CN117830314B - Microscopic coded image reproduction detection method and device, mobile terminal and storage medium - Google Patents
Microscopic coded image reproduction detection method and device, mobile terminal and storage medium Download PDFInfo
- Publication number
- CN117830314B CN117830314B CN202410245524.0A CN202410245524A CN117830314B CN 117830314 B CN117830314 B CN 117830314B CN 202410245524 A CN202410245524 A CN 202410245524A CN 117830314 B CN117830314 B CN 117830314B
- Authority
- CN
- China
- Prior art keywords
- image
- target
- deep learning
- microcosmic
- learning model
- 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
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 87
- 238000013136 deep learning model Methods 0.000 claims abstract description 75
- 238000000034 method Methods 0.000 claims abstract description 68
- 238000012545 processing Methods 0.000 claims abstract description 24
- 238000011176 pooling Methods 0.000 claims description 34
- 238000012549 training Methods 0.000 claims description 22
- 238000010586 diagram Methods 0.000 claims description 16
- 238000012360 testing method Methods 0.000 claims description 13
- 238000012795 verification Methods 0.000 claims description 12
- 238000013138 pruning Methods 0.000 claims description 5
- 238000013139 quantization Methods 0.000 claims description 5
- 238000004590 computer program Methods 0.000 claims description 3
- 230000008569 process Effects 0.000 abstract description 11
- 230000006399 behavior Effects 0.000 abstract description 6
- 238000013135 deep learning Methods 0.000 abstract description 6
- 239000002609 medium Substances 0.000 description 20
- 230000006870 function Effects 0.000 description 7
- 230000004913 activation Effects 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 230000003287 optical effect Effects 0.000 description 3
- 230000009471 action Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 230000000644 propagated effect Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 238000013145 classification model Methods 0.000 description 1
- 230000006835 compression Effects 0.000 description 1
- 238000007906 compression Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 238000010295 mobile communication Methods 0.000 description 1
- 239000012120 mounting media Substances 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000011897 real-time detection Methods 0.000 description 1
- 230000008707 rearrangement Effects 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
- 239000013598 vector Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10056—Microscopic image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20084—Artificial neural networks [ANN]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20112—Image segmentation details
- G06T2207/20132—Image cropping
Landscapes
- Engineering & Computer Science (AREA)
- Quality & Reliability (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
The embodiment of the invention discloses a microscopic coding image reproduction detection method, a microscopic coding image reproduction detection device, a mobile terminal and a storage medium, and relates to the technical field of image processing. The method comprises the following steps: acquiring a target microcosmic coded image to be detected; and inputting the target microcosmic coded image into a trained deep learning model so as to output a flip detection result of the target microcosmic coded image through the deep learning model. According to the technical scheme provided by the embodiment of the invention, the micro coding image is subjected to the reproduction detection by the method based on the deep learning, so that the detection accuracy is greatly improved, the embezzlement and the fraudulent conduct of the micro coding image can be timely found and prevented, the legal use rights of the micro coding image are protected, and the safety of the micro coding image information is improved. If the method can prevent the false behavior in the logistics tracking process, the method has wide application prospect and commercial value, and provides safer and more reliable service for users and enterprises.
Description
Technical Field
The embodiment of the invention relates to the technical field of image processing, in particular to a microscopic coding image reproduction detection method, a microscopic coding image reproduction detection device, a mobile terminal and a storage medium.
Background
Microcoded images refer to a technique for generating images using tiny encoding points or regions that encode information into the microstructure of the image based on the image being minutely processed to enable hiding, encrypting, or embedding additional data. The microcosmic coded image is used as a potential information identification mode, can be used for product tracing, anti-counterfeiting authentication, identity verification, logistics tracking, cultural heritage protection and the like, has better information density, concealment and safety compared with a two-dimensional code, and has more important commercial value and safety value in practical application.
The micro-coded image reproduction refers to secondary acquisition of an image, namely an image obtained by imaging the image through more than two digital images. For example, after shooting the microcosmic coded image by shooting equipment such as a mobile phone, a camera and the like, the microcosmic coded image is displayed on a screen such as a mobile phone screen, a computer screen and the like, and then shooting the microcosmic coded image on the screen by another shooting equipment. Such behavior may be used to fool the code scanning system to gain illegal benefits. For example, in a logistics tracking scene, there may be a phenomenon that a photographed microscopic coded image is sent to be forged in the downstream, that is, a flip image is used in the code scanning process instead of a live real photographed image, and the code scanning system cannot accurately distinguish the flip image from the actual photographed image, so that whether goods actually arrive at a certain place cannot be accurately judged.
