of the classification can be evaluated by a confusion matrix. The two-classification task according to the present embodiment can be divided into four indexes, which are specifically shown in table 1. Where TP represents false positive samples, FN represents false negative samples, TP represents true positive samples, and TN represents true negative samples. Thus, calculation formulas (I) to (IV) of accuracy, specificity, precision rate, and recall rate can be obtained. The accuracy rate represents the correct proportion of the model prediction in all samples, and positive and negative samples are not divided; the accuracy rate represents the proportion of the actual positive sample in all samples of which the model is predicted to be the positive sample; the recall rate represents the proportion of the model which is actually a positive sample and is predicted as the positive sample, and reflects the sensitivity of the model to the positive sample; the specificity rate is opposite to the recall rate, represents the proportion of the model in the actual negative sample predicted as the negative sample, and reflects the sensitivity of the model to the negative sample.
The accuracy is as follows:
Figure RE-BDA0002158014650000091
the precision ratio is as follows:
Figure RE-BDA0002158014650000092
the recall ratio is as follows:
Figure RE-BDA0002158014650000093
the precision ratio is as follows:
Figure RE-BDA0002158014650000094
TABLE 1 confusion matrix
Figure RE-BDA0002158014650000095
In addition, in table 1, type 1 indicates a first type of error in hypothesis testing, which erroneously judges a positive sample as a negative sample, i.e., a "true-out" error; type 2 represents the second type of error in hypothesis testing, erroneously declaring a negative exemplar to be a positive exemplar label, i.e., "false" error.
Hereinafter, a method of classifying wine corks according to an embodiment of the present disclosure will be specifically described with reference to fig. 6 and 7. Fig. 6 and 7 are schematic views each showing an example of a method of classifying a wine cork according to an embodiment of the present disclosure.
Fig. 6 is a schematic diagram illustrating an example of a method of classifying wine corks according to an embodiment of the present disclosure.
In the example shown in fig. 6, the method of classifying a wine cork further comprises obtaining a local texture image of a wine cork containing a known cork type as a reference image; obtaining a reference feature vector of a reference image based on the feature extraction model, and obtaining a target feature vector of a target image based on the feature extraction model; calculating the characteristic distance between the reference characteristic vector and the target characteristic vector; and judging whether the wine cork corresponding to the target image and the wine cork corresponding to the reference image belong to the same type or not based on the characteristic distance. Thereby, it can be verified whether the wine corks corresponding to the target image to be classified and the reference image of the known class belong to the same type.
In some examples, the feature distance may be a euclidean distance L12Euclidean distance L12Is the absolute distance between points in the multidimensional vector space (see equation (V) below). While the Euclidean distance L12Greater than or equal toSetting a threshold T1, judging that the wine cork corresponding to the target image and the wine cork corresponding to the reference image are different types; while the Euclidean distance L12And if the difference is less than or equal to the preset threshold T1, judging that the wine cork corresponding to the target image and the wine cork corresponding to the reference image belong to the same type. Thereby, it can be verified whether the wine corks corresponding to the target image to be classified and the reference image of the known class belong to the same type.
Figure RE-BDA0002158014650000101
Wherein k is a natural number, and k is 1,2,3, … …, n; n represents the dimension of the vector, and k represents the kth dimension of the vector; l is12Representing the distance between the two vectors.
Further, in some examples, the characteristic Distance may also be a Manhattan Distance (Manhattan Distance), a Chebyshev Distance (Chebyshev Distance), a Mahalanobis Distance (Mahalanobis Distance), or a Minkowski Distance (Minkowski Distance).
Fig. 7 is a schematic view showing another example of a wine cork classification method according to an embodiment of the present disclosure.
In the example shown in fig. 7, the method of classifying a wine cork further comprises obtaining a local texture image of a wine cork containing a known cork type as a reference image; carrying out image preprocessing on the reference image; obtaining a reference feature vector of a reference image based on the feature extraction model, and obtaining a target feature vector of a target image based on the feature extraction model; and fusing the reference characteristic vector and the target characteristic vector to form a fused characteristic vector, inputting the fused characteristic vector into an auxiliary artificial neural network with a two-classification function, and judging whether the reference image and the target image belong to the same type. Thereby, it can be verified whether the wine corks corresponding to the target image to be classified and the reference image of the known class belong to the same type.
In addition, in this embodiment, an electronic device according to an embodiment of the present disclosure includes: one or more processors; storage means for storing one or more programs; the one or more programs, when executed by the one or more processors, cause the one or more processors to implement any of the wine cork classification methods as described above. The method described in any of the above implementation modes can be realized by the electronic device related to the embodiment of the present disclosure. Specifically, the characteristic extraction module with the deep learning network structure and the artificial neural network with the output module of the multi-classification network are constructed in the electronic device, and the characteristic extraction model and the classification model of the wine cork are obtained based on the local texture image of the wine cork texture, so that the wine cork to be classified can be conveniently and accurately classified, and the authenticity of the wine cork is judged.
In this embodiment, a third aspect of the disclosure provides a computer readable medium, in which a computer program is stored, which computer program, when being executed by a processor, realizes the method of classifying a wine cork as any one of the above. With the computer-readable medium of the present disclosure, it is also possible to store a computer program that implements the classification method described in any of the implementations described above. Specifically, the artificial neural network with the feature extraction module of the deep learning network structure and the output module of the multi-classification network is constructed in the computer readable medium, and the feature extraction model and the classification model of the wine cork are obtained based on the local texture image of the wine cork texture, so that the wine cork to be classified can be conveniently and accurately classified, and the authenticity of the wine cork is judged.
While the present disclosure has been described in detail in connection with the drawings and examples, it should be understood that the above description is not intended to limit the disclosure in any way. Those skilled in the art can make modifications and variations to the present disclosure as needed without departing from the true spirit and scope of the disclosure, which fall within the scope of the disclosure. It will be understood by those within the art that, in general, terms used in the present disclosure are generally intended as "open" terms (e.g., the term "including" should be interpreted as "including but not limited to," the term "having" should be interpreted as "having at least," the term "includes" should be interpreted as "includes but is not limited to," etc.).
The various methods and techniques described above provide many ways to implement the present disclosure. Of course, it is to be understood that not necessarily all objectives or advantages described may be achieved in accordance with any particular example described in this disclosure. Thus, for example, those skilled in the art will recognize that the methods described in this disclosure may be performed in a manner that achieves or optimizes one advantage or group of advantages as taught by the present disclosure without necessarily achieving other objectives or advantages as may be taught or suggested by the present disclosure.

