Image processing method, device, equipment and storage medium

Abstract

The present invention provides a method, device, equipment and storage medium for image processing, the method comprising: obtaining an image to be processed sent by a user; determining the scene to which the image to be processed belongs according to the environment in which the image to be processed is generated, and extracting feature information of the image to be processed by using a feature extraction method corresponding to the scene; establishing an image library for each scene, performing a numbering and labeling operation on new images entering the image library, and after the numbering and labeling operation is completed for the new images entering the image library, similarity calculation is performed between the new images and the old images previously stored in the image library, and the value of each identifier of the new image is adjusted based on the similarity calculation result; after the value of each identifier of the new image is determined, the new image is reviewed according to the value of each identifier of the new image to obtain a processing result; the method of the present invention can improve the recognition ability of risk images in different scenes and improve the recognition efficiency, and reduce the cost of manual review.

Description

Technical Field

[0001] The present invention relates to the technical field of image risk mining, and in particular to an image processing method, device, equipment and storage medium. Background Art

[0002] With the development of technology, people are no longer limited to sending messages through text. Since images are different from text and are unstructured data, it is difficult to conduct risk mining. Manual review is often required to determine whether they are compliant. On social platforms, a large number of images are generated every day. The illegal content of some images is highly concealed, which is difficult to detect with commonly used risk identification models and easy to be missed by manual review. The existing image review method is to create a list of unqualified images. By extracting image features, adding them to the unqualified list, and calculating similarity, this method has limitations and the following problems: (1) It is necessary to know that a certain image is on the unqualified list. User images are rich and diverse, and the repetition rate is not high. This method is limited in effect. (2) It relies too much on image similarity. If the user changes the background or crops the image, it is easy to be bypassed. (3) Risk mining is impossible. Therefore, a set of effective strategies is needed to conduct risk mining and rapid review of images. Summary of the invention

[0003] The object of the present invention is to provide a picture processing method, apparatus, device and storage medium to improve the above-mentioned problem.

[0004] In order to achieve the above objectives, the present application provides the following technical solutions:

[0005] On the one hand, an embodiment of the present application provides a method for processing an image, the method comprising:

[0006] Get the pictures to be processed sent by the user;

[0007] Determine the scene to which the picture to be processed belongs according to the environment in which the picture to be processed is generated, and extract feature information of the picture to be processed using a feature extraction method corresponding to the scene;

[0008] A picture library is established for each scenario. In each scenario, each time a user sends a picture, the picture will be stored in the picture library. New pictures entering the picture library are numbered and labeled: the new pictures are numbered, and N labels are assigned, and the number is used as the value of each label. After the new pictures entering the picture library complete the numbering and labeling operation, the new pictures are similar to the old pictures previously stored in the picture library, and the value of each label of the new picture is adjusted based on the similarity calculation result, where N is a positive integer greater than zero. After the value of each label of the new picture is determined, the new picture is reviewed according to the value of each label of the new picture. If there is a label and the corresponding value that has been identified as automatically passed or automatically rejected, the new picture is processed as automatically passed or automatically rejected. Otherwise, the new picture is manually reviewed to obtain the processing result.

[0009] The numbering operation for new pictures entering the picture library includes:

[0010] Numbering new pictures entering the picture library, assigning four identifiers to the new pictures after numbering, namely, a first identifier, a second identifier, a third identifier, and a fourth identifier, and using the number of the new picture as the value of the first identifier, the second identifier, the third identifier, and the fourth identifier;

[0011] The similarity between the new image and the old image previously stored in the image library is calculated, and the value of each identifier of the new image is adjusted based on the similarity calculation result, including:

[0012] Based on the feature information, the similarity between the new image currently entering the image library and the old image that already existed in the image library is calculated. After the calculation, the old image corresponding to the maximum similarity value is screened out, and the similarity range where the maximum similarity value is located is analyzed; wherein, if the maximum similarity value is greater than the first percentage, the value of the first identifier of the old image is used as the value of the first identifier of the new image; if the maximum similarity value is greater than the second percentage and is less than or equal to the first percentage, the value of the second identifier of the old image is used as the value of the second identifier of the new image; if the maximum similarity value is greater than the third percentage and is less than or equal to the second percentage, the value of the third identifier of the old image is used as the value of the third identifier of the new image; if the maximum similarity value is greater than the fourth percentage and is less than or equal to the third percentage, the value of the fourth identifier of the old image is used as the value of the fourth identifier of the new image, and the final value of each identifier of the new image is obtained, and the first percentage, the second percentage, the third percentage, and the fourth percentage decrease in sequence.