With the popularity of the mobile internet and smart phones, a microcoded image may be used maliciously as an information recognition method. To solve this problem, anti-roll techniques are generally used to protect the security of the information. Most of the existing anti-rollover technologies are aimed at anti-rollover detection scenes in the processes of credentials, face authentication and the like, the anti-rollover detection recognition rate of microcosmic coded images is very low, and the two situations of rollover and real photographing are easily confused and cannot be accurately recognized. If the traditional image processing method is used, one or more feature vectors such as surface gradient features, HSB tone features, wavelet features, histogram features, LSB features and the like of the image to be detected are extracted, and then training reasoning is carried out by using a target classification model such as SVM and the like to obtain a result of whether the image is a flip image. The method selects specific features conforming to an application scene by manually screening the features to detect whether the image is a reproduction image, so that the method is suitable for single scene and is not suitable for reproduction detection of microcosmic coded images.
Disclosure of Invention
The embodiment of the invention provides a microscopic coding image copying detection method, a microscopic coding image copying detection device, a mobile terminal and a storage medium, which are used for effectively detecting illegal copying behaviors of a microscopic coding image and preventing the microscopic coding image from being maliciously used.
In a first aspect, an embodiment of the present invention provides a method for detecting a flip of a microcoded image, where the method includes:
acquiring a target microcosmic coded image to be detected;
and inputting the target microcosmic coded image into a trained deep learning model so as to output a flip detection result of the target microcosmic coded image through the deep learning model.
Optionally, the deep learning model includes: a backbone network, a neck network, and a classifier; wherein,
The backbone network comprises the first ten layers of mobilenetv networks and is used for processing the target microcoded image to obtain a first characteristic diagram;
The neck network comprises a maximum pooling layer, an average pooling layer and an SE layer, wherein the maximum pooling layer is used for processing the first characteristic map to obtain a second characteristic map, the average pooling layer is used for processing the first characteristic map to obtain a third characteristic map, and the SE layer is used for processing a fourth characteristic map obtained by combining the second characteristic map and the third characteristic map to obtain a fifth characteristic map;
the classifier comprises a full-connection layer and is used for outputting the flap detection result based on the fifth characteristic diagram.
Optionally, before the target microcosmic encoded image is input to the trained deep learning model to output a tap detection result of the target microcosmic encoded image through the deep learning model, the method further includes:
and cutting the target microcosmic coded image based on a preset size according to the code point position in the target microcosmic coded image.
Optionally, the preset size is 3×512×512, the size of the first feature map is 112×32×32, and the size of the fifth feature map is 224×1×1.
Optionally, before the target microcosmic encoded image is input to the trained deep learning model to output a tap detection result of the target microcosmic encoded image through the deep learning model, the method further includes:
collecting a plurality of flip micro-coded images and a plurality of Zhang Shi micro-coded images, and marking to obtain a training sample, a verification sample and a test sample;
And training the deep learning model according to the training sample, the verification sample and the test sample.
Optionally, the method further comprises:
compressing the volume of the deep learning model by pruning and quantization;
And deploying the compressed deep learning model on the mobile terminal.
Optionally, after the target microcosmic encoded image is input to the trained deep learning model to output a tap detection result of the target microcosmic encoded image through the deep learning model, the method further includes:
And if the detection result of the flipping is the flipping, alarming and/or preventing the identification of the target microcosmic coded image.
In a second aspect, an embodiment of the present invention further provides a device for detecting a micro-encoded image reproduction, where the device includes:
The image acquisition module is used for acquiring a target microcosmic coded image to be detected;
And the overturn detection module is used for inputting the target microcosmic coded image into the trained deep learning model so as to output an overturn detection result of the target microcosmic coded image through the deep learning model.
In a third aspect, an embodiment of the present invention further provides a mobile terminal, where the mobile terminal includes:
one or more processors;
a memory 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 method for detecting a micro-coded image flip provided by any embodiment of the present invention.
In a fourth aspect, embodiments of the present invention further provide a computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the method for detecting a micro-encoded image flip provided by any embodiment of the present invention.
The embodiment of the invention provides a method for detecting the flip of a microcosmic coded image, which comprises the steps of firstly acquiring a target microcosmic coded image to be detected, and then inputting the target microcosmic coded image into a trained deep learning model so as to output a flip detection result of the target microcosmic coded image through the deep learning model. According to the method for detecting the micro-coded image reproduction, provided by the embodiment of the invention, the micro-coded image is subjected to reproduction detection by a method based on deep learning, so that the detection accuracy is greatly improved, the micro-coded image can be timely found and prevented from being stolen and fraudulently, the legal use rights of the micro-coded image are protected, and the safety of the micro-coded image information is improved. If the method can prevent the false behavior in the logistics tracking process, the method has wide application prospect and commercial value, and provides safer and more reliable service for users and enterprises.
