WO2019061661A1

WO2019061661A1 - Image tamper detecting method, electronic device and readable storage medium

Info

Publication number: WO2019061661A1
Application number: PCT/CN2017/108765
Authority: WO
Inventors: 王健宗; 王晨羽; 马进; 肖京
Original assignee: 平安科技（深圳）有限公司
Priority date: 2017-09-30
Filing date: 2017-10-31
Publication date: 2019-04-04
Also published as: CN107657259A

Abstract

An image tamper detecting method, an electronic device (1) and a readable storage medium, said method comprising: segmenting an image, which is to be detected, into blocks: segmenting the image, which is to be detected, into several grid image blocks having a preset specification, and extracting, from the several grid image blocks, a preset number of tamper detection features (S10); identifying, on the basis of the preset number of tamper detection features and a pre-determined identification model, the image which is to be detected, and analyzing, according to an identification result, whether the image, which is to be detected, is tampered (S20), the pre-determined identification model being a deep neural network model which is obtained by previously training sample pictures labeled with different tamper trace features. Said method enables that the type of tampered image which can be detected is not only limited to a JPEG format, realizing detection of different types of image tamper.

Description

Image tamper detecting method, electronic device and readable storage medium

The present application is based on the priority of the Chinese Patent Application entitled "Image Tamper Detection Method, Electronic Device and Readable Storage Medium", filed on September 30, 2017, with the application number of CN 201710916508.X, which is filed on September 30, 2017. The overall content is incorporated herein by reference.

Technical field

The present application relates to the field of computer technologies, and in particular, to an image tamper detecting method, an electronic device, and a readable storage medium.

Background technique

With the widespread popularity of digital image editing tools such as Adobe Photoshop and ACDSee, more and more users are free to modify digital photos at will, and some malicious users with illegal purposes can take advantage of them. In the case of illegal operation of the image content, such as synthetic false, illegal editing, etc., resulting in harm. The existing conventional algorithms can only detect tampering of images whose image type is JPEG format, so that the image format capable of image tampering detection is relatively simple.

Summary of the invention

The purpose of the present application is to provide an image tamper detecting method, an electronic device, and a readable storage medium, which are intended to enable detection of tampering image types and not limited to JPEG, to detect tampering of different types of images.

To achieve the above object, a first aspect of the present application provides an electronic device including a memory, a processor, and an image tamper detecting system operable on the processor, the image tampering The detection system implements the following steps when executed by the processor:

A. performing block segmentation on the image to be detected, dividing the image to be detected into a plurality of grid image blocks of preset specifications, and extracting a preset number of tamper detecting features from the plurality of grid image blocks;

B. Identifying the image to be detected based on a preset number of tamper detection features and a predetermined recognition model, and analyzing whether the image to be detected is tampered according to the recognition result; wherein the predetermined recognition model is A deep neural network model obtained by training sample pictures marked with different tamper trace features.

The second aspect of the present application provides an image tampering detection method, where the image tampering detection method includes:

Step 1: performing block segmentation on the detected image, and dividing the image to be detected into several Presetting a grid image block of a specification, and extracting a preset number of tamper detecting features from a plurality of grid image blocks;

Step 2: Identify the image to be detected based on a preset number of tamper detection features and a predetermined recognition model, and analyze whether the image to be detected is tampered according to the recognition result; wherein the predetermined recognition model is A deep neural network model obtained by training a sample picture marked with different tamper trace features in advance.

A third aspect of the present application provides a computer readable storage medium storing an image tamper detecting system, the image tamper detecting system being executable by at least one processor to cause the at least one processor The steps of the image tampering detection method as described above are performed.

The image tamper detecting method, system and readable storage medium proposed by the present application identify a detected image by using a deep neural network model obtained by training a plurality of sample images marked with different tamper trace features, and analyzing the detected image according to the recognition result Whether the image to be detected has been tampered with. The tamper detection feature is extracted from each grid image block by dividing the image to be detected into a plurality of grid image blocks, and the extracted tamper detection feature is identified by using a pre-trained deep neural network model to It is recognized that the image to be detected has tampering marks, and does not depend on the traditional JPEG compression trace detection algorithm, so that the type of tamper-detected image can be detected and is not limited to the JPEG format, and detection of different types of image tampering is realized.