Claims (7)

1. A method for identifying wine cork is characterized in that,
the method comprises the following steps:
collecting a plurality of wine cork images and labels of cork types related to the wine cork images, wherein the labels of the cork types comprise at least one of manufacturers, production places, wines and production years;
performing image preprocessing on each of the wine cork images, obtaining a plurality of local texture images which represent wine cork textures and have different textures from each of the wine cork images, and establishing a picture database which comprises the plurality of local texture images and labels of cork types associated with the plurality of local texture images, wherein the image preprocessing on the wine cork images comprises removing background information and identifying contour information from the wine cork images, and performing image cropping according to the contour information to form the plurality of local texture images, and the local texture images comprise images reflecting the wine cork textures;
constructing an artificial neural network with a feature extraction module of a deep learning network structure and an output module of a multi-classification network, training the artificial neural network based on the image database, and obtaining a feature extraction model and a classification model of the wine cork; and is
The method comprises the steps of obtaining a wine cork image to be classified, carrying out image preprocessing to serve as a target image, classifying the target image based on the artificial neural network, obtaining a classification result of the target image, carrying out fake identification on the wine cork based on the classification result of the target image, and enabling patterns of the wine cork to account for more than 90% of the size of the target image after background information is removed in the image preprocessing of the wine cork image.
2. A method according to claim 1,
further comprising obtaining a local texture image containing wine corks of a known cork type as a reference image;
obtaining a reference feature vector of the reference image based on the feature extraction model, and obtaining a target feature vector of the target image based on the feature extraction model;
calculating the characteristic distance between the reference characteristic vector and the target characteristic vector;
and judging whether the wine cork corresponding to the target image and the wine cork corresponding to the reference image belong to the same type or not based on the characteristic distance.
3. A method according to claim 2,
the characteristic distance is an Euclidean distance, and when the Euclidean distance is larger than a preset threshold value, the wine wooden plug corresponding to the target image and the wine wooden plug corresponding to the reference image are judged to be of different types; and when the Euclidean distance is smaller than or equal to the preset threshold value, judging that the wine cork corresponding to the target image and the wine cork corresponding to the reference image belong to the same type.
4. A method according to claim 1,
in the image preprocessing of the wine cork image, the wine cork image is also subjected to graying processing and binarization processing to form a binarized image.
5. A method according to claim 1,
further comprising obtaining a local texture image containing wine corks of a known cork type as a reference image;
performing the image preprocessing on the reference image;
obtaining a reference feature vector of the reference image based on the feature extraction model, and obtaining a target feature vector of the target image based on the feature extraction model;
and fusing the reference characteristic vector and the target characteristic vector to form a fused characteristic vector, inputting the fused characteristic vector into an auxiliary artificial neural network with a two-classification function, and judging whether the reference image and the target image belong to the same type.
6. An electronic device, comprising:
one or more processors;
storage means for storing one or more programs;
the one or more programs, when executed by the one or more processors, cause the one or more processors to implement the authentication method of any one of claims 1-5.
7. A computer-readable medium comprising, in combination,
in the computer-readable medium, a computer program is stored which, when being executed by a processor, carries out the authentication method according to any one of claims 1 to 5.
CN201910723327.4A 2019-08-06 2019-08-06 Authenticity identification method for wine cork Active CN110443306B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910723327.4A CN110443306B (en) 2019-08-06 2019-08-06 Authenticity identification method for wine cork