Drawings
FIG. 1 is a flowchart of a method for detecting a flip of a microcoded image according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of a deep learning model according to an embodiment of the present invention;
fig. 3 is a schematic structural diagram of a micro-encoded image reproduction detection device according to a second embodiment of the present invention;
Fig. 4 is a schematic structural diagram of a mobile terminal according to a third embodiment of the present invention.
Detailed Description
The invention is described in further detail below with reference to the drawings and examples. It is to be understood that the specific embodiments described herein are merely illustrative of the invention and are not limiting thereof. It should be further noted that, for convenience of description, only some, but not all of the structures related to the present invention are shown in the drawings.
Before discussing exemplary embodiments in more detail, it should be mentioned that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although a flowchart depicts steps as a sequential process, many of the steps may be implemented in parallel, concurrently, or with other steps. Furthermore, the order of the steps may be rearranged. The process may be terminated when its operations are completed, but may have additional steps not included in the figures. The processes may correspond to methods, functions, procedures, subroutines, and the like.
Furthermore, the terms "first," "second," and the like, may be used herein to describe various directions, acts, steps, or elements, etc., but these directions, acts, steps, or elements are not limited by these terms. These terms are only used to distinguish one direction, action, step or element from another direction, action, step or element. For example, a first feature map may be referred to as a second feature map, and similarly, a second feature map may be referred to as a first feature map, without departing from the scope of embodiments of the present invention. Both the first feature map and the second feature map are feature maps, but they are not the same feature map. The terms "first," "second," and the like, are not to be construed as indicating or implying a relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defining "a first" or "a second" may explicitly or implicitly include one or more such feature. In the description of the embodiments of the present invention, the meaning of "plurality" is at least two, for example, two, three, etc., unless explicitly defined otherwise.
Example 1
Fig. 1 is a flowchart of a method for detecting a flip of a microcoded image according to an embodiment of the present invention. The embodiment is suitable for the situation that whether the microcoded image is a reproduction or not is detected before the microcoded image is required to be identified in the aspects of product tracing, anti-counterfeiting authentication, identity verification, logistics tracking, cultural heritage protection and the like. As shown in fig. 1, the method specifically comprises the following steps:
s11, acquiring a target microcosmic coded image to be detected.
S12, inputting the target microcosmic coded image into a trained deep learning model so as to output a reproduction detection result of the target microcosmic coded image through the deep learning model.
Specifically, the target microcosmic coded image can be obtained by shooting through a mobile terminal such as a mobile phone, and whether the target microcosmic coded image is a turnup image or not can be detected before the target microcosmic coded image is identified. Specifically, the photographed target microcosmic coded image can be input into a trained deep learning model, so that a flip detection result of the target microcosmic coded image is obtained through prediction of the deep learning model. The test result of the turnup can include turnup and non-turnup, further, when the test result of the turnup is non-turnup, the target microcosmic code image can be used for identification directly, so as to read the information therein for verification.
Optionally, as shown in fig. 2, the deep learning model includes: a backbone network (backbone), a neck network (back), and a classifier (classifier) 310; the backbone network comprises a first ten layers (mobilenetv [:13 ]) 110 of mobilenetv < 3 > network, which is used for processing the target microcoded image to obtain a first characteristic diagram; the neck network comprises a maximum pooling layer (maxpool) 210, an average pooling layer (avgpool) 220 and a SE layer (seLayer) 230, wherein the maximum pooling layer 210 is used for processing the first feature map to obtain a second feature map, the average pooling layer 220 is used for processing the first feature map to obtain a third feature map, and the SE layer 230 is used for processing a fourth feature map obtained by combining the second feature map and the third feature map to obtain a fifth feature map; the classifier 310 includes a full-connection layer, configured to output the beat detection result based on the fifth feature map.
Specifically, a deep learning model can be used as a feature extractor, and the model can be modified to be modularized in order to facilitate the operation of the model on the mobile terminal, so that the number of modules and the network structure in the modules can be conveniently controlled. The backbone network extracts the first feature map of the target microcoded image using the first ten layers 110 of the mobilenetv network (featureMap), which may include thirteen modules (blockNet). The depth separable convolution is used in each module to replace all standard convolution operations, so that the calculation amount and the parameter number of the model are reduced, the network characterization capability is improved, and meanwhile, an SE (sequential-and-specification) module is introduced, so that the channel weight in the channel characteristic diagram can be adaptively adjusted, the model is focused on key characteristics of an image, a hard-swish activation function is used for replacing an original ReLu activation function, the hard-swish activation function has higher calculation efficiency, the nonlinear characteristic is stronger, the activation range is wider, and better gradient propagation is achieved. After the backbone network, a parallel maximum pooling layer 210 and an average pooling layer 220 can be added as neck extraction information, the maximum pooling layer 210 and the average pooling layer 220 respectively process the first feature map to obtain a second feature map and a third feature map, then the second feature map and the third feature map are combined to obtain a fourth feature map, and the number of compressed images and parameters of the two pooling layers can reduce the overfitting. After the two pooling layers, an SE layer 230 may be further added to serve as a channel attention module, and the fourth feature map is processed by the SE layer 230 to obtain a fifth feature map, and the fifth feature map is sent to the classifier 310, so that the attention of the model to the key features can be improved. The classifier 310 is a fully-connected network, and is configured to output a tap detection result based on the fifth feature map, and includes a fully-connected layer (Linear), and may further include a Dropout layer to prevent overfitting, where the tap detection result output by the classifier 310 may be 0/1, where 0 may represent non-tap, and 1 may represent tap. By combining the deep learning technology and the image feature extraction algorithm and carrying out customized model construction aiming at a specific application scene, the accuracy of the model is greatly improved. The method specifically compares different common network structures, selects the network structure with the best effect of detecting the flip of the microcosmic coded image, and in order to improve the accuracy, the whole network structure is not directly used, but the structure is optimized and modified, thirteen blockNet are combined to be used as a backstone of the feature extractor, seLayer is added to be used as an attention module, and compared with the method of directly using a lightweight network, the accuracy of the flip detection is higher, and the robustness is better. In addition, the existing method for detecting the flip is mainly aimed at plane flip scenes, but in practical application, situations such as inclined flip and curved flip can be met, and by applying the deep learning model provided by the embodiment, the detection of the flip of the microcosmic coded image under various conditions can be effectively carried out, the method has stronger generalization capability, and reliable detection and identification can be realized even in the face of diversified flip methods such as transformation, rotation, scaling and the like.
Further optionally, before the target microcosmic encoded image is input to the trained deep learning model to output a tap detection result of the target microcosmic encoded image through the deep learning model, the method further includes: and cutting the target microcosmic coded image based on a preset size according to the code point position in the target microcosmic coded image, so that the influence of excessive background content on subsequent detection is avoided. Optionally, the preset size (i.e. the input size of the backbone network) is 3×512×512, the size of the first feature map (i.e. the output size of the backbone network and the input size of the neck network) is 112×32×32, and the size of the fifth feature map (i.e. the output size of the neck network and the input size of the classifier) is 224×1×1.
On the basis of the above technical solution, optionally, before the inputting the target microcosmic encoded image to the trained deep learning model to output the tap detection result of the target microcosmic encoded image through the deep learning model, the method further includes: collecting a plurality of flip micro-coded images and a plurality of Zhang Shi micro-coded images, and marking to obtain a training sample, a verification sample and a test sample; and training the deep learning model according to the training sample, the verification sample and the test sample. Specifically, a large number of actual scene flipped micro-encoded images and a large number of actual photographed actual photo micro-encoded images may be collected and marked to distinguish which are flipped and which are actual, and then the collected images may be divided into training samples, verification samples and test samples, so that the training samples are used to train the used deep learning model, so that the training samples learn the features of the two types (flipped and non-flipped) of images, such as various features such as moire, screen frames, etc. may exist on the flipped images. The model may then be further training optimized using the validation sample and the test sample. Furthermore, after the trained deep learning model is used for detecting the target microcosmic coding image, training samples can be accumulated and updated continuously by using the target microcosmic coding image, and training is repeated, so that the performance of the model is optimized, the detection accuracy and the detection robustness are further improved, and a closed loop is formed.
On the basis of the above technical solution, optionally, the method further includes: compressing the volume of the deep learning model by pruning and quantization; and deploying the compressed deep learning model on the mobile terminal. Specifically, most of the existing methods for detecting the beats at present implement reasoning at a server, and a client is only used for collecting images, so that local real-time beat detection cannot be implemented. The method for detecting the turner can be realized on the mobile terminal, namely the mobile terminal can realize the turner detection in real time after collecting the target microcoded image, so that the efficiency and the accuracy of the turner detection are improved. The mobile terminal generally has limited computing resources and memory capacity, in order to deploy a model on the mobile terminal and accelerate reasoning, the trained deep learning model can be optimized in a customized mode, and the model volume is compressed to be within 1M by utilizing pruning, quantization and other technologies, so that the model volume is smaller, the reasoning speed is higher, and the model can be improved to be more than 30 FPS. And then the compressed deep learning model can be deployed to the mobile terminal, so that the mobile terminal can directly use the model for real-time detection, and meanwhile, the model can also be supported for use on a plurality of operating systems and devices, so that a user can perform the flap detection on different devices, the flexibility and convenience of use are improved, and the application field of deep learning at the mobile terminal is expanded.