DRAWINGS

1 is a schematic diagram of an operating environment of a preferred embodiment of an image tamper detecting system 10 of the present application;

2 is a schematic structural diagram of an optional identification model according to various embodiments of the present application;

FIG. 3 is a schematic flowchart diagram of an embodiment of an image tampering detection method according to the present application.

Detailed ways

In order to make the objects, technical solutions, and advantages of the present application more comprehensible, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It is understood that the specific embodiments described herein are merely illustrative of the application and are not intended to be limiting. All other embodiments obtained by a person of ordinary skill in the art based on the embodiments of the present application without departing from the inventive scope are the scope of the present application.

It should be noted that the descriptions of "first", "second" and the like in the present application are for the purpose of description only, and are not to be construed as indicating or implying their relative importance or implicitly indicating the number of technical features indicated. . Thus, features defining "first" and "second" may include at least one of the features, either explicitly or implicitly. In addition, the technical solutions between the various embodiments may be combined with each other, but must be implemented by those skilled in the art. On the basis of this, when the combination of technical solutions is contradictory or impossible to achieve, it should be considered that the combination of such technical solutions does not exist and is not within the protection scope of the present application.

The application provides an image tampering detection system. Please refer to FIG. 1 , which is a schematic diagram of an operating environment of a preferred embodiment of the image tamper detecting system 10 of the present application.

In the embodiment, the image tamper detecting system 10 is installed and operated in the electronic device 1. The electronic device 1 may include, but is not limited to, a memory 11, a processor 12, and a display 13. Figure 1 shows only the electronic device 1 with components 11-13, but it should be understood that not all illustrated components may be implemented, and more or fewer components may be implemented instead.

The memory 11 comprises at least one type of readable storage medium, which in some embodiments may be an internal storage unit of the electronic device 1, such as a hard disk or memory of the electronic device 1. The memory 11 may also be an external storage device of the electronic device 1 in other embodiments, such as a plug-in hard disk equipped on the electronic device 1, a smart memory card (SMC), and a secure digital device. (Secure Digital, SD) card, flash card, etc. Further, the memory 11 may also include both an internal storage unit of the electronic device 1 and an external storage device. The memory 11 is configured to store application software and various types of data installed in the electronic device 1, such as program codes of the image tamper detecting system 10, and the like. The memory 11 can also be used to temporarily store data that has been output or is about to be output.

The processor 12, in some embodiments, may be a central processing unit (CPU), a microprocessor or other data processing chip for running program code or processing data stored in the memory 11, for example The image tampering detection system 10 and the like are executed.

The display 13 in some embodiments may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, an OLED (Organic Light-Emitting Diode) touch sensor, or the like. The display 13 is for displaying information processed in the electronic device 1 and a user interface for displaying visualization, such as an image to be detected, image tampering information, and the like. The components 11-13 of the electronic device 1 communicate with one another via a system bus.

The image tamper detecting system 10 includes at least one computer readable instructions stored in the memory 11, the at least one computer readable instructions being executable by the processor 12 to implement various embodiments of the present application.

Wherein, when the image tampering detection system 10 is executed by the processor 12, the following steps are implemented:

Step S1: performing block segmentation on the image to be detected, dividing the image to be detected into a plurality of grid image blocks of a preset specification, and extracting a preset number of tamper detecting features from the plurality of grid image blocks.

In this embodiment, the image tampering detection system 10 receives an image tampering detection request sent by a user, which includes, but is not limited to, JPEG, PNG, and the image to be detected. Various formats such as GIF. For example, receiving an image tampering detection request sent by a user through a terminal such as a mobile phone, a tablet computer, or a self-service terminal device, such as receiving an image tampering detection request sent by a user on a client pre-installed in a terminal such as a mobile phone, a tablet computer, or a self-service terminal device. Or receiving an image tampering detection request sent by the user on a browser system in a terminal such as a mobile phone, a tablet, or a self-service terminal device.

