Information processing device, information processing method, and recording medium

Abstract

An information processing device 1 is provided with: a feature acquisition unit 11 that alternately acquires a portion of the features of a target image subject to determination about whether or not said image has been processed and a portion of the features of a registered image; a feature generation unit 12 that generates features of a paired image of the target image and the registered image, and generates features of a current paired image on the basis of the features of the previous paired image generated in a previous operation and the portions of the features; a calculation unit 13 that calculates, on the basis of the features of the current paired image, a score pertaining to whether or not the target image has been processed; a determination unit 14 that determines, on the basis of the score, whether or not the target image has been processed; a repetition unit 15 that repeatedly executes operations performed by the feature acquisition unit 11, the feature generation unit 12, the calculation unit 13, and the determination unit 14 until completion conditions for the operations are satisfied; and an output unit 16 that performs output according to the determination result by the determination unit 14.

Description

Information processing device, information processing method, and recording medium

[0001] The present disclosure relates to the technical fields of an information processing device, an information processing method, and a recording medium.

[0002] Non-Patent Document 1 discloses that a processed image of a target image is detected by calculating the difference between the facial matching feature extracted from a registered image and the facial matching feature extracted from the target image.

[0003] Deep Face Representation for Differential Morphing Attack Detection (Scherhag, 2020)

[0004] An object of the present disclosure is to provide an information processing device, an information processing method, and a recording medium that can accurately detect whether an image to be determined has been edited.

[0005] One aspect of the information processing device comprises a feature acquisition means that alternately acquires some features of a target image to be determined as to whether it has been edited and some features of a registered image; a feature generation means that generates features of a pair of images of the target image and the registered image, the feature generation means generating features of the current pair of images based on the features of the previous pair of images generated in the previous operation and the some features; a calculation means that calculates a score as to whether the target image has been edited based on the features of the current pair of images; a judgment means that judges whether the target image has been edited based on the score; a repetition means that repeatedly executes the operations of the feature acquisition means, the feature generation means, the calculation means, and the judgment means until a completion condition for the operation is met; and an output means that outputs according to the judgment result by the judgment means.

[0006] One aspect of the information processing method involves acquiring partial features of one of a pair of images, which is a target image to be determined as having been edited or not, and a registered image, generating features for the current pair of images based on the partial features and the features of the previous pair of images generated in the previous operation, calculating a score as to whether the target image has been edited or not based on the features of the current pair of images, determining whether the target image has been edited or not based on the score, and repeating the feature acquisition operation, feature generation operation, calculation operation, and determination operation until a completion condition for the operation is met, and in acquiring the partial features, alternately acquiring partial features of the target image and partial features of the registered image, and outputting according to the determination result.

[0007] One aspect of the recording medium is a computer program recorded on the recording medium to cause a computer to execute an information processing method, which involves acquiring partial features of one of a pair of images, a target image to be determined as to whether it has been edited and a registered image, generating features for the current pair of images based on the partial features and the features of the previous pair of images generated in the previous operation, calculating a score as to whether the target image has been edited based on the features of the current pair of images, determining whether the target image has been edited based on the score, repeating the feature acquisition operation, feature generation operation, calculation operation, and determination operation until a completion condition for the operation is met, and in acquiring the partial features, alternately acquiring partial features of the target image and partial features of the registered image, and outputting an output according to the determination result.

[0008] The information processing device, information processing method, and recording medium according to the present disclosure can accurately detect whether an image to be determined has been edited.

[0009] FIG. 1 is a block diagram showing an example of a configuration of an information processing device according to the present disclosure. FIG. 2 is a block diagram showing an example of a configuration of an information processing device according to the present disclosure. FIG. 3 is a flowchart showing an example of a flow of information processing operation of an information processing device according to the present disclosure. FIG. 4 is a block diagram showing an example of a configuration of an information processing device according to the present disclosure. FIG. 5 is a flowchart showing an example of a flow of information processing operation of an information processing device according to the present disclosure. FIG. 6 is a schematic diagram showing an example of a flow of information processing operation of an information processing device according to the present disclosure. FIG. 7 is a block diagram showing an example of a configuration of an information processing device according to the present disclosure.

[0010] Hereinafter, an information processing device, an information processing method, and a recording medium according to an embodiment will be described with reference to the drawings. [1: First Embodiment]

[0011] A first embodiment of an information processing device, an information processing method, and a recording medium will be described. Hereinafter, the first embodiment of the information processing device, the information processing method, and the recording medium will be described using an information processing device 1 according to the present disclosure. [1-1: Configuration of Information Processing Device 1]

[0012] 1 is a block diagram showing the configuration of an information processing device 1 according to the present disclosure. As shown in FIG. 1, the information processing device 1 includes a feature acquisition unit 11, a feature generation unit 12, a calculation unit 13, a determination unit 14, a repetition unit 15, and an output unit 16.

[0013] The feature acquisition unit 11 alternately acquires some features of a target image to be determined as to whether it has been edited and some features of a registered image. Whether an image has been edited can be rephrased as whether the image contains a morphing artifact. The registered image is an image that has been registered and has not been edited.

[0014] The feature generation unit 12 generates features for the paired images of the target image and the registered image. The feature generation unit 12 generates the features for the current paired images based on the features for the paired images generated in the previous operation and some of the acquired features.

[0015] The calculation unit 13 calculates a score regarding whether or not the target image has been edited based on the features of the current paired image. The determination unit 14 determines whether or not the target image has been edited based on the score.

[0016] The repeating unit 15 repeatedly executes the operations of the feature acquiring unit 11, the feature generating unit 12, the calculating unit 13, and the determining unit 14 until the completion condition of the operations is satisfied. The output unit 16 outputs according to the determination result of the determining unit 14. [1-2: Technical Effects of the Information Processing Device 1]

[0017] The information processing device 1 according to the present disclosure determines whether or not a target image has been edited based on some of the acquired features, and therefore can determine whether or not the target image has been edited even if the target image has been partially edited. In other words, the information processing device 1 can accurately detect whether or not the image to be determined has been edited. [2: Second Embodiment]

[0018] Next, a second embodiment of an information processing device, an information processing method, and a recording medium will be described. Hereinafter, the second embodiment of the information processing device, the information processing method, and the recording medium will be described using an information processing device 2 according to the present disclosure. [2-1: Configuration of the information processing device 2]

[0019] 2 is a block diagram showing the configuration of an information processing device 2 according to the present disclosure. As shown in FIG. 2, the information processing device 2 includes a calculation device 21 and a storage device 22. The information processing device 2 may further include a communication device 23, an input device 24, and an output device 25. However, the information processing device 2 does not necessarily have to include at least one of the communication device 23, the input device 24, and the output device 25. The calculation device 21, the storage device 22, the communication device 23, the input device 24, and the output device 25 may be connected via a data bus 26.

[0020] The arithmetic device 21 includes, for example, at least one of a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), and an FPGA (Field Programmable Gate Array). The arithmetic device 21 reads a computer program. For example, the arithmetic device 21 may read a computer program stored in the storage device 22. For example, the arithmetic device 21 may read a computer program stored in a computer-readable, non-transitory recording medium using a recording medium reading device (e.g., an input device 24 described later) not shown in the drawings that is included in the information processing device 2. The arithmetic device 21 may acquire (i.e., download or read) the computer program from a device (not shown) located outside the information processing device 2 via the communication device 23 (or another communication device). The arithmetic device 21 executes the read computer program. As a result, logical functional blocks for executing the operations to be performed by the information processing device 2 are realized within the arithmetic device 21. In other words, the arithmetic device 21 can function as a controller for realizing logical functional blocks for executing the operations (in other words, processing) to be performed by the information processing device 2.

[0021] FIG. 2 shows an example of logical functional blocks implemented within the arithmetic device 21 to perform information processing operations. As shown in FIG. 3, the arithmetic device 21 includes a feature acquisition unit 211, which is a specific example of a "feature acquisition means" described in the appendix, a feature generation unit 212, which is a specific example of a "feature generation means" described in the appendix, a calculation unit 213, which is a specific example of a "calculation means" described in the appendix, a determination unit 214, which is a specific example of a "determination means" described in the appendix, a repetition unit 215, which is a specific example of a "repetition means" described in the appendix, an output unit 216, which is a specific example of an "output means" described in the appendix, and a paired image acquisition unit 217. However, the paired image acquisition unit 217 does not necessarily have to be implemented within the arithmetic device 21. The feature acquisition unit 211 may include a portion selection unit 2111 and an extraction unit 2112. The operations of the feature acquisition unit 211, feature generation unit 212, calculation unit 213, determination unit 214, repetition unit 215, output unit 216, and image pair acquisition unit 217 will be described in detail later with reference to FIG.