On the basis of the above technical solution, optionally, after the target microcosmic encoded image is input to the trained deep learning model to output a tap detection result of the target microcosmic encoded image through the deep learning model, the method further includes: and if the flip detection result is a flip, alarming and/or preventing the identification of the target microcosmic coded image so as to prompt the risk of a user and prevent the microcosmic coded image from being maliciously used.
According to the technical scheme provided by the embodiment of the invention, the target microcoded image to be detected is firstly obtained, and then the target microcoded image is input into the trained deep learning model, so that the flip detection result of the target microcoded image is output through the deep learning model. The detection accuracy is greatly improved by performing the flap detection on the microcosmic coded image based on the deep learning method, so that the embezzlement and the fraud of the microcosmic coded image can be timely found and prevented, the legal use rights and interests of the microcosmic coded image are protected, and the safety of the microcosmic coded image information is improved. If the method can prevent the false behavior in the logistics tracking process, the method has wide application prospect and commercial value, and provides safer and more reliable service for users and enterprises.
Example two
Fig. 3 is a schematic structural diagram of a micro-encoded image reproduction detection device according to a second embodiment of the present invention, where the device may be implemented in hardware and/or software, and may be generally integrated in a mobile terminal, for executing the micro-encoded image reproduction detection method according to any embodiment of the present invention. As shown in fig. 3, the apparatus includes:
an image acquisition module 31 for acquiring a target microcoded image to be detected;
The flip detection module 32 is configured to input the target microcoded image into a trained deep learning model, so as to output a flip detection result of the target microcoded image through the deep learning model.
According to the technical scheme provided by the embodiment of the invention, the target microcoded image to be detected is firstly obtained, and then the target microcoded image is input into the trained deep learning model, so that the flip detection result of the target microcoded image is output through the deep learning model. The detection accuracy is greatly improved by performing the flap detection on the microcosmic coded image based on the deep learning method, so that the embezzlement and the fraud of the microcosmic coded image can be timely found and prevented, the legal use rights and interests of the microcosmic coded image are protected, and the safety of the microcosmic coded image information is improved. If the method can prevent the false behavior in the logistics tracking process, the method has wide application prospect and commercial value, and provides safer and more reliable service for users and enterprises.
On the basis of the above technical solution, optionally, the deep learning model includes: a backbone network, a neck network, and a classifier; wherein,
The backbone network comprises the first ten layers of mobilenetv networks and is used for processing the target microcoded image to obtain a first characteristic diagram;
The neck network comprises a maximum pooling layer, an average pooling layer and an SE layer, wherein the maximum pooling layer is used for processing the first characteristic map to obtain a second characteristic map, the average pooling layer is used for processing the first characteristic map to obtain a third characteristic map, and the SE layer is used for processing a fourth characteristic map obtained by combining the second characteristic map and the third characteristic map to obtain a fifth characteristic map;
the classifier comprises a full-connection layer and is used for outputting the flap detection result based on the fifth characteristic diagram.
On the basis of the technical scheme, the microscopic coded image reproduction detection device optionally further comprises:
And the image clipping module is used for clipping the target microcosmic coding image based on a preset size according to the code point position in the target microcosmic coding image before the target microcosmic coding image is input into the trained deep learning model so as to output the flip detection result of the target microcosmic coding image through the deep learning model.
Based on the above technical solution, optionally, the preset size is 3×512×512, the size of the first feature map is 112×32×32, and the size of the fifth feature map is 224×1×1.
On the basis of the technical scheme, the microscopic coded image reproduction detection device optionally further comprises:
The sample acquisition module is used for acquiring a plurality of flip micro-coded images and a plurality of Zhang Shi micro-coded images and marking the flip micro-coded images before the target micro-coded images are input into the trained deep learning model to output a flip detection result of the target micro-coded images through the deep learning model, so as to obtain a training sample, a verification sample and a test sample;
And the model training module is used for training the deep learning model according to the training sample, the verification sample and the test sample.
On the basis of the technical scheme, the microscopic coded image reproduction detection device optionally further comprises:
The model compression module is used for compressing the volume of the deep learning model by pruning and quantization;
And the model deployment module is used for deploying the compressed deep learning model on the mobile terminal.