Performing block division on the received image to be detected, dividing the image to be detected into a plurality of grid image blocks of preset specifications (such as N*M, N and M are positive integers), and extracting from a plurality of grid image blocks A preset number of tamper detection features are presented. For example, in an optional implementation manner, the image to be detected is first converted from an original color space (for example, an RGB color space) to a YCrCb color space; and the image to be detected is subjected to block segmentation, and the image is to be detected. The detected image is divided into a number of 16*16 grid image blocks. Applying a preset wavelet function to each YCrCb color space component on each grid image block, such as a plurality of levels (for example, 3 levels) of 2D Daubechies Wavelet decomposition, to obtain a plurality of (for example, 30) corresponding pixel coefficient maps (Coefficient Maps), and calculating summary statistical coefficients for each of the pixel coefficient maps (for example, pixel standard deviation, pixel mean value corresponding to each grid image block pixel coefficient map) The sum of the pixels and the coefficients in the horizontal, vertical, and diagonal directions), and apply wavelet functions such as Doubeches Orthogonal wavelets D2-D5 to obtain multiple summary statistical coefficients on each grid image block, and for all The summary statistical coefficients of the grid image blocks are transformed to obtain a plurality of summary statistical coefficients as tamper detection features. For example, 450 features can be obtained as tamper detection features.

In this embodiment, the image to be detected is divided into 16*16 pixel grid image blocks for feature extraction, which is generally divided into 32*32 pixels in the conventional segmentation process, and the grid image is in this embodiment. The block is reduced to a quarter of the conventional segmentation specification, and each detail feature of the image to be detected can be extracted in a more detailed manner, and each 16*16 is obtained by using Dobecy wavelet, Dobecy orthogonal, and the like. The summary statistical coefficient of the grid image block of the pixel is used as the tamper detection feature, which can make the extracted tamper detection feature more accurate.

Step S2: Identify the image to be detected based on a preset number of tamper detecting features and a predetermined recognition model, and analyze whether the image to be detected is tampered according to the recognition result; wherein the predetermined recognition model is A deep neural network model obtained by training a sample picture marked with different tamper trace features in advance.

After extracting a preset number of tamper detecting features from the plurality of mesh image blocks after the image block to be detected is divided, the image to be detected may be identified based on the extracted preset number of tamper detecting features and a predetermined recognition model. And analyzing whether the image to be detected is tampered with according to the recognition result. The recognition model can be continuously trained, learned, verified, optimized, etc. by identifying a large number of sample images marked with different tamper trace features, so as to train them into models capable of accurately identifying various tamper trace features. For example, a plurality of sample pictures corresponding to common picture tampering trace features may be prepared, such as in a picture. A specific location (such as the upper left corner, the lower left corner, the upper right corner, the lower right corner, the top end, the micro end, the middle right) is pasted with a tampering image of a signature, a watermark, etc., and the corresponding tampering trace features can be marked, and different tampering marks are marked for the label. The sample picture of the feature is trained, learned, verified, optimized, etc. to generate a recognition model. The recognition model may be a Convolutional Neural Network (CNN) model, a stack-type automatic encoder, etc., and is not limited herein.

In an optional implementation manner, the training process of the recognition model is as follows:

C. Obtain a preset number (for example, 100,000) of image samples;

D. Perform block segmentation on each image sample, divide each image sample into 16*16 grid image blocks, and mark each grid image block, and mark the tampered grid image block as 1, untamed The grid image block is labeled as 0 to tamper with the trace features of each image sample. In this embodiment, the final label is not the tampering part of the image sample, but the tampering trace of the image sample, which is equivalent to the feature of training the tampering trace, instead of tampering with the content, and marking the tampering trace as 1 ( Positive class), other labels are 0 (negative class), the segmented grid image block is reduced to a quarter of the traditional segmentation specification (32*32 pixels to 16*16 pixels), which can better train and identify The traces of slender tampering that are common in practical applications are more practical.

E. All image samples are divided into a first data set and a second data set according to a ratio of X:Y (for example, 8:2), and the number of image samples in the first data set is greater than the number of image samples in the second data set, first The data set is used as a training set, and the second data set is used as a test set, wherein X is greater than 0 and Y is greater than 0;

F. training the predetermined recognition model by using each image sample in the first data set;

G. verifying the accuracy of the trained recognition model by using each image sample in the second data set. If the accuracy is greater than or equal to the preset accuracy (for example, 95%), the training ends, or if the accuracy is less than the preset accuracy Rate, then increase the number of image samples and re-execute the above steps D, E, F and G until the accuracy of the trained recognition model is greater than or equal to the preset accuracy.