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910723327.4A CN110443306B (en) 2019-08-06 2019-08-06 Authenticity identification method for wine cork

Publications (2)

Publication Number Publication Date
CN110443306A CN110443306A (en) 2019-11-12
CN110443306B true CN110443306B (en) 2020-12-18

Family

ID=68433569

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910723327.4A Active CN110443306B (en) 2019-08-06 2019-08-06 Authenticity identification method for wine cork

Country Status (1)

Country Link
CN (1) CN110443306B (en)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112101309A (en) * 2020-11-12 2020-12-18 北京道达天际科技有限公司 Ground object target identification method and device based on deep learning segmentation network
CN112598008B (en) * 2020-12-25 2021-12-03 上海大学 Thin film pattern database establishing and classification identification method for non-duplicable anti-counterfeit label

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103177280A (en) * 2012-07-27 2013-06-26 张震历 Radiofrequency anti-counterfeiting tag and anti-counterfeiting system
CN106650866A (en) * 2015-12-18 2017-05-10 海南亚元防伪技术研究所（普通合伙） Glue spraying texture anti-counterfeit method and package thereof

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103632388A (en) * 2013-12-19 2014-03-12 百度在线网络技术（北京）有限公司 Semantic annotation method, device and client for image
CN105046280B (en) * 2015-08-10 2018-05-04 北京小豹科技有限公司 A kind of wardrobe intelligent management apapratus and method
CN107977609B (en) * 2017-11-20 2021-07-20 华南理工大学 Finger vein identity authentication method based on CNN
CN108734211B (en) * 2018-05-17 2019-12-24 腾讯科技（深圳）有限公司 Image processing method and device

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103177280A (en) * 2012-07-27 2013-06-26 张震历 Radiofrequency anti-counterfeiting tag and anti-counterfeiting system
CN106650866A (en) * 2015-12-18 2017-05-10 海南亚元防伪技术研究所（普通合伙） Glue spraying texture anti-counterfeit method and package thereof

Also Published As

Publication number Publication date
CN110443306A (en) 2019-11-12

Similar Documents

Publication Publication Date Title
CN109829914B (en) Method and device for detecting product defects
Shahab et al. ICDAR 2011 robust reading competition challenge 2: Reading text in scene images
KR101515256B1 (en) Document verification using dynamic document identification framework
KR101446376B1 (en) Identification and verification of an unknown document according to an eigen image process
CN106610969A (en) Multimodal information-based video content auditing system and method
CN105574550A (en) Vehicle identification method and device
CN108108731B (en) Text detection method and device based on synthetic data
CN106203539B (en) Method and device for identifying container number
TW201732651A (en) Word segmentation method and apparatus
WO2020164278A1 (en) Image processing method and device, electronic equipment and readable storage medium
CN111126240B (en) Three-channel feature fusion face recognition method
CN110598693A (en) Ship plate identification method based on fast-RCNN
CN111583180B (en) Image tampering identification method and device, computer equipment and storage medium
CN107622489A (en) A kind of distorted image detection method and device
CN110443306B (en) Authenticity identification method for wine cork
CN108073940B (en) Method for detecting 3D target example object in unstructured environment
CN109753962B (en) Method for processing text region in natural scene image based on hybrid network
CN115082776A (en) Electric energy meter automatic detection system and method based on image recognition
CN111737478A (en) Text detection method, electronic device and computer readable medium
CN108921006B (en) Method for establishing handwritten signature image authenticity identification model and authenticity identification method
CN111950556A (en) License plate printing quality detection method based on deep learning
CN115731198A (en) Intelligent detection system for leather surface defects
CN115713776A (en) General certificate structured recognition method and system based on deep learning
CN113158745B (en) Multi-feature operator-based messy code document picture identification method and system
CN113111882B (en) Card identification method and device, electronic equipment and storage medium

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant