A matching method and device for camera fingerprints affected by software noise

Abstract

The present invention discloses a method and apparatus for matching camera fingerprints affected by software noise. The method includes: obtaining a to-be-tested image I q and a target matching camera fingerprint K r ; according to the to-be-tested image I q , extracting the corresponding to-be-tested camera fingerprint K q and calculating the basic similarity Cor_pair with the target matching camera fingerprint K r ; estimating the software noise similarity Cor_software_noise of the device model and the shooting software corresponding to the target matching camera fingerprint K r ; calculating the ratio Ratio of Cor_pair to Cor_software_noise as the new camera fingerprint similarity, and comparing the ratio Ratio with a preset threshold to obtain a matching result. By calculating the similarity between the to-be-tested camera fingerprint and the camera fingerprints of other devices of the same model and taking the average value, the influence degree of software noise on the similarity is estimated, and complex schemes such as reverse engineering are avoided to deduce the generation process of software noise. At the same time, the ratio Ratio is obtained in a ratio manner, and it is directly determined whether there are identical components in the similarity between the to-be-tested camera fingerprint and the target matching camera fingerprint, and the matching result is obtained.

Description

Technical Field

[0001] The present invention belongs to the technical field of hardware fingerprint matching, and in particular relates to a matching method and device for camera fingerprints affected by software noise. Background Art

[0002] The camera fingerprint based on the PRNU (photo-response non-uniformity) property is a reliable hardware fingerprint of the imaging sensor that exists in the photos it takes. Each imaging sensor is composed of millions of silicon chips and obtains image data through photoelectric conversion. However, due to the imperfection of the production process, each silicon chip cannot be guaranteed to be completely consistent, so it is almost impossible to have a completely consistent camera fingerprint. When it is necessary to determine whether the image pair is taken by the same device, the image is usually extracted through a denoising filter, and then the similarity is calculated and compared with the preset threshold to draw a judgment conclusion. If the similarity result is greater than the threshold, then the same camera fingerprint exists in the extracted noise component, and the two photos are taken by the same device; otherwise, the similarity result tends to 0, the camera fingerprint in the noise component is different, and the two photos are not taken by the same device.

[0003] Nowadays, whether it is digital cameras or mobile phones, post-processing programs are becoming more and more powerful. While enhancing the image, they will also have a certain impact on the extracted camera fingerprint. The camera fingerprint extracted from the photos taken by such devices not only contains the hardware fingerprint caused by PRNU, but also the software fingerprint left after software processing. Therefore, when calculating the similarity of two camera fingerprints, the result includes the similarity of software fingerprints in addition to the camera fingerprint similarity. The same software post-processing algorithm produces the same software noise. Therefore, even if two photos are taken by different devices, as long as the shooting program is the same, the similarity result is no longer 0, and it is likely to exceed the preset threshold. Therefore, both the threshold selection and the discrimination accuracy are affected. In particular, in the scenario of large-scale user login verification, the discrimination method of unified threshold cannot be applied to devices of different brands and models (with different software noise intensities).

[0004] In the process of implementing the present invention, the inventors found that there are at least the following problems in the prior art:

[0005] It is difficult to effectively model software noise: because image processing algorithms are core secrets for major brand manufacturers and third-party camera software, reverse engineering is not feasible from the perspective of intellectual property protection and cracking difficulty; even if the post-processing algorithm can be obtained, it is particularly troublesome to model and set thresholds for each algorithm because the intensity of software noise varies; traditional denoising filters are almost unable to remove software fingerprints while retaining hardware fingerprints; traditional camera fingerprint enhancement algorithms cannot independently enhance hardware fingerprints to separate software fingerprints. Summary of the invention

[0006] In view of the shortcomings of the prior art, the present invention proposes a matching method for camera fingerprints affected by software noise, which can reduce the impact of camera software on existing hardware fingerprint matching. By detecting the relevant components of hardware fingerprint correlation from the traditional similarity value results, the accuracy of recognition and matching is improved.

[0007] According to a first aspect of an embodiment of the present application, a matching method for a camera fingerprint affected by software noise is provided, comprising:

[0008] Get the image to be tested I q Match the camera fingerprint K with the target r ;

[0009] According to the image to be tested I q , extract the corresponding camera fingerprint K to be tested q And match the camera fingerprint K with the target r Calculate the basic similarity Cor_pair;

[0010] Estimated target matching camera fingerprint K r The software noise similarity between the corresponding device model and the shooting software is Cor_software_noise;

[0011] The ratio of Cor_pair to Cor_software_noise is calculated as the new camera fingerprint similarity, and the ratio is compared with the preset threshold to obtain the matching result.