[0013] In a second aspect, an embodiment of the present application provides a picture processing device, the device comprising:

[0014] The acquisition module is used to obtain the pictures to be processed sent by the user;

[0015] An extraction module, used to determine the scene to which the image to be processed belongs according to the environment in which the image to be processed is generated or the information contained therein, and to extract feature information of the image to be processed using a feature extraction method corresponding to the scene;

[0016] The processing module is used to establish a picture library for each scenario. In each scenario, each time a user sends a picture, the picture will be stored in the picture library; the new pictures entering the picture library are numbered and labeled: the new pictures are numbered, and N identifiers are assigned, and the number is used as the value of each identifier; after the new pictures entering the picture library complete the numbering and labeling operation, the new pictures and the old pictures previously stored in the picture library are similarly calculated, and the value of each identifier of the new picture is adjusted based on the similarity calculation result, and N is a positive integer greater than zero; after the value of each identifier of the new picture is determined, the new picture is reviewed according to the value of each identifier of the new picture, wherein, if there is an identifier and a corresponding value that have been identified as automatically passed or automatically rejected, the new picture is processed as automatically passed or automatically rejected, otherwise the new picture is manually reviewed to obtain the processing result;

[0017] The numbering operation for new pictures entering the picture library includes:

[0018] Numbering new pictures entering the picture library, assigning four identifiers to the new pictures after numbering, namely, a first identifier, a second identifier, a third identifier, and a fourth identifier, and using the number of the new picture as the value of the first identifier, the second identifier, the third identifier, and the fourth identifier;

[0019] The similarity between the new image and the old image previously stored in the image library is calculated, and the value of each identifier of the new image is adjusted based on the similarity calculation result, including:

[0020] Based on the feature information, the similarity between the new image currently entering the image library and the old image that already existed in the image library is calculated. After the calculation, the old image corresponding to the maximum similarity value is screened out, and the similarity range where the maximum similarity value is located is analyzed; wherein, if the maximum similarity value is greater than the first percentage, the value of the first identifier of the old image is used as the value of the first identifier of the new image; if the maximum similarity value is greater than the second percentage and is less than or equal to the first percentage, the value of the second identifier of the old image is used as the value of the second identifier of the new image; if the maximum similarity value is greater than the third percentage and is less than or equal to the second percentage, the value of the third identifier of the old image is used as the value of the third identifier of the new image; if the maximum similarity value is greater than the fourth percentage and is less than or equal to the third percentage, the value of the fourth identifier of the old image is used as the value of the fourth identifier of the new image, and the final value of each identifier of the new image is obtained, and the first percentage, the second percentage, the third percentage, and the fourth percentage decrease in sequence.

[0021] In a third aspect, an embodiment of the present application provides an image processing device, the device comprising a memory and a processor. The memory is used to store a computer program; the processor is used to implement the steps of the above-mentioned image processing method when executing the computer program.

[0022] In a fourth aspect, an embodiment of the present application provides a readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the above-mentioned image processing method are implemented.

[0023] The beneficial effects of the present invention are:

[0024] 1. The present invention can improve the ability to identify and mine multimodal content in social scenarios; solve the problem that some illegal content in pictures is highly concealed and difficult to be found by commonly used risk identification models; solve the limitation that the unqualified list must be added in advance, and the pictures have timeliness and other limitations of the unqualified list; reduce the cost of manual review, and improve timeliness and accuracy. 2. The method of the present invention can improve the ability to identify and mine multimodal content in social scenarios. Some illegal content in pictures is highly concealed and difficult to be found by commonly used risk identification models. Different business scenarios have different aggregation of illegal pictures. The present invention adopts different algorithm models and strategies for risk mining according to different scenarios to improve the accuracy and effectiveness of the strategy. In some scenarios, by detecting the number, position, size, etc. of the subject and the subject information, the strategy of combining business scenario characteristics and portrait characteristics is used for risk mining. The manual review results are combined with the characteristic vector, and the strategy of learning from one example is used to expand the results of manual review. The similarity aggregation statistical strategy is used to effectively mine new forms of violations. The historical retrieval strategy uses the vector database to quickly and effectively discover hidden illegal users and crack down on them.