On the basis of the technical scheme, the microscopic coded image reproduction detection device optionally further comprises:
And the flipping processing module is used for alarming and/or preventing the identification of the target microcosmic coded image if the flipping detection result is a flipping after the target microcosmic coded image is input into the trained deep learning model so as to output the flipping detection result of the target microcosmic coded image through the deep learning model.
The micro-coded image reproduction detection device provided by the embodiment of the invention can execute the micro-coded image reproduction detection method provided by any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the execution method.
It should be noted that, in the embodiment of the above-mentioned microscopic encoded image reproduction detection apparatus, each unit and module included are only divided according to the functional logic, but are not limited to the above-mentioned division, as long as the corresponding functions can be realized; in addition, the specific names of the functional units are also only for distinguishing from each other, and are not used to limit the protection scope of the present invention.
Example III
Fig. 4 is a schematic structural diagram of a mobile terminal according to a third embodiment of the present invention, and shows a block diagram of an exemplary mobile terminal suitable for implementing an embodiment of the present invention. The mobile terminal shown in fig. 4 is only an example, and should not be construed as limiting the functionality and scope of use of the embodiments of the present invention. As shown in fig. 4, the mobile terminal includes a processor 41, a memory 42, an input device 43 and an output device 44; the number of processors 41 in the mobile terminal may be one or more, in fig. 4, one processor 41 is taken as an example, and the processors 41, the memory 42, the input device 43 and the output device 44 in the mobile terminal may be connected by a bus or other means, in fig. 4, by a bus connection is taken as an example.
The memory 42 is a computer readable storage medium, and may be used to store software programs, computer executable programs, and modules, such as program instructions/modules corresponding to the method for detecting a micro-encoded image in an embodiment of the present invention (e.g., the image acquisition module 31 and the image detection module 32 in the micro-encoded image reproduction detection apparatus). The processor 41 executes various functional applications of the mobile terminal and data processing, namely, implements the above-described micro-coded image roll-over detection method by running software programs, instructions and modules stored in the memory 42.
The memory 42 may mainly include a storage program area and a storage data area, wherein the storage program area may store an operating system, at least one application program required for functions; the storage data area may store data created according to the use of the mobile terminal, etc. In addition, memory 42 may include high-speed random access memory, and may also include non-volatile memory, such as at least one magnetic disk storage device, flash memory device, or other non-volatile solid-state storage device. In some examples, memory 42 may further comprise memory located remotely from processor 41, which may be connected to the mobile terminal via a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.
The input means 43 may be used for acquiring a target microcoded image to be detected, generating key signal inputs related to user settings and function controls of the mobile terminal, etc. The output device 44 may include a display that may be used to present the results of the test to the user, etc.
Example IV
A fourth embodiment of the present invention also provides a storage medium containing computer-executable instructions, which when executed by a computer processor, are for performing a method of detecting a micro-encoded image flip, the method comprising:
acquiring a target microcosmic coded image to be detected;
and inputting the target microcosmic coded image into a trained deep learning model so as to output a flip detection result of the target microcosmic coded image through the deep learning model.
The storage medium may be any of various types of memory devices or storage devices. The term "storage medium" is intended to include: mounting media such as CD-ROM, floppy disk or tape devices; computer system memory or random access memory such as DRAM, DDR RAM, SRAM, EDO RAM, lanbus (Rambus) RAM, etc.; nonvolatile memory such as flash memory, magnetic media (e.g., hard disk or optical storage); registers or other similar types of memory elements, etc. The storage medium may also include other types of memory or combinations thereof. In addition, the storage medium may be located in a computer system in which the program is executed, or may be located in a different second computer system connected to the computer system through a network (such as the internet). The second computer system may provide program instructions to the computer for execution. The term "storage medium" may include two or more storage media that may reside in different locations (e.g., in different computer systems connected by a network). The storage medium may store program instructions (e.g., embodied as a computer program) executable by one or more processors.
Of course, the storage medium containing the computer executable instructions provided in the embodiments of the present invention is not limited to the above-described method operations, and may also perform the related operations in the micro-encoded image flip detection method provided in any embodiment of the present invention.
The computer readable signal medium may include a propagated data signal with computer readable program code embodied therein, either in baseband or as part of a carrier wave. Such a propagated data signal may take any of a variety of forms, including, but not limited to, electro-magnetic, optical, or any suitable combination of the foregoing. A computer readable signal medium may also be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device.
Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing.
From the above description of embodiments, it will be clear to a person skilled in the art that the present invention may be implemented by means of software and necessary general purpose hardware, but of course also by means of hardware, although in many cases the former is a preferred embodiment. Based on such understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art in the form of a software product, which may be stored in a computer readable storage medium, such as a floppy disk, a Read-Only Memory (ROM), a random access Memory (Random Access Memory, RAM), a FLASH Memory (FLASH), a hard disk, or an optical disk of a computer, etc., and include several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method according to the embodiments of the present invention.