[0022] The storage device 22 can store desired data. For example, the storage device 22 may temporarily store a computer program executed by the arithmetic device 21. The storage device 22 may temporarily store data that the arithmetic device 21 temporarily uses when the arithmetic device 21 is executing a computer program. The storage device 22 may store data that the information processing device 2 stores long-term. The storage device 22 may include at least one of a RAM (Random Access Memory), a ROM (Read Only Memory), a hard disk device, a magneto-optical disk device, an SSD (Solid State Drive), and a disk array device. In other words, the storage device 22 may include a non-temporary recording medium.

[0023] The communication device 23 is capable of communicating with devices external to the information processing device 2 via a communication network (not shown). The communication device 23 may be a communication interface based on standards such as Ethernet (registered trademark), Wi-Fi (registered trademark), Bluetooth (registered trademark), or USB (Universal Serial Bus).

[0024] The input device 24 is a device that accepts information input to the information processing device 2 from outside the information processing device 2. For example, the input device 24 may include an operation device (e.g., at least one of a keyboard, a mouse, and a touch panel) that can be operated by an operator of the information processing device 2. For example, the input device 24 may include a reading device that can read information recorded as data on a recording medium that can be externally attached to the information processing device 2.

[0025] The output device 25 is a device that outputs information to the outside of the information processing device 2. For example, the output device 25 may output information as an image. That is, the output device 25 may include a display device (a so-called display) that can display an image showing the information to be output. For example, the output device 25 may output information as sound. That is, the output device 25 may include an audio device (a so-called speaker) that can output sound. For example, the output device 25 may output information on paper. That is, the output device 25 may include a printing device (a so-called printer) that can print desired information on paper. [2-2: Information Processing Operation Performed by the Information Processing Device 2]

[0026] The flow of information processing operations performed by the information processing device 2 will be described with reference to Fig. 3. Fig. 3 is a flowchart showing an example of the flow of information processing operations performed by the information processing device 2.

[0027] 3, the paired image acquisition unit 217 acquires a paired image of a target image and a registered image (step S20). The paired image acquisition unit 217 may acquire the paired image via the communication device 23 or the input device 24. The paired image acquisition unit 217 may acquire the paired image stored in the storage device 22. The paired image acquisition unit 217 may acquire the target image acquired via the communication device 23 or the input device 24 and the registered image stored in the storage device 22 as the paired image.

[0028] The part selection unit 2111 selects a partial area of ​​the target image or the registered image based on the features of the paired image (step S21). Hereinafter, the partial area may be referred to as a "patch." The part selection unit 2111 may output position information of the patch. The position information of the patch may include information indicating the coordinates, size, etc. of the part in the target image or the registered image. The part selection unit 2111 may select the patch using a selection model. The selection model is a model that outputs information indicating the patch when the features of the paired image are input. The selection model may be a model that includes any neural network in its architecture.

[0029] The extraction unit 2112 extracts patch features from the selected patches (step S22). Hereinafter, the "patch features" may be referred to as "patch features." The extraction unit 2112 may extract patch features using an extraction model. The extraction model is a model that outputs patch features when a patch is input. The extraction model may also be said to convert the patch into features. The extraction model may be, for example, a model that includes a convolutional neural network (CNN) in its architecture. The feature acquisition unit 211 alternately acquires patch features of the target image and patch features of the registered images. The feature acquisition unit 211 may also be said to sequentially acquire patch features. Each of the sequentially acquired patch features may be understood as an element of sequence data.