[0025] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or be understood by implementing the embodiments of the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description, claims, and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0027] Figure 1 is a schematic diagram of the process of the picture processing method described in an embodiment of the present invention;

[0028] Figure 2 is a schematic diagram of the structure of the image processing device described in an embodiment of the present invention;

[0029] Figure 3 It is a schematic diagram of the structure of the image processing device described in an embodiment of the present invention. DETAILED DESCRIPTION

[0030] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. The components of the embodiments of the present invention generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0031] It should be noted that similar reference numerals or letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in the subsequent drawings. At the same time, in the description of the present invention, the terms "first", "second", etc. are only used to distinguish the description and cannot be understood as indicating or implying relative importance.

[0032] Example 1

[0033] like Figure 1 As shown, this embodiment provides a picture processing method, which includes step S1, step S2 and step S3.

[0034] Step S1, obtaining the picture to be processed sent by the user;

[0035] Step S2, determining the scene to which the image to be processed belongs according to the environment in which the image to be processed is generated, and extracting feature information of the image to be processed using a feature extraction method corresponding to the scene;

[0036] The specific implementation steps of this step include step S21;

[0037] Step S21, analyzing the image to be processed, wherein, if the image to be processed is a picture generated by the interaction of two users in the social software, the image to be processed belongs to scene one, and the image to be processed is calculated through 13 convolutional layers and 3 fully connected layers to obtain feature information; if the image to be processed is a picture generated by the interaction of three or more users in the social software, the image to be processed belongs to scene two, and the image to be processed is calculated through 43 convolutional layers and 1 fully connected layer to obtain feature information; in addition, the image to be processed generated in the social software belongs to scene three, and other feature extraction rules are used to extract feature information of the image to be processed.

[0038] In this step, scenario one can be understood as a scenario in which two users chat privately in a social software, that is, a picture sent by one of the users; scenario two can be understood as a scenario in which three, four or more users chat in a group in a social software, that is, a picture sent by one of the users; scenario three can be understood as a picture that a user wants to upload in the social software except for scenario one and scenario two; except for scenario one and scenario two, feature information is extracted from other pictures to be processed using other feature extraction rules, and the specific extraction method includes step S211;

[0039] Step S211, grayscale the image to be processed, extract the subject information in the image to be processed after the grayscale processing, the subject information includes the number of subjects, the size of the subjects, and the position information of the subjects; judge whether the image to be processed violates the rules according to the number of subjects, the size of the subjects, the position information of the subjects and preset rules, and if so, refuse to pass the image to be processed; otherwise, obtain the user's personal information, judge whether the image to be processed violates the rules according to the personal information and the subject information of the image to be processed, and if so, refuse to pass the image to be processed, otherwise, cut out the subject in the image to be processed, and use the CNN algorithm to extract feature information of the subject.

[0040] In this step, it can be understood that the subject includes people, and the person information in the image to be processed is extracted, including the number of people, the size of the people and the position information of the people. Different rules will be set for different scenes to determine whether it is in violation. For example, the rule is that there can only be one subject. If multiple subjects appear, it is considered a violation. If the size of the subject is smaller than the preset size, it is considered a violation. If the position of the subject deviates from the preset position range, it is considered a violation. After obtaining the number of people, the size of the person and the position information of the person, as long as one of them does not meet the regulations, the image to be processed is deemed to be in violation and is rejected. If it is not in violation, the user's personal information will continue to be obtained, and the image to be processed will be determined to be in violation based on the personal information and the subject information of the image to be processed. Specifically, the gender information in the personal information is matched with the gender information in the subject information. If the two are inconsistent, it is in violation.