[0012] Furthermore, the steps of estimating software noise similarity are as follows:

[0013] Use the camera fingerprint K that matches the target r The corresponding device model and shooting software are the same as the device Z = {Z1,…,Z n}Several photos taken, extract the corresponding camera fingerprint Calculate the fingerprint K of the camera to be tested q With camera fingerprint K z Basic similarity of camera fingerprint Cor z Take the average to get the estimated software noise similarity Cor_software_noise.

[0014] Furthermore, the ratio of Cor_pair to Cor_software_noise is calculated as:

[0015]

[0016] or

[0017]

[0018] Wherein, C is a preset constant.

[0019] Furthermore, the step of obtaining the matching result is as follows: comparing the preset threshold Threshold with Ratio, if Ratio is greater than Threshold, then the camera fingerprint K q With K r Match; otherwise camera fingerprint K q With K r No match.

[0020] Furthermore, the target matches the camera fingerprint K r Photo I taken by device r r Estimated.

[0021] Further, estimate the camera fingerprint K q and K r Here are the steps:

[0022] I will take the photo r Use the denoising algorithm to denoise and get the denoised image

[0023] I will take the photo r And the denoised image Subtract and get the noise residual W contains the camera fingerprint K r and other noises.

[0024] Furthermore, a peak correlation energy ratio, a cross-correlation coefficient or a normalized cross-correlation coefficient is used as the basic similarity.

[0025] According to a second aspect of an embodiment of the present application, a matching device for a camera fingerprint affected by software noise is provided, comprising:

[0026] Acquisition module, used to acquire the image to be tested I q Match the camera fingerprint K with the target r ;

[0027] Extraction module, used for extracting the image I q , extract the corresponding camera fingerprint K to be tested q ;

[0028] The first calculation module is used to calculate the target matching camera fingerprint K according to the selected basic similarity calculation method. r And the camera fingerprint K to be tested q Camera fingerprint similarity Cor_pair;

[0029] The second computing module is used to match the camera fingerprint K with the targetr The corresponding device model and shooting software are the same as the device Z = {Z1,…,Z n}Several photos taken, extract the corresponding camera fingerprint Calculate the image K to be tested q With camera fingerprint K z Camera fingerprint similarity Cor z Take the average to get the estimated software noise similarity Cor_software_noise;

[0030] The third calculation module is used to calculate the ratio Ratio of Cor_pair and Cor_software_noise;

[0031] The comparison module is used to compare the preset threshold Threshold with Ratio. If Ratio is greater than Threshold, the camera fingerprint K q With K r Match; otherwise camera fingerprint K q With K r No match.

[0032] According to a third aspect of an embodiment of the present application, there is provided an electronic device, including:

[0033] one or more processors;

[0034] A memory for storing one or more programs;

[0035] When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in the first aspect.

[0036] According to a fourth aspect of an embodiment of the present application, a computer-readable storage medium is provided, on which computer instructions are stored, and when the instructions are executed by a processor, the steps of the method described in the first aspect are implemented.

[0037] The technical solution provided by the embodiments of the present application may have the following beneficial effects:

[0038] It can be seen from the above embodiments that the present application calculates the similarity between the camera fingerprint to be tested and the camera fingerprints of other devices of the same model and takes the average value to estimate the influence of software noise on the similarity, avoiding the complex schemes such as reverse engineering to deduce the generation process of software noise. At the same time, the Ratio is obtained by the camera fingerprint similarity ratio, and the similarity between the camera fingerprint to be tested and the target matching camera fingerprint is directly judged whether there is the same component, and the matching judgment result is obtained.

[0039] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0041] Figure 1 The invention is a flow chart showing a method for matching a camera fingerprint affected by software noise according to an exemplary embodiment.

[0042] Figure 2 is a distribution diagram of a matching situation of a traditional algorithm (PCE) according to an exemplary embodiment.

[0043] Figure 3 The algorithm of the present invention (Ratio + _PCE) matching situation distribution diagram.

[0044] Figure 4 The invention is a block diagram of a camera fingerprint matching device affected by software noise according to an exemplary embodiment.