[0030] The feature generation unit 212 generates integrated features as features of the pair of images between the target image and the registered image (step S23). The feature generation unit 212 generates the current integrated features based on the integrated features generated in the previous operation and the patch features acquired this time. The feature generation unit 212 may generate the integrated features using a model that handles sequential data. For example, the feature generation unit 212 may generate the integrated features using a model whose architecture includes a recurrent neural network (RNN) that employs long short-term memory (LSTM). Alternatively, the feature generation unit 212 may generate the integrated features using a model whose architecture includes a simple RNN. The model used by the feature generation unit 212 may be a temporal integrator. The temporal integrator receives patch features and outputs integrated features. In other words, the temporal integrator updates the integrated feature when a patch feature is input. The temporal integrator may integrate the currently acquired patch feature with the previous integrated feature and output the current integrated feature.

[0031] Alternatively, the feature generation unit 212 may generate an integrated feature by integrating patch features acquired sequentially. The integrated feature may be understood as a feature of series data. Furthermore, in step S21, the portion selection unit 2111 may select a partial region of the target image or the registered image based on the integrated feature generated in the previous operation. That is, the portion selection unit 2111 may recognize a portion to which attention should be paid in the current repetition based on the integrated feature based on each of the patch features handled by the information processing device 3 up to the previous operation.

[0032] The calculation unit 213 calculates a score indicating whether the target image has been edited or not based on the current integrated feature (step S24). For example, the score may be an index indicating whether the target image is appropriate to belong to an edited class indicating that the target image has been edited, or an unedited class indicating that the target image has not been edited. Alternatively, the score may be an index indicating whether the paired images are appropriate to belong to an edited class indicating that the paired images include an edited image, or an unedited class indicating that the paired images do not include an edited image. The calculation unit 213 may calculate the score using a score calculator. The score calculator may calculate the score using any method. The determination unit 214 determines whether the target image has been edited or not based on the score (step S25).

[0033] The repeating unit 215 determines whether the completion condition of the operation is satisfied (step S26). The case where the completion condition of the operation is satisfied may be, for example, a case where the determination by the determination unit 214 is finalized. The case where the determination by the determination unit 214 is finalized may be, for example, a case where the number of determinations by the determination unit 214 reaches a predetermined number. Alternatively, for example, the case where the time during which the information processing device 2 has been performing the information processing operation reaches a predetermined time.

[0034] If the operation completion condition is not satisfied (step S26: No), the process returns to step S21. That is, the operations from step S21 to step S25 are executed. In other words, the repetition unit 215 repeatedly executes the operations of the feature acquisition unit 211, the feature generation unit 212, the calculation unit 213, and the determination unit 214 until the operation completion condition is satisfied. In other words, the repetition unit 215 executes sequential operations until the operation completion condition is satisfied. The sequential operations include a patch feature acquisition operation (a patch selection operation and a patch feature extraction operation), an integrated feature generation operation, a score calculation operation, and a score determination operation.

[0035] If the operation completion condition is satisfied (step S26: Yes), the output unit 216 outputs a result according to the determination result by the determination unit 214 (step S27). The output unit 216 may, for example, output a confirmed determination result. The output unit 216 may, for example, display the determination result on a display or the like. Furthermore, the output unit 216 may, for example, output the determination result as sound via a speaker or the like.

[0036] Each of the selection model, extraction model, and temporal integrator described above may be a trained model. Each of the trained models may be a model that has been machine-learned using training data prepared in advance.

[0037] Furthermore, when comparing two images, a technique of cutting out a part of the image, converting it into a feature, and repeatedly comparing the feature of the part of the image is also called attentive recurrent comparators (ARC) (see References 1 and 2 below).

[0038] [Reference 1] P. Shyam, S. Gupta, and A. Dukkipati, “Attentive recurrent comparators,” in arXiv:1703.00767, 2015. [Reference 2] [[1804.11027] Deep Co-attention based Comparators For Relative Representation Learning in Person Re-identification (arxiv.org)] [2-3: Technical effects of information processing device 2]

[0039] Non-Patent Document 1, cited above, describes a technology for detecting whether an image to be subjected to facial recognition has been edited by comparing the entirety of a registration image registered by a user in advance with the entirety of an image to be subjected to facial recognition. The technology described in Non-Patent Document 1 is referred to as a comparative example. The comparative example detects whether an image has been edited by calculating the difference between authentication features extracted from the registration image and face matching features extracted from the image to be subjected to facial recognition. That is, the comparative example performs detection using features extracted from the entirety of the registration image and features extracted from the entirety of the image to be subjected to facial recognition. However, if only a portion of the image has been edited, the features extracted from the entirety of the registration image and the features extracted from the entirety of the image to be subjected to facial recognition may be similar, making it difficult to distinguish whether the image has been edited.