[0041] Step S3, establish a picture library for each scenario, in each scenario, each time a user sends a picture, the picture will be stored in the picture library; perform numbering and labeling operations on new pictures entering the picture library: number the new pictures, assign N identifiers and use the number as the value of each identifier; after the new pictures entering the picture library complete the numbering and labeling operation, the new pictures will be similarly calculated with the old pictures previously stored in the picture library, and the value of each identifier of the new picture will be adjusted based on the similarity calculation result, and N is a positive integer greater than zero; after the value of each identifier of the new picture is determined, the new picture will be reviewed according to the value of each identifier of the new picture, wherein, if there is an identifier and the corresponding value that has been identified as automatically passed or automatically rejected, the new picture will be processed as automatically passed or automatically rejected, otherwise the new picture will be manually reviewed to obtain the processing result;

[0042] The numbering operation for new pictures entering the picture library includes:

[0043] Numbering new pictures entering the picture library, assigning four identifiers to the new pictures after numbering, namely, a first identifier, a second identifier, a third identifier, and a fourth identifier, and using the number of the new picture as the value of the first identifier, the second identifier, the third identifier, and the fourth identifier;

[0044] The similarity between the new image and the old image previously stored in the image library is calculated, and the value of each identifier of the new image is adjusted based on the similarity calculation result, including:

[0045] Based on the feature information, the similarity between the new image currently entering the image library and the old image that already existed in the image library is calculated. After the calculation, the old image corresponding to the maximum similarity value is screened out, and the similarity range where the maximum similarity value is located is analyzed; wherein, if the maximum similarity value is greater than the first percentage, the value of the first identifier of the old image is used as the value of the first identifier of the new image; if the maximum similarity value is greater than the second percentage and is less than or equal to the first percentage, the value of the second identifier of the old image is used as the value of the second identifier of the new image; if the maximum similarity value is greater than the third percentage and is less than or equal to the second percentage, the value of the third identifier of the old image is used as the value of the third identifier of the new image; if the maximum similarity value is greater than the fourth percentage and is less than or equal to the third percentage, the value of the fourth identifier of the old image is used as the value of the fourth identifier of the new image, and the final value of each identifier of the new image is obtained, and the first percentage, the second percentage, the third percentage, and the fourth percentage decrease in sequence.

[0046] This step can be understood as follows: each scenario corresponds to a picture library. Initially, there are no pictures in the picture library. Each subsequent picture sent by the user will be stored in the picture library. In addition, when each picture enters the picture library, it will be numbered and labeled. The specific numbering and labeling operation is: number the pictures, assign four identifiers and use the number as the value of each identifier. For example, when the first picture enters, the first picture is assigned the number 1001, and four identifiers are generated for the first picture. The four identifiers are simlarId1, simlarId2, simlarId3, and simlarId4, and simlarId1, simlarId2, simlarId3, and simlarId4 are assigned Values, simlarId1, simlarId2, simlarId3, simlarId4 are all the numbers of this picture, that is, "1001"; when the second picture enters the picture library, the second picture is assigned the number 1002, and four identifiers are generated for the second picture, namely simlarId1, simlarId2, simlarId3, simlarId4, and values ​​are assigned to simlarId1, simlarId2, simlarId3, simlarId4, and the values ​​of simlarId1, simlarId2, simlarId3, simlarId4 are all the numbers of this picture, that is, "1002";

[0047] The first percentage, the second percentage, the third percentage, and the fourth percentage are 95%, 90%, 85%, and 80% respectively; after each picture is numbered and labeled, a similarity calculation will be performed, and the values ​​of the four identifiers will be adjusted based on the similarity calculation results. Specifically, based on the feature information, the new picture currently entering the picture library and the old pictures that already exist in the picture library will be calculated for similarity. After the calculation, the old picture corresponding to the maximum similarity value will be screened out, and the similarity range of the maximum similarity value will be analyzed. Specifically, if the maximum similarity value is greater than 95%, the value of simlarId1 of the old picture will be adjusted. As the value of simlarId1 of the new image, that is, overwrite the value. If the maximum similarity is greater than 90% and less than or equal to 95%, the value of simlarId2 of the old image is used as the value of simlarId2 of the new image. If the maximum similarity is greater than 85% and less than or equal to 90%, the value of simlarId3 of the old image is used as the value of simlarId3 of the new image. If the maximum similarity is greater than 80% and less than or equal to 85%, the value of simlarId4 of the old image is used as the value of simlarId4 of the new image, and the final value of each identifier of each image is obtained;

[0048] After the values ​​of the four identifiers are adjusted, a judgment will be made as to whether there is a violation. In the process of violation judgment, manual judgment is first used. In the continuous judgment process, if there are H pictures with the same simlarId1 value that are manually reviewed as violations, and H is greater than the preset threshold, then simlarId1 and its value will be put into the cache. When the simlarId1 value of the new picture entering the picture library is the same as the simlarId1 value in the cache, it will be automatically judged as a violation and rejected. This method can improve the timeliness and efficiency of the review.