When the image to be detected is identified by using the trained recognition model, and the image to be detected is falsified according to the recognition result, if the recognition model is used to identify that the image to be detected has a tampering mark feature, It is analyzed that the image to be detected has been tampered with; if the image to be detected does not have the tampering trace feature by using the predetermined recognition model, then the image to be detected is analyzed and not falsified.

Compared with the prior art, the embodiment identifies the image to be detected by using a deep neural network model obtained by training a plurality of sample pictures marked with different tamper trace features, and analyzes whether the image to be detected is tampered with according to the recognition result. . Because it is through The image to be detected is divided into a plurality of grid image blocks, the tamper detection features are extracted from the respective grid image blocks, and the extracted tamper detection features are identified by the pre-trained deep neural network model to identify Whether the image to be detected has tampering marks does not depend on the traditional JPEG compression trace detection algorithm, so that the type of tamper-detected image can be detected and is not limited to the JPEG format, and detection of different types of image tampering is realized.

In an optional embodiment, based on the foregoing embodiment of FIG. 1, the predetermined recognition model is a deep neural network model that is a fully connected layer, and the number of neurons in the deep neural network model They are 450, 500, 256, 128, and 2, respectively, where 450 is the number of input features, 500, 256, and 128 are the preset number of hidden features, and 2 is the number of final classification categories.

Referring to FIG. 2, it is a schematic structural diagram of an optional identification model according to various embodiments of the present application. Extracting a preset number, such as 450 tamper detection features, from a plurality of grid image blocks after the image block to be detected is segmented, and inputting the extracted 450 tamper detection features as input features of the deep neural network model into the depth neural network. The input layer of the model is processed by the hidden layer 1 of the deep neural network model, the hidden layer 2, and the hidden layer 3 to perform feature reduction filtering, etc., to reach the final classification layer, and output the final two classification results. In the present embodiment, it is "not falsified" and "tampered", thereby completing image tampering detection of the image to be detected.

As shown in FIG. 3, FIG. 3 is a schematic flowchart of an image tampering detection method according to an embodiment of the present invention. The image tampering detection method includes the following steps:

Step S10: Perform block segmentation on the image to be detected, divide the image to be detected into a plurality of grid image blocks of a preset specification, and extract a preset number of tamper detecting features from the plurality of grid image blocks.

In this embodiment, the image tampering detection system receives an image tampering detection request sent by the user, which includes, but is not limited to, JPEG, PNG, GIF, and the like. For example, receiving an image tampering detection request sent by a user through a terminal such as a mobile phone, a tablet computer, or a self-service terminal device, such as receiving an image tampering detection request sent by a user on a client pre-installed in a terminal such as a mobile phone, a tablet computer, or a self-service terminal device. Or receiving an image tampering detection request sent by the user on a browser system in a terminal such as a mobile phone, a tablet, or a self-service terminal device.

Performing block division on the received image to be detected, dividing the image to be detected into a plurality of grid image blocks of preset specifications (such as N*M, N and M are positive integers), and extracting from a plurality of grid image blocks A preset number of tamper detection features are presented. For example, in an optional implementation manner, the image to be detected is first converted from an original color space (for example, an RGB color space) to a YCrCb color space; and the image to be detected is subjected to block segmentation, and the image is to be detected. The detected image is divided into a number of 16*16 grid image blocks. On each grid image block A preset wavelet function is applied to each YCrCb color space component such as a plurality of levels (for example, 3 levels) of 2D Daubechies Wavelet decomposition to obtain a plurality of (for example, 30) corresponding ones. a coefficient map (Coefficient Map), which calculates a summary statistical coefficient for each of the pixel coefficient mappings (for example, a pixel standard deviation corresponding to a pixel coefficient map of each grid image block, a pixel mean and a pixel sum, and a horizontal direction, a vertical direction, and a pair The coefficient of the angular direction), and apply wavelet functions such as Doubeches Orthogonal wavelets D2-D5 to obtain multiple summary statistical coefficients on each grid image block, and transform the summary statistical coefficients of all grid image blocks. , multiple summary statistical coefficients are obtained as tamper detection features. For example, 450 features can be obtained as tamper detection features.