[0045] Figure 5 is a schematic diagram of an electronic device according to an exemplary embodiment. DETAILED DESCRIPTION

[0046] Here, exemplary embodiments will be described in detail, and examples thereof are shown in the accompanying drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application.

[0047] The terms used in this application are for the purpose of describing specific embodiments only and are not intended to limit this application. The singular forms of "a", "said" and "the" used in this application and the appended claims are also intended to include plural forms unless the context clearly indicates other meanings. It should also be understood that the term "and / or" used herein refers to and includes any or all possible combinations of one or more associated listed items.

[0048] It should be understood that although the terms first, second, third, etc. may be used in the present application to describe various information, these information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of the present application, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".

[0049] Figure 1 is a flow chart of a method for matching a camera fingerprint affected by software noise according to an exemplary embodiment. Figure 1 As shown, the method is applied in a terminal and may include the following steps:

[0050] Step S11: Obtain the image I to be tested q Match the camera fingerprint K with the target r ;

[0051] Step S12: According to the image to be tested I q , extract the corresponding camera fingerprint K to be tested q And match the camera fingerprint K with the target r Calculate the basic similarity Cor_pair;

[0052] Step S13: Estimate the target matching camera fingerprint K r The software noise similarity between the corresponding device model and the shooting software is Cor_software_noise;

[0053] Step S14: Calculate the ratio of Cor_pair to Cor_software_noise as the new camera fingerprint similarity, and compare the ratio with a preset threshold to obtain a matching result.

[0054] It can be seen from the above embodiments that the present application calculates the similarity between the camera fingerprint to be tested and the camera fingerprints of other devices of the same model and takes the average value to estimate the influence of software noise on the similarity, avoiding the complex schemes such as reverse engineering to deduce the generation process of software noise. At the same time, the Ratio is obtained by the camera fingerprint similarity ratio, and the similarity between the camera fingerprint to be tested and the target matching camera fingerprint is directly judged whether there is the same component, and the matching judgment result is obtained.

[0055] In the specific implementation of step S11, the image to be tested I is obtained. q Match the camera fingerprint K with the target r ;

[0056] Specifically, the discriminant image I q Whether Kr The corresponding device is used to take the photo. Among them, K r Photos taken in advance by device r I r Estimated. Estimated camera fingerprint K r The steps are as follows: First, take the photo I r Use the denoising algorithm to denoise and get the denoised image Then, subtract them to get the noise residual The camera fingerprint is mainly in the high-frequency area of ​​the image, so most of the image content is filtered, and W contains the camera fingerprint K r and other noises. In order to obtain a higher quality camera fingerprint K r , usually take multiple photos first, and then use the fingerprint enhancement algorithm to increase the signal-to-noise ratio of the camera fingerprint to obtain a camera fingerprint with less noise. Among them, the denoising algorithm can be one of Wavelet-Based Filter, Context-Adaptive Interpolator (CAI), 2-Pixel Approach, Adaptive Spatial (AS) Filtering, Content Adaptive Guided Image (CAGI) Filtering, Perona-Malik Diffusion (PMD) Filter, Total Variation (TV) Filter and Block-Matching and 3D (BM3D) Algorithm. In one embodiment, Wavelet-Based Filter can be used as the basic denoising algorithm. The fingerprint enhancement algorithm can be one of Maximum Likelihood Estimator (MLE) and Weighted Averaging (WA). In one embodiment, MLE can be used as the basic fingerprint enhancement algorithm.

[0057] In the specific implementation of step S12, according to the image to be tested I q , extract the corresponding camera fingerprint K to be tested q And match the camera fingerprint K with the target r Calculate the basic similarity Cor_pair;

[0058] Specifically, first, the photo I q Use the denoising algorithm to denoise and get the denoised image Then, subtract them to get the noise residual W contains the camera fingerprint K r and other noise, denoted as K q .

[0059] In one embodiment, the Peak to Correlation Energy (PCE) may be used as the basic similarity, and the Cross Correlation (CC) and the Normalized Cross Correlation (NCC) may also be used.

[0060] In the specific implementation of step S13, the target matching camera fingerprint K is estimated r The software noise similarity between the corresponding device model and the shooting software is Cor_software_noise;

[0061] Specifically, the camera fingerprint K is matched with the target r The corresponding device model and shooting software are the same as the device Z = {Z1,…,Z n}Several photos taken, extract the corresponding camera fingerprint Calculate the image K to be tested q With camera fingerprint K z Camera fingerprint similarity Cor z Take the average to get the estimated software noise similarity Cor_software_noise;

[0062] Specifically, extract the camera fingerprint K r The method is the same as that in step S11 and will not be described in detail here.