Note that the above is only a preferred embodiment of the present invention and the technical principle applied. It will be understood by those skilled in the art that the present invention is not limited to the particular embodiments described herein, but is capable of various obvious changes, rearrangements and substitutions as will now become apparent to those skilled in the art without departing from the scope of the invention. Therefore, while the invention has been described in connection with the above embodiments, the invention is not limited to the embodiments, but may be embodied in many other equivalent forms without departing from the spirit or scope of the invention, which is set forth in the following claims.
Claims (8)
1. The method for detecting the micropoding image reproduction is characterized by comprising the following steps of:
acquiring a target microcosmic coded image to be detected;
Inputting the target microcosmic coded image into a trained deep learning model so as to output a reproduction detection result of the target microcosmic coded image through the deep learning model;
The deep learning model includes: a backbone network, a neck network, and a classifier; wherein,
The backbone network comprises the first ten layers of mobilenetv networks and is used for processing the target microcoded image to obtain a first characteristic diagram;
The neck network comprises a maximum pooling layer, an average pooling layer and an SE layer, wherein the maximum pooling layer and the average pooling layer are connected in parallel and then are used for extracting information from the neck after being connected to the backbone network, and the SE layer is arranged after the maximum pooling layer and the average pooling layer and is used as a channel attention module; the maximum pooling layer is used for processing the first feature map to obtain a second feature map, the average pooling layer is used for processing the first feature map to obtain a third feature map, and the SE layer is used for processing a fourth feature map obtained by combining the second feature map and the third feature map to obtain a fifth feature map;
The classifier comprises a full-connection layer and is used for outputting the flap detection result based on the fifth characteristic diagram;
Before the target micro-coded image is input into the trained deep learning model to output the flip detection result of the target micro-coded image through the deep learning model, the method further comprises:
and cutting the target microcosmic coded image based on a preset size according to the code point position in the target microcosmic coded image.
2. The method of claim 1, wherein the predetermined size is 3×512×512, the size of the first feature map is 112×32×32, and the size of the fifth feature map is 224×1×1.
3. The method according to claim 1, further comprising, before the inputting the target microcoded image to a trained deep learning model to output a reproduction detection result of the target microcoded image by the deep learning model:
collecting a plurality of flip micro-coded images and a plurality of Zhang Shi micro-coded images, and marking to obtain a training sample, a verification sample and a test sample;
And training the deep learning model according to the training sample, the verification sample and the test sample.
4. The method of detecting a flip of a microcoded image according to claim 1, further comprising:
compressing the volume of the deep learning model by pruning and quantization;
And deploying the compressed deep learning model on the mobile terminal.
5. The method according to claim 1, further comprising, after the inputting the target microcoded image to a trained deep learning model to output a result of the detection of the target microcoded image by the deep learning model:
And if the detection result of the flipping is the flipping, alarming and/or preventing the identification of the target microcosmic coded image.
6. A microscopic encoded image roll-over detection device, comprising:
The image acquisition module is used for acquiring a target microcosmic coded image to be detected;
the device comprises a target microcosmic coding image detection module, a training deep learning model and a reproduction detection module, wherein the target microcosmic coding image detection module is used for inputting the target microcosmic coding image into the training deep learning model so as to output a reproduction detection result of the target microcosmic coding image through the deep learning model;
The deep learning model includes: a backbone network, a neck network, and a classifier; wherein,
The backbone network comprises the first ten layers of mobilenetv networks and is used for processing the target microcoded image to obtain a first characteristic diagram;
The neck network comprises a maximum pooling layer, an average pooling layer and an SE layer, wherein the maximum pooling layer and the average pooling layer are connected in parallel and then are used for extracting information from the neck after being connected to the backbone network, and the SE layer is arranged after the maximum pooling layer and the average pooling layer and is used as a channel attention module; the maximum pooling layer is used for processing the first feature map to obtain a second feature map, the average pooling layer is used for processing the first feature map to obtain a third feature map, and the SE layer is used for processing a fourth feature map obtained by combining the second feature map and the third feature map to obtain a fifth feature map;
The classifier comprises a full-connection layer and is used for outputting the flap detection result based on the fifth characteristic diagram;
The apparatus further comprises:
And the image clipping module is used for clipping the target microcosmic coding image based on a preset size according to the code point position in the target microcosmic coding image before the target microcosmic coding image is input into the trained deep learning model so as to output the flip detection result of the target microcosmic coding image through the deep learning model.