[0040] In the information processing device 2 according to the present disclosure, the part selection unit 2111 selects a patch, extracts patch features, and determines whether the target image has been edited based on the patch features. Therefore, even if the target image has been partially edited, it is possible to determine whether the target image has been edited. In other words, even if only a portion of the image has been edited, the information processing device 2 acquires some features, so it is possible to detect whether the image has been edited. The information processing device 2 can accurately detect whether there are edited portions, such as morphing artifacts. Furthermore, the information processing device 2 performs multiple repeated operations, so it can observe multiple aspects of both images, and therefore accurately detect whether the images have been edited. [3: Third Embodiment]

[0041] Next, a third embodiment of an information processing device, an information processing method, and a recording medium will be described. Hereinafter, the third embodiment of an information processing device, an information processing method, and a recording medium will be described using an information processing device 3 according to the present disclosure. In the third embodiment, the operation of the calculation unit 313 and the operation of the determination unit 314 differ from those in the second embodiment. [3-1: Information processing operation performed by the information processing device 3]

[0042] 5 and 6, the flow of the information processing operation performed by the information processing device 3 will be described. FIG. 5 is a flowchart showing the flow of the information processing operation performed by the information processing device 3.

[0043] As shown in FIG. 5 , the paired image acquisition unit 217 acquires paired images of a target image and a registered image (step S20). The part selection unit 3111 selects patches of the target image or the registered image based on the previous integrated features (step S21). The extraction unit 3112 extracts patch features from the selected patches (step S22). The feature generation unit 312 generates integrated features (step S23). The feature generation unit 312 generates the current integrated features based on the integrated features generated in the previous operation and the patch features.

[0044] The calculation unit 313 calculates a score indicating the likelihood that the target image has been edited and the likelihood that the target image has not been edited, based on the current integrated feature (step S30). For example, the score may include a ratio (likelihood ratio) between a probability indicating the likelihood that the target image belongs to an edited class indicating that the target image has been edited and a probability indicating the likelihood that the target image belongs to an unedited class indicating that the target image has not been edited. Alternatively, the score may include a ratio (likelihood ratio) between a probability indicating the likelihood that the paired image belongs to an edited class indicating that the paired image includes an edited image and a probability indicating the likelihood that the paired image belongs to an unedited class indicating that the paired image does not include an edited image.

[0045] The determination unit 314 compares the score with a predetermined threshold (step S31). The determination unit 314 determines whether the score reaches the predetermined threshold (step S32). The determination unit 314 may perform an operation using a sequential probability ratio test (SPRT). The determination unit 314 may perform an operation using a probability ratio tester that performs the SPRT.

[0046] The predetermined threshold may include, for example, a threshold for an edited class and a threshold for an unedited class. For example, if the score increases when the target image is edited and decreases when the target image is edited, the threshold for the edited class may be set to a value close to the upper limit of the score, and the threshold for the unedited class may be set to a value close to the lower limit of the score. In this case, when the score reaches the predetermined threshold, this includes when the score exceeds the threshold for the edited class and when the score falls below the threshold for the unedited class. The predetermined threshold may be a value that can be set depending on the application.

[0047] If the score reaches a predetermined threshold (step S32: Yes), the determination unit 314 determines whether the target image has been edited (step S33). In the third embodiment, the completion condition of the operation is satisfied when the determination unit 314 determines whether the target image has been edited. That is, in the third embodiment, the case where the completion condition of the operation is satisfied may include the case where the score reaches the threshold.

[0048] In the third embodiment, the score may be referred to as reliability. When referred to as reliability, it can be understood that the determination unit 314 makes a determination by successively updating the reliability. It may also be said that the determination unit 314 finalizes the determination when sufficient accuracy is obtained.

[0049] When the score reaches a predetermined threshold, the determination unit 314 may classify at least one of the target image and the paired image into either an edited class or an unedited class. The determination unit 314 may make a determination regarding at least one of the target image and the paired image by class classification.