[0063] In the specific implementation of step S14, the ratio Ratio of Cor_pair to Cor_software_noise is calculated as the new camera fingerprint similarity, and the ratio Ratio is compared with a preset threshold to obtain a matching result;

[0064] Specifically, the ratio of Cor_pair to Cor_software_noise is calculated as follows:

[0065]

[0066] or

[0067]

[0068] Wherein, C is a preset constant.

[0069] Compare the preset threshold Threshold with Ratio. If Ratio is greater than Threshold, it means that there is a hardware fingerprint similarity component in Ratio, and the camera fingerprint K q With K rMatch; otherwise, it means that there is no hardware fingerprint similarity component in Ratio, and the camera fingerprint K q With K r No match.

[0070] Specifically, if the image pair to be tested does not match, then Cor_pair only contains software noise similarity, and Cor_software_noise also contains software noise similarity, then the theoretical result of Ratio is 1, and a value near 1 will be obtained; if the image pair to be tested matches each other, then Cor_pair contains hardware fingerprint similarity components in addition to software noise similarity, then the theoretical result of Ratio will be > 1, and a value far from 1 will be obtained. At this time, a division threshold Threshold can be set between the two results to determine whether they match.

[0071] Experimental data description:

[0072] Specifically, a model (Samsung A51) is used as an example. Similarity Cor is exemplified by PCE.

[0073] Example of judging the image to be tested I q The corresponding camera fingerprint K q Match the camera fingerprint K with the target r First, prepare the fingerprint of the camera of other devices according to the formula requirements. In this example, n is 28. Calculate The calculation results are as follows:

[0074]

[0075] Cor z Take the average to get the estimated software noise similarity Cor_software_noise = 5.818573697899166e+04

[0076] In the case of mismatch, Cor(K q ,K r )=57727.0147591175

[0077]

[0078] In the matching case, Cor(K q ,K r )=1.387994478578573e+05

[0079]

[0080] An empirical threshold between 1.25 and 1.85 can be used to distinguish between matching and mismatching cases.

[0081] Since the above formula can only be calculated when software noise exists, if there is no software noise, the denominator will be meaningless. In order to make the present invention compatible with models that are not affected by software noise, the following improvements will be made: a constant C with a smaller similarity than the corresponding traditional one (for example, PCE can take C=4-8) is added to both the numerator and the denominator to avoid taking the denominator as 0, and at the same time, it will not cause too much interference to the result:

[0082]

[0083] In this case, we also conducted experiments on the model affected by software noise (Samsung A51)

[0084] In case of mismatch: Cor(K q ,K r )=57727.0147591175

[0085]

[0086] Matching case: Cor(K Q ,K R )=1.387994478578573e+05

[0087]

[0088] The results show that the difference with Ratio is very small, because the constant is very small and has almost no effect on the results.

[0089] For devices not affected by software noise (VIVO IQOO U1X), first prepare the camera fingerprint of other devices according to the public requirements. In this example, n is 28. Calculate The calculation results are as follows:

[0090] -1.11295547860759 1.61567732644653 0.3716598670588400.372778275160196 -2.69205632274085 1.22414184229707 0.0532123673450589 -2.33416400772466 -0.0119119759846827 0.151930002420765 -0.5860492481899163.71586899808310 -3.53780479210240 -2.20741942083776 0.285220277302892 -2.31405254902106 4.88042966667315e-05 1.32175564054650 2.234782149823931.05166290604463 0.294067908720449 0.656156762087179 -0.07432097897938833.48683564672390 0.797253125616931 0.0705461322532100 -0.298039271240209 -2.43782697838117

[0091] Cor z Take the average to get the estimated software noise similarity Cor_software_noise = 0.0035

[0092] In the case of mismatch, Cor(K q ,K r )=0.9628

[0093]

[0094] Matching case: Cor(K Q ,K R )=42816.12145079920

[0095]

[0096] An empirical threshold between 1.25 and 1.85 can be used to distinguish between matching and mismatching cases.