Step S20: Identify the image to be detected based on a preset number of tamper detecting features and a predetermined recognition model, and analyze whether the image to be detected is tampered according to the recognition result; wherein the predetermined recognition model is A deep neural network model obtained by training a sample picture marked with different tamper trace features in advance.

After extracting a preset number of tamper detecting features from the plurality of mesh image blocks after the image block to be detected is divided, the image to be detected may be identified based on the extracted preset number of tamper detecting features and a predetermined recognition model. And analyzing whether the image to be detected is tampered with according to the recognition result. The recognition model can be continuously trained, learned, verified, optimized, etc. by identifying a large number of sample images marked with different tamper trace features, so as to train them into models capable of accurately identifying various tamper trace features. For example, a plurality of sample pictures corresponding to a common picture tampering trace feature may be prepared, such as a signature, a watermark at a specific position of the picture (eg, upper left corner, lower left corner, upper right corner, lower right corner, top end, micro end, medium middle). The tampering pictures can be labeled with different tamper trace features, and the sample images with different tamper trace features are trained, learned, verified, optimized, etc. to generate the recognition model. The recognition model may be a Convolutional Neural Network (CNN) model, a stack-type automatic encoder, etc., and is not limited herein.

C. Obtain a preset number (for example, 100,000) of image samples;

D. Perform block segmentation on each image sample, divide each image sample into 16*16 grid image blocks, and mark each grid image block, and mark the tampered grid image block as 1, untamed The grid image block is labeled 0 to tamper with each image sample Marking of trace features. In this embodiment, the final label is not the tampering part of the image sample, but the tampering trace of the image sample, which is equivalent to the feature of training the tampering trace, instead of tampering with the content, and marking the tampering trace as 1 ( Positive class), other labels are 0 (negative class), the segmented grid image block is reduced to a quarter of the traditional segmentation specification (32*32 pixels to 16*16 pixels), which can better train and identify The traces of slender tampering that are common in practical applications are more practical.

Compared with the prior art, the embodiment identifies the image to be detected by using a deep neural network model obtained by training a plurality of sample pictures marked with different tamper trace features, and analyzes whether the image to be detected is tampered with according to the recognition result. . The tamper detection feature is extracted from each grid image block by dividing the image to be detected into a plurality of grid image blocks, and the extracted tamper detection feature is identified by using a pre-trained deep neural network model to It is recognized that the image to be detected has tampering marks, and does not depend on the traditional JPEG compression trace detection algorithm, so that the type of tamper-detected image can be detected and is not limited to the JPEG format, and detection of different types of image tampering is realized.

In an optional embodiment, based on the foregoing embodiment, the predetermined recognition model is a deep neural network model that is a fully connected layer, and the number of neurons in the deep neural network model is 450 respectively. 500, 256, 128, 2, wherein 450 is the number of input features, 500, 256, 128 are the number of preset hidden features, and 2 is the number of final classification categories.

Moreover, the present application also provides a computer readable storage medium storing an image tamper detecting system, the image tamper detecting system being executable by at least one processor to cause the at least one processor The steps of the image tampering detection method in the above embodiment are as follows, and the specific implementation processes of the steps S10 and S20 of the image tampering detection method are as described above, and are not described herein again.

It is to be understood that the term "comprises", "comprising", or any other variants thereof, is intended to encompass a non-exclusive inclusion, such that a process, method, article, or device comprising a series of elements includes those elements. It also includes other elements that are not explicitly listed, or elements that are inherent to such a process, method, article, or device. An element that is defined by the phrase "comprising a ..." does not exclude the presence of additional equivalent elements in the process, method, item, or device that comprises the element.

Through the description of the above embodiments, those skilled in the art can clearly understand that the foregoing embodiment method can be implemented by means of software plus a necessary general hardware platform, and can also be implemented by hardware, but in many cases, the former is A better implementation. Based on such understanding, the technical solution of the present application, which is essential or contributes to the prior art, may be embodied in the form of a software product stored in a storage medium (such as ROM/RAM, disk, The optical disc includes a number of instructions for causing a terminal device (which may be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to perform the methods described in various embodiments of the present application.

The preferred embodiments of the present application have been described above with reference to the drawings, and are not intended to limit the scope of the application. The serial numbers of the embodiments of the present application are merely for the description, and do not represent the advantages and disadvantages of the embodiments. Additionally, although logical sequences are shown in the flowcharts, in some cases the steps shown or described may be performed in a different order than the ones described herein.