[0050] The output unit 216 outputs a result according to the determination result by the determination unit 314 (step S27). The output unit 216 may output, for example, a confirmed determination result.

[0051] If the score does not reach the predetermined threshold (step S32: No), the determination unit 314 does not classify the paired image into either the edited class or the unedited class. Returning to step S21, the feature acquisition unit 311 further acquires patch features.

[0052] Fig. 6 is a schematic diagram showing an example of the flow of information processing operations performed by the information processing device 3. Fig. 6 illustrates a case where the determination is finalized in the t+1th determination.

[0053] The score calculator receives the integrated feature (t-1) generated in the t-1th action and outputs a score (step S1). The probability ratio tester receives the score and compares it with a predetermined threshold (step S2). The probability ratio tester determines whether to continue the action and determines that it should be continued at the relevant step (step S3). If the probability ratio tester determines that it should not be continued, it is determined whether the action has been altered.

[0054] The selection model receives the integrated feature (t-1) generated in the t-1th operation and outputs the patch position information (step S4). The extraction model receives the target image patch corresponding to the position information and outputs the patch feature (step S5).

[0055] The temporal integrator receives the integrated feature (t-1) generated in the t-1th operation and the patch feature, and outputs the integrated feature (t) (step S6). The integrated feature (t) is the integrated feature generated in the tth operation.

[0056] The score calculator receives the integrated feature (t) generated in the tth action and outputs a score (step S7). The probability ratio tester receives the score and compares it with a predetermined threshold (step S8). The probability ratio tester determines whether to continue the action and determines that it should be continued at the relevant step (step S9).

[0057] The selection model receives the integrated feature (t) generated in the tth operation and outputs the patch position information (step S10). The extraction model receives the patch of the registered image corresponding to the position information and outputs the patch feature (step S11).

[0058] The temporal integrator receives the integrated feature (t) generated in the t-th operation and the patch feature, and outputs the integrated feature (t+1) (step S12). The integrated feature (t+1) is the integrated feature generated in the t+1-th operation.

[0059] The score calculator receives the integrated feature (t+1) generated in the t+1th action and outputs a score (step S13). The probability ratio tester receives the score and compares it with a predetermined threshold (step S14). The probability ratio tester determines whether to continue the action, determines not to continue at the relevant step, and determines whether the action has been modified (step S15).

[0060] The integrated feature (t+1) generated in the t+1th operation may be input, and the selection model and extraction model may operate in parallel with steps S13 to S15. If there is no input indicating continuation to the temporal integrator, the information processing device 3 completes the operation. [3-2: Modification]

[0061] The feature acquiring unit 311 may acquire patch features from sources other than the paired images. For example, the feature acquiring unit 311 may acquire patch features from information related to the paired images (related information).

[0062] In this case, the part selection unit 3111 selects a part of one or the other of the pair of related information based on the previous integrated feature. The extraction unit 3112 extracts patch features of the selected part. The feature acquisition unit 311 alternately acquires patch features of one of the pair of related information and patch features of the other of the pair of related information.

[0063] The feature generating unit 312 may generate an integrated feature based on a patch feature acquired from a portion of the related information. The related information may be a pair of information generated based on a pair of images. Each piece of the related information may have a feature of the target image and a feature of the registered image. In this case, the integrated feature generated from each of the patch features acquired from one or the other of the pair of related information includes a partial feature of the target image and a partial feature of the registered image, so that it is possible to obtain an effect equivalent to that obtained when patch features are acquired from a pair of images. [3-3: Technical Effects of the Information Processing Device 3]

[0064] In a method that repeatedly judges by acquiring and comparing features from parts of an image, the more comparisons are made, the more accurate the judgment tends to be. However, since the number of comparisons must be determined in advance, reducing the number of comparisons to reduce the amount of calculations, or increasing the number of comparisons to improve calculation accuracy, requires retraining the model.