[0049] After the new pictures entering the picture library are numbered and labeled in step S3, step S4 may also be included;

[0050] Step S4, for each scenario, record an identifier and its value as a mark, count the number of pictures containing the same mark, sort them after counting, send all pictures corresponding to the top M marks to the staff for manual review, and manually determine whether this mark corresponds to the unqualified list or the qualified list. When a new picture enters the picture library and calculates the value corresponding to each identifier, determine whether it contains the mark corresponding to the unqualified list or the qualified list. If it contains the mark corresponding to the unqualified list, it will be rejected; if it contains the mark corresponding to the qualified list, it will be allowed to pass; otherwise, it will be manually determined to pass or refuse to pass, and M is a positive integer greater than zero.

[0051] This step can be understood as: count the pictures with the identifier simlarId1 1001, count the pictures with simlarId2 1001, and count the pictures with simlarId2 1002, that is, count the pictures with the same identifier and the same identifier value; after the statistics, conduct manual review. After the review, if the picture with the identifier simlarId1 1001 is identified as unqualified, then after the subsequent new pictures calculate the value of each identifier, if the value of simlarId1 of the new picture is also 1001, it will be automatically rejected and no manual review is required;

[0052] In addition to the content in step S4, all the images corresponding to the first M marks can also be sent to the staff for manual review, and the staff can manually determine whether the image corresponding to this mark contains a new violation; when it is manually determined that the image corresponding to this mark contains a new violation, all users corresponding to the images containing this mark are recorded as violating users, so that corresponding treatment measures can be taken against the violating users. In this way, the operation of historical retrieval can be completed to more comprehensively mine illegal images and illegal users.

[0053] The solution in this embodiment can be applied in social software. According to the solution in this embodiment, the processing of pictures generated in social software can be completed. When it is in scenario one or scenario two, when it is rejected, it can be shown that the picture sent by the user cannot be successfully sent out and cannot be received by others; when it is allowed to pass, it can be shown that the picture sent by the user can be successfully sent out and can be received by others; when it is in scenario three, when it is rejected, it can be shown that the picture sent by the user cannot be successfully uploaded; when it is allowed to pass, it can be shown that the picture sent by the user can be successfully uploaded;

[0054] Through the method in this embodiment, the ability to identify and mine multimodal content in social scenarios can be improved. The illegal content of some pictures is highly concealed and difficult to detect with commonly used risk identification models. Different business scenarios have different aggregation of illegal pictures. The present invention adopts different algorithm models and strategies for risk mining according to different scenarios to improve the accuracy and effectiveness of the strategy. In some scenarios, risk mining is performed by detecting the number, position, size, etc. of the subject and the subject information, combining the business scenario characteristics and portrait characteristics. The manual review results are combined with the characteristic vector, and the strategy of learning from one example can expand the results of manual review. The similarity aggregation statistical strategy is used to effectively mine new forms of violations. The historical retrieval strategy uses the vector database to quickly and effectively discover hidden illegal users and crack down on them.

[0055] Example 2

[0056] like Figure 2 As shown, this embodiment provides a picture processing device, which includes an acquisition module 1, an extraction module 2 and a processing module 3.