A person skilled in the art can implement the present application in various variants without departing from the scope and spirit of the present application. For example, the features of one embodiment can be used in another embodiment to obtain another embodiment. Any modifications, equivalent substitutions and improvements made within the technical concept of the application should be within the scope of the application.

Claims

An electronic device, comprising: a memory, a processor, on the memory, an image tamper detecting system operable on the processor, wherein the image tamper detecting system is The following steps are implemented during execution:

A. performing block segmentation on the image to be detected, dividing the image to be detected into a plurality of grid image blocks of preset specifications, and extracting a preset number of tamper detecting features from the plurality of grid image blocks;

B. Identifying the image to be detected based on a preset number of tamper detection features and a predetermined recognition model, and analyzing whether the image to be detected is tampered according to the recognition result; wherein the predetermined recognition model is A deep neural network model obtained by training sample pictures marked with different tamper trace features.
The electronic device according to claim 1, wherein when the image tampering detection system is executed by the processor to implement the step A, the method comprises:

Mapping the image to be detected from the original color space to the YCrCb color space;

Performing block division on the image to be detected, and dividing the image to be detected into a plurality of 16*16 grid image blocks;

Applying a two-dimensional Dobey wavelet decomposition operation on each YCrCb color space component on each grid image block to obtain a corresponding pixel coefficient mapping, calculating a summary statistical coefficient for each of the pixel coefficient mappings, and applying Dobecy Orthogonal Daubechies Orthogonal wavelets D2-D5 transforms the summary statistical coefficients of all grid image blocks to obtain 450 features as tamper detection features.
The electronic device according to claim 2, wherein the predetermined recognition model is a deep neural network model which is a fully connected layer, and the number of neurons in the deep neural network model is 450, 500, respectively. 256, 128, 2, wherein 450 is the number of input features, 500, 256, 128 are the number of preset hidden features, and 2 is the number of final classification categories.
The electronic device according to claim 1, wherein the training process of the predetermined recognition model is as follows:

C. Obtain a preset number of image samples;

D. Perform block segmentation on each image sample, divide each image sample into 16*16 grid image blocks, and mark each grid image block, and mark the tampered grid image block as 1, untamed The grid image block is labeled as 0 to mark the tampering trace features of each image sample;

E. Divide all image samples into the first data set and the second number according to the ratio of X:Y According to the data set, the number of image samples in the first data set is greater than the number of image samples in the second data set, the first data set is used as a training set, and the second data set is used as a test set, wherein X is greater than 0 and Y is greater than 0;

F. training the predetermined recognition model by using each image sample in the first data set;

G. verifying the accuracy of the trained recognition model by using each image sample in the second data set. If the accuracy is greater than or equal to the preset accuracy, the training ends, or if the accuracy is less than the preset accuracy, the image sample is added. The number and re-execute steps D, E, F and G above.
The electronic device according to claim 2, wherein the training process of the predetermined recognition model is as follows:

C. Obtain a preset number of image samples;

D. Perform block segmentation on each image sample, divide each image sample into 16*16 grid image blocks, and mark each grid image block, and mark the tampered grid image block as 1, untamed The grid image block is labeled as 0 to mark the tampering trace features of each image sample;

E. Dividing all image samples into a first data set and a second data set according to a ratio of X:Y, the number of image samples in the first data set is greater than the number of image samples in the second data set, and the first data set is used as a training set, The second data set is used as a test set, wherein X is greater than 0 and Y is greater than 0;

F. training the predetermined recognition model by using each image sample in the first data set;

G. verifying the accuracy of the trained recognition model by using each image sample in the second data set. If the accuracy is greater than or equal to the preset accuracy, the training ends, or if the accuracy is less than the preset accuracy, the image sample is added. The number and re-execute steps D, E, F and G above.
The electronic device according to claim 1, wherein the analyzing whether the image to be detected is tampered according to the recognition result comprises:

If the image to be detected has a tamper-evident feature by using the predetermined recognition model, analyzing that the image to be detected has been tampered with;