[0065] The information processing device 3 according to the present disclosure sequentially tests (SPRT) scores, and can immediately complete the comparison process when the score reaches a threshold. Since the information processing device 3 SPRTs scores, the number of times features are acquired and compared can be changed by changing the SPRT threshold. For example, increasing the SPRT threshold increases the amount of calculation but can improve accuracy. On the other hand, decreasing the SPRT threshold reduces the amount of calculation but can lower accuracy. In other words, since the information processing device 3 SPRTs scores, it is possible to adjust the trade-off between the amount of calculation and accuracy without re-learning.

[0066] The information processing device 3 can make a determination without wasting unnecessary time. Furthermore, the determination by the information processing device 3 can ensure sufficient accuracy. In other words, the information processing device 3 can increase both the speed and accuracy of determining the features of a pair of images including patch features that are sequentially acquired. In other words, the information processing device 3 can optimize the determination speed while ensuring the determination accuracy. [4: Fourth Embodiment]

[0067] Next, a fourth embodiment of an information processing device, an information processing method, and a recording medium will be described. Hereinafter, the fourth embodiment of an information processing device, an information processing method, and a recording medium will be described using an information processing device 4 according to the present disclosure.

[0068] 7 , the information processing device 4 differs from the information processing device 2 and the information processing device 3 in that a threshold setting unit 418 is further implemented in the calculation device 21. Other features of the information processing device 4 may be the same as other features of at least one of the information processing device 2 and the information processing device 3.

[0069] The threshold setting unit 418 sets the SPRT threshold. For example, the threshold setting unit 418 may automatically set the SPRT threshold depending on the operating conditions. Alternatively, the threshold setting unit 418 may set the SPRT threshold by accepting input from a user. For example, the user may be able to specify the threshold when inputting an image to be determined. [Technical Effects of the Information Processing Device 4]

[0070] The information processing device 4 according to the present disclosure can set the SPRT threshold, and therefore can easily change the threshold to a desired value. The information processing device 4 can easily adjust the trade-off between the amount of calculation and accuracy in determining paired images. [5: Supplementary Note]

[0071] The following supplementary notes are further disclosed with respect to the above-described embodiment. [Supplementary Note 1] An information processing apparatus comprising: feature acquisition means for alternately acquiring partial features of a target image to be determined as to whether it has been edited and partial features of a registered image; feature generation means for generating features of a pair of images of the target image and the registered image, the feature generation means generating features of the current pair of images based on the partial features and the previous feature of the pair of images generated in a previous operation; calculation means for calculating a score as to whether the target image has been edited based on the features of the current pair of images; determination means for determining whether the target image has been edited based on the score; repetition means for repeatedly executing the operations of the feature acquisition means, the feature generation means, the calculation means, and the determination means until an operation completion condition is satisfied; and output means for outputting an output according to the determination result by the determination means. [Supplementary Note 2] The information processing apparatus according to Supplementary Note 1, wherein the feature acquisition means includes: part selection means for selecting the part of the target image or the registered image based on the feature of the previous pair of images; and extraction means for extracting the partial features. [Supplementary Note 3] The information processing device according to Supplementary Note 1 or 2, wherein the calculation means calculates the score indicating the likelihood that the target image has been edited and the likelihood that the target image has not been edited, and the determination means makes a determination based on a comparison of the score with a threshold. [Supplementary Note 4] The determination means determines whether or not the target image has been edited when the score reaches a threshold, and the operation completion condition is satisfied when the determination means determines whether or not the target image has been edited. [Supplementary Note 5] The information processing device according to Supplementary Note 2, wherein the part selection means selects a part of one or the other of a pair of related information related to the pair of images based on features of the previous pair of images, and the feature acquisition means alternately acquires the part of features of one of the pair of related information and the part of features of the other of the pair of related information. [Supplementary Note 6] The information processing device according to Supplementary Note 3, further comprising a setting means for setting the threshold.[Supplementary Note 7] An information processing method comprising: acquiring partial features of one of a pair of images, which is a target image to be determined as to whether it has been edited or not, and a registered image; generating features of the current pair of images based on the partial features and the previous features of the pair of images generated in the previous operation; calculating a score as to whether the target image has been edited or not based on the features of the current pair of images; determining whether the target image has been edited or not based on the score; repeating the feature acquisition operation, feature generation operation, calculation operation, and determination operation until an operation completion condition is met; in acquiring the partial features, alternately acquiring partial features of the target image and partial features of the registered image; and outputting according to the determination result. [Supplementary Note 8] A recording medium having recorded thereon a computer program for causing a computer to execute an information processing method comprising: acquiring partial features of one of a pair of images, which is a target image to be determined as to whether it has been edited or not, and a registered image; generating features of the current pair of images based on the partial features and the previous features of the pair of images generated in the previous operation; calculating a score as to whether the target image has been edited or not based on the features of the current pair of images; determining whether the target image has been edited or not based on the score; repeating the feature acquisition operation, the feature generation operation, the calculation operation, and the determination operation until an operation completion condition is satisfied; in acquiring the partial features, alternately acquiring partial features of the target image and partial features of the registered image; and outputting according to the determination result.