[0057] Acquisition module 1, used to acquire the image to be processed sent by the user;

[0058] Extraction module 2, used to determine the scene to which the image to be processed belongs according to the environment in which the image to be processed is generated, and extract feature information of the image to be processed using a feature extraction method corresponding to the scene;

[0059] Processing module 3 is used to establish a picture library for each scenario. In each scenario, each time a user sends a picture, the picture will be stored in the picture library; the new pictures entering the picture library are numbered and labeled: the new pictures are numbered, and N identifiers are assigned and the number is used as the value of each identifier; after the new pictures entering the picture library complete the numbering and labeling operation, the new pictures are similar to the old pictures previously stored in the picture library, and the value of each identifier of the new picture is adjusted based on the similarity calculation result, and N is a positive integer greater than zero; after the value of each identifier of the new picture is determined, the new picture is reviewed according to the value of each identifier of the new picture, wherein, if there is an identifier and the corresponding value that have been identified as automatically passed or automatically rejected, the new picture is processed as automatically passed or automatically rejected, otherwise the new picture is manually reviewed to obtain the processing result;

[0060] The numbering operation for new pictures entering the picture library includes:

[0061] Numbering new pictures entering the picture library, assigning four identifiers to the new pictures after numbering, namely, a first identifier, a second identifier, a third identifier, and a fourth identifier, and using the number of the new picture as the value of the first identifier, the second identifier, the third identifier, and the fourth identifier;

[0062] The similarity between the new image and the old image previously stored in the image library is calculated, and the value of each identifier of the new image is adjusted based on the similarity calculation result, including:

[0063] Based on the feature information, the similarity between the new image currently entering the image library and the old image that already existed in the image library is calculated. After the calculation, the old image corresponding to the maximum similarity value is screened out, and the similarity range where the maximum similarity value is located is analyzed; wherein, if the maximum similarity value is greater than the first percentage, the value of the first identifier of the old image is used as the value of the first identifier of the new image; if the maximum similarity value is greater than the second percentage and is less than or equal to the first percentage, the value of the second identifier of the old image is used as the value of the second identifier of the new image; if the maximum similarity value is greater than the third percentage and is less than or equal to the second percentage, the value of the third identifier of the old image is used as the value of the third identifier of the new image; if the maximum similarity value is greater than the fourth percentage and is less than or equal to the third percentage, the value of the fourth identifier of the old image is used as the value of the fourth identifier of the new image, and the final value of each identifier of the new image is obtained, and the first percentage, the second percentage, the third percentage, and the fourth percentage decrease in sequence.

[0064] It should be noted that, regarding the device in the above embodiment, the specific manner in which each module performs the operation has been described in detail in the embodiment of the method, and will not be elaborated here.

[0065] Example 3

[0066] Corresponding to the above method embodiment, the embodiment of the present disclosure further provides a picture processing device, and the picture processing device described below and the picture processing method described above can refer to each other.

[0067] Figure 3 FIG. 3 is a block diagram of a picture processing device 300 according to an exemplary embodiment. Figure 3 As shown, the image processing device 300 may include: a processor 301 and a memory 302. The image processing device 300 may also include one or more of a multimedia component 303, an I / O interface 304, and a communication component 305.

[0068] The processor 301 is used to control the overall operation of the image processing device 300 to complete all or part of the steps in the above-mentioned image processing method. The memory 302 is used to store various types of data to support the operation of the image processing device 300. For example, these data may include instructions for any application or method used to operate on the image processing device 300, and application-related data, such as contact data, sent and received messages, pictures, audio, video, etc. The memory 302 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read-Only Memory, referred to as EPROM), programmable read-only memory (Programmable Read-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic memory, flash memory, magnetic disk or optical disk. The multimedia component 303 may include a screen and an audio component. The screen may be, for example, a touch screen, and the audio component is used to output and / or input audio signals. For example, the audio component may include a microphone for receiving external audio signals. The received audio signal may be further stored in the memory 302 or sent through the communication component 305. The audio component also includes at least one speaker for outputting audio signals. The I / O interface 304 provides an interface between the processor 301 and other interface modules, and the above-mentioned other interface modules may be keyboards, mice, buttons, etc. These buttons may be virtual buttons or physical buttons. The communication component 305 is used for wired or wireless communication between the image processing device 300 and other devices. Wireless communication, such as Wi-Fi, Bluetooth, Near Field Communication (NFC), 2G, 3G or 4G, or a combination of one or more of them, so the corresponding communication component 305 may include: Wi-Fi module, Bluetooth module, NFC module.