7. A mobile terminal, comprising:
one or more processors;
a memory 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 micro-encoded image tap detection method of any of claims 1-5.
8. A computer readable storage medium having stored thereon a computer program, wherein the program when executed by a processor implements a micro-coded image roll-over detection method according to any of claims 1-5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410245524.0A CN117830314B (en) | 2024-03-05 | 2024-03-05 | Microscopic coded image reproduction detection method and device, mobile terminal and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410245524.0A CN117830314B (en) | 2024-03-05 | 2024-03-05 | Microscopic coded image reproduction detection method and device, mobile terminal and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117830314A CN117830314A (en) | 2024-04-05 |
| CN117830314B true CN117830314B (en) | 2024-05-03 |
Family
ID=90513820
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410245524.0A Active CN117830314B (en) | 2024-03-05 | 2024-03-05 | Microscopic coded image reproduction detection method and device, mobile terminal and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117830314B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115147705A (en) * | 2022-09-06 | 2022-10-04 | 平安银行股份有限公司 | Face copying detection method and device, electronic equipment and storage medium |
| CN116109878A (en) * | 2023-04-10 | 2023-05-12 | 南京掌控网络科技有限公司 | Image reproduction identification method, system, device and storage medium |
| CN117611907A (en) * | 2023-11-30 | 2024-02-27 | 中国农业银行股份有限公司 | Method and device for detecting flip image and training model, electronic equipment and medium |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US7512249B2 (en) * | 2004-04-26 | 2009-03-31 | Graphic Security Systems Corporation | System and method for decoding digital encoded images |
| WO2022221719A2 (en) * | 2021-04-16 | 2022-10-20 | Strong Force Vcn Portfolio 2019, Llc | Systems, methods, kits, and apparatuses for digital product network systems and biology-based value chain networks |
-
2024
- 2024-03-05 CN CN202410245524.0A patent/CN117830314B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115147705A (en) * | 2022-09-06 | 2022-10-04 | 平安银行股份有限公司 | Face copying detection method and device, electronic equipment and storage medium |
| CN116109878A (en) * | 2023-04-10 | 2023-05-12 | 南京掌控网络科技有限公司 | Image reproduction identification method, system, device and storage medium |
| CN117611907A (en) * | 2023-11-30 | 2024-02-27 | 中国农业银行股份有限公司 | Method and device for detecting flip image and training model, electronic equipment and medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117830314A (en) | 2024-04-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Marra et al. | A full-image full-resolution end-to-end-trainable CNN framework for image forgery detection | |
| Rocha et al. | Vision of the unseen: Current trends and challenges in digital image and video forensics | |
| Redi et al. | Digital image forensics: a booklet for beginners | |
| CN112052830B (en) | Method, device and computer storage medium for face detection | |
| Cozzolino et al. | Combining PRNU and noiseprint for robust and efficient device source identification | |
| Mareen et al. | Comprint: Image forgery detection and localization using compression fingerprints | |
| CN114758113A (en) | Confrontation sample defense training method, classification prediction method and device, and electronic equipment | |
| Gani et al. | A novel method for digital image copy-move forgery detection and localization using evolving cellular automata and local binary patterns | |
| Ekhande et al. | Review on effectiveness of deep learning approach in digital forensics. | |
| CN117830314B (en) | Microscopic coded image reproduction detection method and device, mobile terminal and storage medium | |
| Nataraj et al. | Seam carving detection and localization using two-stage deep neural networks | |
| CN113269123B (en) | Certificate identification method and system | |
| Koshy et al. | Video forgery detection using CNN | |
| CN115775409A (en) | Face image tamper-proof fusion detection method | |
| Kaur et al. | A comparative review of various techniques for image splicing detection and localization | |
| Alharbi et al. | Spoofing Face Detection Using Novel Edge-Net Autoencoder for Security. | |
| Patil et al. | Securing Visual Integrity: Machine learning approaches for forged image detection | |
| Alzamil | Image Forgery Detection with Machine Learning | |
| CN114694196A (en) | Living body classifier establishing method, human face living body detection method and device | |
| Saleem et al. | Credibility assessment of social media images shared during disasters | |
| Singh et al. | Fake Image Detection Using Ensemble Learning | |
| Meenakshi Sundaram et al. | ASRD: algorithm for spliced region detection in digital image forensics | |
| Selvarathi et al. | Copy Move Counterfeiting Identification based on CNN using Scalar Invariant Feature Technique | |
| CN116778534B (en) | Image processing method, device, equipment and medium | |
| CN114092994B (en) | Face living body detection method based on multi-view feature learning |
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 |