[0097] Multiple device situation:

[0098] In the case of multiple models, it is impossible to define an effective threshold to distinguish between matches and mismatches based on the calculation results of traditional similarity. The example uses multiple devices to take photos, including three mobile phones from Samsung, VIVO and Motorola, and two digital cameras from Fuji and Sony (including models affected by software noise and models not affected by software noise). For each model, the similarity between matches and mismatches between devices is calculated, and then the mismatch similarity values ​​of all models are classified into H0, and the matching similarity values ​​are classified into H1, and the distribution of H0 and H1 is plotted.

[0099] like Figure 2 As shown in the figure, the dot-filled bar graph is the similarity distribution in the case of mismatch, and the oblique-filled bar graph is the similarity distribution in the case of match. It can be seen that in the traditional similarity result distribution, the two types of bar graphs blend with each other, and it is impossible to set a suitable threshold to distinguish between matching and mismatching situations in the whole domain.

[0100] Specifically, the presence of software noise will increase the similarity values ​​in matching and mismatching situations, while the software noise intensity of some models is higher than the camera fingerprint intensity, resulting in excessively high similarity in mismatching situations, even exceeding the similarity in matching situations between devices not affected by software noise, resulting in an interlaced distribution of matching and mismatching in a device verification system with a unified threshold.

[0101] like Figure 3 As shown, in the results obtained by the algorithm of the present invention, the mismatching situations of each model form a distribution with a mean of 1 (H0 distribution), and when combined together, it is also a distribution with a mean of 1. Therefore, the H0 distribution is no longer affected by the strength of the software noise similarity and changes dramatically. The matching situation forms a distribution with a mean of 2 (H1 distribution), which is basically distinguished from the mismatching situation. The example results show that the equal error rate (EER) drops from 43.37% to 5.85%. The present invention makes it possible to define a unified threshold for the complex situation of multiple models that are affected by software noise and those that are not affected by software noise.

[0102] Corresponding to the aforementioned embodiment of the method for matching camera fingerprints affected by software noise, the present application also provides an embodiment of a device for matching camera fingerprints affected by software noise.

[0103] Figure 4 FIG. 1 is a block diagram of a camera fingerprint matching device affected by software noise according to an exemplary embodiment. Figure 4 , the device may include:

[0104] The acquisition module 21 is used to acquire the image I to be tested. q Match the camera fingerprint K with the target r ;

[0105] The extraction module 22 is used to extract the image I q , extract the corresponding camera fingerprint K to be tested q ;

[0106] The first calculation module 23 is used to calculate the target matching camera fingerprint K according to the selected basic similarity calculation method. r And the camera fingerprint K to be tested q Camera fingerprint similarity Cor_pair;

[0107] The second computing module 24 is used to match the camera fingerprint K with the target r The corresponding device model and shooting software are the same as the device Z = {Z1,…,Z n}Several photos taken, extract the corresponding camera fingerprint Calculate the image K to be tested q With camera fingerprint K z Camera fingerprint similarity Cor z Take the average to get the estimated software noise similarity Cor_software_noise;

[0108] The third calculation module 25 is used to calculate the ratio Ratio of Cor_pair and Cor_software_noise;

[0109] The comparison module 26 is used to compare the preset threshold Threshold with Ratio. If Ratio is greater than Threshold, it means that there is a hardware fingerprint similarity component in Ratio, and the camera fingerprint K q With K r Match; otherwise, it means that there is no hardware fingerprint similarity component in Ratio, and the camera fingerprint K q With K r No match.

[0110] Regarding the device in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.

[0111] For the device embodiment, since it basically corresponds to the method embodiment, the relevant parts can refer to the partial description of the method embodiment. The device embodiment described above is only schematic, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the present application scheme. A person of ordinary skill in the art can understand and implement it without paying any creative work.

[0112] Accordingly, the present application also provides an electronic device, comprising: one or more processors; a memory for storing one or more programs; when the one or more programs are executed by the one or more processors, the one or more processors implement the above-mentioned camera fingerprint matching method affected by software noise. Figure 5 As shown, a hardware structure diagram of a camera fingerprint matching method affected by software noise provided by an embodiment of the present invention is provided for any device with data processing capability, except Figure 5 In addition to the processor, memory and network interface shown, any device with data processing capability in which the apparatus in the embodiment is located may also include other hardware, generally based on the actual functions of the device with data processing capability, which will not be described in detail.