[0072] This disclosure may be modified as appropriate within the scope of the claims and the technical idea that can be read from the entire specification. Information processing devices, information processing methods, and recording media that involve such modifications are also included in the technical idea of ​​this disclosure.

[0073] 1, 2, 3, 4 Information processing device 11, 211, 311 Feature acquisition unit 12, 212, 312 Feature generation unit 13, 213, 313 Calculation unit 14, 214, 314 Determination unit 15, 215 Repetition unit 16, 216 Output unit 2111, 3111 Part selection unit 2112, 3112 Extraction unit 217 Image pair acquisition unit 418 Threshold setting unit

Claims

1. An information processing device comprising: feature acquisition means for alternately acquiring partial features of a target image to be judged as to whether it has been edited and partial features of a registered image; feature generation means for generating features of a paired image of the target image and the registered image, the feature generation means generating features of the current paired image based on the previous feature of the paired image generated in a previous operation and the partial features; calculation means for calculating a score as to whether the target image has been edited or not based on the features of the current paired image; judgment means for judging whether the target image has been edited or not based on the score; repetition means for repeatedly executing operations by the feature acquisition means, the feature generation means, the calculation means and the judgment means until a completion condition for the operation is satisfied; and output means for providing an output according to the judgment result by the judgment means.

2. The information processing device according to claim 1, wherein the feature acquisition means includes: a portion selection means for selecting the portion of the target image or the registered image based on the features of the previous paired image; and an extraction means for extracting the features of the portion.

3. An information processing device as described in claim 1 or 2, wherein the calculation means calculates the score indicating the likelihood that the target image has been edited and the likelihood that the target image has not been edited, and the judgment means makes a judgment based on a comparison of the score with a threshold value.

4. The information processing device according to claim 3, wherein the determination means determines whether or not the target image has been edited when the score reaches a threshold value, and the completion condition for the operation is satisfied when the determination means determines whether or not the target image has been edited.

5. The information processing device according to claim 2, wherein the portion selection means selects a portion of one or the other of a pair of related information related to the pair of images based on the features of the previous pair of images, and the feature acquisition means alternately acquires the portion of the features of one of the pair of related information and the portion of the features of the other of the pair of related information.

6. The information processing device according to claim 3, further comprising a setting means for setting the threshold value.

7. An information processing method comprising the steps of: acquiring partial features of one of a pair of images, which is a target image to be determined as having been edited or not, and a registered image; generating features of the current pair of images based on the partial features and the features of the previous pair of images generated in a previous operation; calculating a score as to whether the target image has been edited or not based on the features of the current pair of images; determining whether the target image has been edited or not based on the score; repeating the feature acquisition operation, feature generation operation, calculation operation, and determination operation until a completion condition for the operation is satisfied; and in acquiring the partial features, alternately acquiring partial features of the target image and partial features of the registered image; and outputting according to the determination result.

8. A recording medium having recorded thereon a computer program for causing a computer to execute an information processing method comprising the steps of: acquiring partial features of one of a pair of images, which is a target image to be determined as having been edited or not, and a registered image; generating features of the current pair of images based on the partial features and the previous features of the pair of images generated in the previous operation; calculating a score as to whether the target image has been edited or not based on the features of the current pair of images; determining whether the target image has been edited or not based on the score; repeating the feature acquisition operation, feature generation operation, calculation operation, and determination operation until an operation completion condition is satisfied; and, in acquiring the partial features, alternately acquiring partial features of the target image and partial features of the registered image; and outputting according to the determination result.

Images