If the predetermined recognition model is used to identify that the image to be detected does not have a tampering trace feature, then analyzing the image to be detected is not falsified.
The electronic device according to claim 2, wherein the analyzing whether the image to be detected is tampered according to the recognition result comprises:

Recognizing that the image to be detected has a flaw by using the predetermined recognition model Changing the trace feature, analyzing that the image to be detected has been tampered with;

If the predetermined recognition model is used to identify that the image to be detected does not have a tampering trace feature, then analyzing the image to be detected is not falsified.
An image tampering detecting method, wherein the image tampering detecting method comprises:

Step 1: Perform block segmentation on the detected image, divide the image to be detected into a plurality of grid image blocks of preset specifications, and extract a preset number of tamper detecting features from the plurality of grid image blocks;

Step 2: Identify the image to be detected based on a preset number of tamper detection features and a predetermined recognition model, and analyze whether the image to be detected is tampered according to the recognition result; wherein the predetermined recognition model is A deep neural network model obtained by training a sample picture marked with different tamper trace features in advance.
The image tampering detecting method according to claim 8, wherein the step one comprises:

Mapping the image to be detected from the original color space to the YCrCb color space;

Performing block division on the image to be detected, and dividing the image to be detected into a plurality of 16*16 grid image blocks;

Applying a two-dimensional Dobey wavelet decomposition operation on each YCrCb color space component on each grid image block to obtain a corresponding pixel coefficient mapping, calculating a summary statistical coefficient for each of the pixel coefficient mappings, and applying Dobecy Orthogonal Daubechies Orthogonal wavelets D2-D5 transforms the summary statistical coefficients of all grid image blocks to obtain 450 features as tamper detection features.
The image tampering detecting method according to claim 9, wherein the predetermined recognition model is a deep neural network model which is a fully connected layer, and the number of neurons in the deep neural network model is 450, respectively. 500, 256, 128, 2, wherein 450 is the number of input features, 500, 256, 128 are the number of preset hidden features, and 2 is the number of final classification categories.
The image tampering detecting method according to claim 8, wherein the training process of the predetermined recognition model is as follows:

C. Obtain a preset number of image samples;

D. Perform block segmentation on each image sample, divide each image sample into 16*16 grid image blocks, and mark each grid image block, and mark the tampered grid image block as 1, untamed The grid image block is labeled as 0 to mark the tampering trace features of each image sample;

E. Dividing all image samples into a first data set and a second data set according to a ratio of X:Y, the number of image samples in the first data set is greater than the number of image samples in the second data set, and the first data set is used as a training set, The second data set is used as a test set, wherein X is greater than 0 and Y is greater than 0;

F. training the predetermined recognition model by using each image sample in the first data set;

G. verifying the accuracy of the trained recognition model by using each image sample in the second data set. If the accuracy is greater than or equal to the preset accuracy, the training ends, or if the accuracy is less than the preset accuracy, the image sample is added. The number and re-execute steps D, E, F and G above.
The image tampering detecting method according to claim 9, wherein the training process of the predetermined recognition model is as follows:

C. Obtain a preset number of image samples;

D. Perform block segmentation on each image sample, divide each image sample into 16*16 grid image blocks, and mark each grid image block, and mark the tampered grid image block as 1, untamed The grid image block is labeled as 0 to mark the tampering trace features of each image sample;

E. Dividing all image samples into a first data set and a second data set according to a ratio of X:Y, the number of image samples in the first data set is greater than the number of image samples in the second data set, and the first data set is used as a training set, The second data set is used as a test set, wherein X is greater than 0 and Y is greater than 0;

F. training the predetermined recognition model by using each image sample in the first data set;

G. verifying the accuracy of the trained recognition model by using each image sample in the second data set. If the accuracy is greater than or equal to the preset accuracy, the training ends, or if the accuracy is less than the preset accuracy, the image sample is added. The number and re-execute steps D, E, F and G above.
The image tampering detecting method according to claim 8, wherein the step of analyzing whether the image to be detected is falsified according to the recognition result comprises:

If the image to be detected has a tamper-evident feature by using the predetermined recognition model, analyzing that the image to be detected has been tampered with;

If the predetermined recognition model is used to identify that the image to be detected does not have a tampering trace feature, then analyzing the image to be detected is not falsified.
The image tampering detecting method according to claim 9, wherein the step of analyzing whether the image to be detected is falsified according to the recognition result comprises:

If the image to be detected has a tamper-evident feature by using the predetermined recognition model, analyzing that the image to be detected has been tampered with;

If the predetermined recognition model is used to identify that the image to be detected does not have a tampering trace feature, then analyzing the image to be detected is not falsified.
A computer readable storage medium, characterized in that the computer readable storage medium stores an image tamper detecting system, the image tamper detecting system being executable by at least one processor to cause the at least one processor to execute The following steps:

S1, performing block segmentation on the image to be detected, dividing the image to be detected into a plurality of grid image blocks of a preset specification, and extracting a preset number of tamper detecting features from the plurality of grid image blocks;

S2, identifying the image to be detected based on a preset number of tamper detection features and a predetermined recognition model, and analyzing whether the image to be detected is tampered according to the recognition result; wherein the predetermined recognition model is A deep neural network model obtained by training sample pictures marked with different tamper trace features.
The computer readable storage medium according to claim 15, wherein when the image tampering detection system is executed by the processor to implement the step S1, the method comprises:

Mapping the image to be detected from the original color space to the YCrCb color space;

Performing block division on the image to be detected, and dividing the image to be detected into a plurality of 16*16 grid image blocks;

Applying a two-dimensional Dobey wavelet decomposition operation on each YCrCb color space component on each grid image block to obtain a corresponding pixel coefficient mapping, calculating a summary statistical coefficient for each of the pixel coefficient mappings, and applying Dobecy Orthogonal Daubechies Orthogonal wavelets D2-D5 transforms the summary statistical coefficients of all grid image blocks to obtain 450 features as tamper detection features.
The computer readable storage medium according to claim 16, wherein the predetermined recognition model is a deep neural network model that is a fully connected layer, and the number of neurons in the deep neural network model is 450 respectively. 500, 256, 128, 2, wherein 450 is the number of input features, 500, 256, 128 are the number of preset hidden features, and 2 is the number of final classification categories.
The computer readable storage medium of claim 15, wherein the training process of the predetermined recognition model is as follows:

C. Obtain a preset number of image samples;

D. Perform block segmentation on each image sample, divide each image sample into 16*16 grid image blocks, and mark each grid image block to tamper with the grid image block. Labeled as 1, the untampered grid image block is marked as 0 to tamper with the trace features of each image sample;

E. Dividing all image samples into a first data set and a second data set according to a ratio of X:Y, the number of image samples in the first data set is greater than the number of image samples in the second data set, and the first data set is used as a training set, The second data set is used as a test set, wherein X is greater than 0 and Y is greater than 0;

F. training the predetermined recognition model by using each image sample in the first data set;

G. verifying the accuracy of the trained recognition model by using each image sample in the second data set. If the accuracy is greater than or equal to the preset accuracy, the training ends, or if the accuracy is less than the preset accuracy, the image sample is added. The number and re-execute steps D, E, F and G above.
The computer readable storage medium of claim 16 wherein the predetermined training process of the recognition model is as follows:

C. Obtain a preset number of image samples;

D. Perform block segmentation on each image sample, divide each image sample into 16*16 grid image blocks, and mark each grid image block, and mark the tampered grid image block as 1, untamed The grid image block is labeled as 0 to mark the tampering trace features of each image sample;

E. Dividing all image samples into a first data set and a second data set according to a ratio of X:Y, the number of image samples in the first data set is greater than the number of image samples in the second data set, and the first data set is used as a training set, The second data set is used as a test set, wherein X is greater than 0 and Y is greater than 0;

F. training the predetermined recognition model by using each image sample in the first data set;

G. verifying the accuracy of the trained recognition model by using each image sample in the second data set. If the accuracy is greater than or equal to the preset accuracy, the training ends, or if the accuracy is less than the preset accuracy, the image sample is added. The number and re-execute steps D, E, F and G above.
The computer readable storage medium according to claim 15, wherein the step of analyzing whether the image to be detected is falsified according to the recognition result comprises:

If the image to be detected has a tamper-evident feature by using the predetermined recognition model, analyzing that the image to be detected has been tampered with;

If the predetermined recognition model is used to identify that the image to be detected does not have a tampering trace feature, then analyzing the image to be detected is not falsified.