[0069] In an exemplary embodiment, the image processing device 300 can be implemented by one or more application specific integrated circuits (ASIC), digital signal processors (DSP), digital signal processing devices (DSPD), programmable logic devices (PLD), field programmable gate arrays (FPGA), controllers, microcontrollers, microprocessors or other electronic components to execute the above-mentioned image processing method.

[0070] In another exemplary embodiment, a computer-readable storage medium including program instructions is also provided, and when the program instructions are executed by the processor, the steps of the above-mentioned image processing method are implemented. For example, the computer-readable storage medium can be the above-mentioned memory 302 including program instructions, and the above-mentioned program instructions can be executed by the processor 301 of the image processing device 300 to complete the above-mentioned image processing method.

[0071] Example 4

[0072] Corresponding to the above method embodiment, the embodiment of the present disclosure further provides a readable storage medium. The readable storage medium described below and the image processing method described above can refer to each other.

[0073] A readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the image processing method of the above method embodiment are implemented.

[0074] The readable storage medium may specifically be a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, or other readable storage medium that can store program codes.

[0075] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for processing an image, characterized in that: include: Get the pictures to be processed sent by the user; Determine the scene to which the picture to be processed belongs according to the environment in which the picture to be processed is generated, and extract feature information of the picture to be processed using a feature extraction method corresponding to the scene; A picture library is established for each scenario. In each scenario, each time a user sends a picture, the picture will be stored in the picture library. New pictures entering the picture library are numbered and labeled: the new pictures are numbered and assigned N identifiers, and the number is used as the value of each identifier. After the new pictures entering the picture library are numbered and labeled, the similarity between the new pictures and the old pictures previously stored in the picture library is calculated, and the value of each identifier of the new picture is adjusted based on the similarity calculation result, where N is a positive integer greater than zero. After the value of each identifier of the new image is determined, the new image is reviewed according to the value of each identifier of the new image. If there is an identifier and a corresponding value that has been determined to be automatically passed or automatically rejected, the new image is processed as automatically passed or automatically rejected. Otherwise, the new image is manually reviewed to obtain a processing result. The operation of numbering and labeling new pictures entering the picture library includes: Numbering new pictures entering the picture library, assigning four identifiers to the new pictures after numbering, namely, a first identifier, a second identifier, a third identifier, and a fourth identifier, and using the number of the new picture as the value of the first identifier, the second identifier, the third identifier, and the fourth identifier; The similarity between the new image and the old image previously stored in the image library is calculated, and the value of each identifier of the new image is adjusted based on the similarity calculation result, including: Based on the feature information, the similarity between the new image currently entering the image library and the old image that already existed in the image library is calculated. After the calculation, the old image corresponding to the maximum similarity value is screened out, and the similarity range where the maximum similarity value is located is analyzed; wherein, if the maximum similarity value is greater than the first percentage, the value of the first identifier of the old image is used as the value of the first identifier of the new image; if the maximum similarity value is greater than the second percentage and is less than or equal to the first percentage, the value of the second identifier of the old image is used as the value of the second identifier of the new image; if the maximum similarity value is greater than the third percentage and is less than or equal to the second percentage, the value of the third identifier of the old image is used as the value of the third identifier of the new image; if the maximum similarity value is greater than the fourth percentage and is less than or equal to the third percentage, the value of the fourth identifier of the old image is used as the value of the fourth identifier of the new image, and the final value of each identifier of the new image is obtained, and the first percentage, the second percentage, the third percentage, and the fourth percentage decrease in sequence.

2. The image processing method according to claim 1, characterized in that: Determining the scene to which the picture to be processed belongs according to the environment in which the picture to be processed is generated, and extracting feature information of the picture to be processed using a feature extraction method corresponding to the scene, including: The image to be processed is analyzed. If the image to be processed is a picture generated by the interaction of two users in the social software, the image to be processed belongs to scene one, and the image to be processed is calculated through 13 convolutional layers and 3 fully connected layers to obtain feature information; if the image to be processed is a picture generated by the interaction of three or more users in the social software, the image to be processed belongs to scene two, and the image to be processed is calculated through 43 convolutional layers and 1 fully connected layer to obtain feature information; in addition, the image to be processed generated in the social software belongs to scene three, and other feature extraction rules are used to extract feature information of the image to be processed.