[0113] Accordingly, the present application also provides a computer-readable storage medium on which computer instructions are stored, and when the instructions are executed by the processor, the matching method of the camera fingerprint affected by software noise as described above is implemented. The computer-readable storage medium can be an internal storage unit of any device with data processing capabilities described in any of the aforementioned embodiments, such as a hard disk or a memory. The computer-readable storage medium can also be an external storage device, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), an SD card, a flash card (Flash Card), etc. equipped on the device. Furthermore, the computer-readable storage medium can also include both an internal storage unit of any device with data processing capabilities and an external storage device. The computer-readable storage medium is used to store the computer program and other programs and data required by any device with data processing capabilities, and can also be used to temporarily store data that has been output or is to be output.

[0114] Those skilled in the art will readily appreciate other embodiments of the present application after considering the specification and practicing the contents disclosed herein. The present application is intended to cover any variations, uses or adaptations of the present application, which follow the general principles of the present application and include common knowledge or customary technical means in the art that are not disclosed in the present application.

[0115] It should be understood that the present application is not limited to the exact construction that has been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof.

Claims

1. A camera fingerprint matching method affected by software noise, characterized in that: include: Get the image to be tested I q And the target matching camera fingerprint k r , the target matches the camera fingerprint K r Photo I taken by device r r It is estimated that According to the image to be tested I q , extract the corresponding camera fingerprint K to be tested q And match the camera fingerprint K with the target r Calculate the basic similarity Cor_pair; Estimated target matching camera fingerprint K r The software noise similarity between the corresponding device model and the shooting software is Cor_software_noise; Calculate the ratio of Cor_pair to Cor_software_noise as the new camera fingerprint similarity, and compare the ratio with the preset threshold to obtain the matching result; The steps of estimating the software noise similarity are as follows: Use the camera fingerprint K that matches the target r The corresponding device model and shooting software are the same as the device Z = {Z1,…,Z n }Several photos taken, extract the corresponding camera fingerprint Calculate the fingerprint k of the camera to be tested q With camera fingerprint k z Basic similarity of camera fingerprint Cor z Take the average to get the estimated software noise similarity Cor_software_noise.

2. The method according to claim 1, characterized in that The ratio of Cor_pair to Cor_software_noise is calculated as: or Wherein, C is a preset constant.

3. The method according to claim 1, characterized in that Compare the ratio Ratio with the preset threshold to obtain the matching result. Specifically, compare the preset threshold Threshold with Ratio. If Ratio is greater than Threshold, then the camera fingerprint k q With K r Match; otherwise camera fingerprint K q With k r No match.

4. The method according to claim 1, characterized in that: Estimating camera fingerprint k q and K r Here are the steps: I will take the photo r Use the denoising algorithm to denoise and get the denoised image I will take the photo r And the denoised image Subtract and get the noise residual W contains the camera fingerprint K r and other noises.

5. The method according to claim 1, characterized in that The peak correlation energy ratio, the cross-correlation coefficient or the normalized cross-correlation coefficient is used as the basic similarity.

6. A matching device for camera fingerprints affected by software noise, characterized in that: include: Acquisition module, used to acquire the image to be tested I q Match the camera fingerprint K with the target r , the target matches the camera fingerprint K r Photo I taken by device r r It is estimated that Extraction module, used for extracting the image I q , extract the corresponding camera fingerprint K to be tested q ; The first calculation module is used to calculate the target matching camera fingerprint K according to the selected basic similarity calculation method. r And the camera fingerprint K to be tested q Camera fingerprint similarity Cor_pair; The second computing module is used to match the camera fingerprint K with the target r The corresponding device model and shooting software are the same as the device Z = {Z1,…,Z n }Several photos taken, extract the corresponding camera fingerprint Calculate the image K to be tested q With camera fingerprint K z Camera fingerprint similarity Cor z Take the average to get the estimated software noise similarity Cor_software_noise; The third calculation module is used to calculate the ratio Ratio of Cor_pair and Cor_software_noise; The comparison module is used to compare the preset threshold Threshold with Ratio. If Ratio is greater than Threshold, the camera fingerprint K q With K r Match; otherwise camera fingerprint K q With K r No match.

7. An electronic device, characterized in that: include: one or more processors; A memory for storing one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 5.

8. A computer-readable storage medium having computer instructions stored thereon, characterized in that: When the instruction is executed by a processor, the steps of the method according to any one of claims 1 to 5 are implemented.

Images