3. The image processing method according to claim 2, characterized in that: Other feature extraction rules are used to extract feature information of the image to be processed, including: The image to be processed is gray-scaled, and the subject information of the image to be processed is extracted after the gray-scale processing, the subject information including the number of subjects, the size of the subjects, and the position information of the subjects; according to the number of subjects, the size of the subjects, the position information of the subjects and preset rules, it is judged whether the image to be processed violates the rules, and if so, the image to be processed is rejected; otherwise, the user's personal information is obtained, and according to the personal information and the subject information of the image to be processed, it is judged whether the image to be processed violates the rules, and if so, the image to be processed is rejected, otherwise the subject in the image to be processed is cut out, and the feature information of the subject is extracted by using the CNN algorithm.

4. The image processing method according to claim 1, characterized in that: After the new pictures entering the picture library are numbered and labeled, the following steps are also included: For each scenario, an identifier and its value are recorded as a mark, and the number of pictures containing the same mark is counted and sorted. All pictures corresponding to the top M marks are sent to the staff for manual review, and the staff manually determines whether the mark corresponds to the unqualified list or the qualified list. When a new picture enters the picture library and the value corresponding to each identifier is calculated, it is determined whether it contains the mark corresponding to the unqualified list or the qualified list. If it contains the mark corresponding to the unqualified list, it is rejected; if it contains the mark corresponding to the qualified list, it is allowed to pass; otherwise, it is manually determined to pass or reject it. M is a positive integer greater than zero.

5. A picture processing device, characterized in that: include: The acquisition module is used to obtain the pictures to be processed sent by the user; An extraction module, used to determine the scene to which the image to be processed belongs according to the environment in which the image to be processed is generated or the information contained therein, and to extract feature information of the image to be processed using a feature extraction method corresponding to the scene; The processing module is used to establish a picture library for each scenario. In each scenario, each time a user sends a picture, the picture will be stored in the picture library; the new pictures entering the picture library are numbered and labeled: the new pictures are numbered and assigned N identifiers, and the number is used as the value of each identifier; after the new pictures entering the picture library complete the numbering and labeling operation, the new pictures are calculated for similarity with the old pictures previously stored in the picture library, and the value of each identifier of the new picture is adjusted based on the similarity calculation result, where N is a positive integer greater than zero; After the value of each identifier of the new image is determined, the new image is reviewed according to the value of each identifier of the new image. If there is an identifier and a corresponding value that has been determined to be automatically passed or automatically rejected, the new image is processed as automatically passed or automatically rejected. Otherwise, the new image is manually reviewed to obtain a processing result. The operation of numbering and labeling new pictures entering the picture library includes: Numbering new pictures entering the picture library, assigning four identifiers to the new pictures after numbering, namely, a first identifier, a second identifier, a third identifier, and a fourth identifier, and using the number of the new picture as the value of the first identifier, the second identifier, the third identifier, and the fourth identifier; The similarity between the new image and the old image previously stored in the image library is calculated, and the value of each identifier of the new image is adjusted based on the similarity calculation result, including: Based on the feature information, the similarity between the new image currently entering the image library and the old image that already existed in the image library is calculated. After the calculation, the old image corresponding to the maximum similarity value is screened out, and the similarity range where the maximum similarity value is located is analyzed; wherein, if the maximum similarity value is greater than the first percentage, the value of the first identifier of the old image is used as the value of the first identifier of the new image; if the maximum similarity value is greater than the second percentage and is less than or equal to the first percentage, the value of the second identifier of the old image is used as the value of the second identifier of the new image; if the maximum similarity value is greater than the third percentage and is less than or equal to the second percentage, the value of the third identifier of the old image is used as the value of the third identifier of the new image; if the maximum similarity value is greater than the fourth percentage and is less than or equal to the third percentage, the value of the fourth identifier of the old image is used as the value of the fourth identifier of the new image, and the final value of each identifier of the new image is obtained, and the first percentage, the second percentage, the third percentage, and the fourth percentage decrease in sequence.

6. A picture processing device, characterized in that: include: Memory for storing computer programs; A processor, configured to implement the steps of the image processing method according to any one of claims 1 to 4 when executing the computer program.

7. A storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the image processing method according to any one of claims 1 to 4 are implemented.

Images