CN108024148B

CN108024148B - Behavior feature-based multimedia file identification method, processing method and device

Info

Publication number: CN108024148B
Application number: CN201610929276.7A
Authority: CN
Inventors: 刘杰
Original assignee: Tencent Technology Shenzhen Co Ltd
Current assignee: Tencent Technology Shenzhen Co Ltd
Priority date: 2016-10-31
Filing date: 2016-10-31
Publication date: 2020-02-28
Anticipated expiration: 2036-10-31
Also published as: CN108024148A

Abstract

The invention relates to the technical field of internet, in particular to a multimedia file identification method, a multimedia file processing method and a multimedia file processing device based on behavior characteristics. The method analyzes the correlation between the internet surfing behavior of the user and the watching of the specific content, and proposes that in the playing process of the multimedia file, the probability of the specific content contained in the multimedia file is calculated according to the portrait characteristic value of each user and the first intention characteristic value by obtaining the portrait characteristic value of the audience user and the first intention characteristic value for expressing that the user wants to watch the specific content within the preset time, and the probability is compared with the preset value to determine whether the multimedia file needs to be further detected, so that the multimedia file to be analyzed is obtained by the auxiliary screening of the user behavior characteristics, the specific content of the screened multimedia file is detected, and the identification efficiency and the accuracy of the multimedia file are improved. The method is used for pornographic video detection, can greatly improve the detection rate and reliability, and is convenient for the management and control of multimedia videos.

Description

Behavior feature-based multimedia file identification method, processing method and device

Technical Field

The invention relates to the technical field of internet, in particular to a multimedia file identification method, a multimedia file processing method and a multimedia file processing device based on behavior characteristics.

Background

With the rapid development of the internet and multimedia technology, video media in the network is ubiquitous and becomes an important part of our life and entertainment. However, the social cultural life is seriously influenced by the full abundance of pornographic videos. How to effectively prevent the spread of the pornographic videos on the network has great significance for maintaining physical and mental health of teenagers and purifying network video resources.

Pornographic video detection methods have been emerging continuously over the past few years. These methods fall into three categories: the first type is that key frames of the video are extracted, and pornographic content detection is carried out on the key frames; the second type is to extract the audio frequency in the video and identify the audio frequency content; and in the third category, motion vectors of the video are acquired, and the motion characteristics are analyzed. But the biggest difference of pornographic video relative to normal video is that: the pornographic video frame contains a large amount of naked skin. Thus, the first method has a higher accuracy than the other two methods. In the existing patent, a plurality of key frames in a video are extracted, pornographic content detection is carried out on the key frames, and decision fusion judgment is carried out according to the detection result of pornographic content of a single key frame and the detection results of other previously obtained key frames. However, the first method also has disadvantages, namely: this method has a reduced possibility of recognizing erotic content if the erotic content occupies only a small portion of a long video segment, and furthermore, it has a low recognition efficiency to recognize erotic video from a large amount of video.

Disclosure of Invention

The inventor finds out through research that: the internet behavior of the user can reflect the preference of the user, for example, the user of the favorite drama generally browses the drama plates of various movie evaluation websites, and the user can watch the videos and also watches the dramas; users who prefer pornographic content pay attention to various adult websites, pornographic forums and the like, and the users who prefer pornographic content often see videos of human content when watching the videos. User preferences and user behavior may be directed to videos of particular content, for example, when a user views pornographic text or pornographic pictures and then views the video or live broadcast, the video or live broadcast has a greater likelihood of containing pornographic content; meanwhile, if 10 people are watching a video, of which 8 are users who prefer pornographic content or have just browsed pornography text or pornography pictures, there is a high possibility that the video contains pornography content.

Based on the association of user preference and user behavior with specific content videos, the behavior and preference of the user on the Internet can be added into pornographic content identification as an auxiliary dimension to identify whether the multimedia file contains pornographic content or not more effectively; based on the above, the invention provides a multimedia file identification method, a multimedia file processing method and a multimedia file identification device based on behavior characteristics.

In a first aspect, the present invention provides a multimedia file identification method based on behavioral characteristics, including:

in the multimedia file playing process, obtaining a portrait characteristic value and a first intention characteristic value of a viewer user, wherein the portrait characteristic value is used for identifying the preference of the user for specific content, and the first intention characteristic value is used for identifying the intention of the user wishing to watch the specific content within a preset time period;

calculating the probability that the multimedia file contains specific content according to the portrait characteristic value and the first intention characteristic value;

judging whether the probability exceeds a preset value, if so, performing feature detection on the multimedia file;

and judging whether the multimedia file is a multimedia file with specific content or not according to the characteristic detection result.

Preferably, the calculating the probability that the multimedia file contains the specific content according to the portrait characteristic value and the first intention characteristic value includes:

determining a second intention characteristic value of each user according to the portrait characteristic value and the first intention characteristic value;

and calculating the probability that the multimedia file contains the specific content according to the second intention characteristic values of all the users.

Preferably, the determining a second intention characteristic value of each user according to the portrait characteristic value and the first intention characteristic value includes: and summing the portrait characteristic value and the first intention characteristic value to obtain the second intention characteristic value, or carrying out weighted average on the portrait characteristic value and the first intention characteristic value according to weights preset for the portrait characteristic value and the first intention characteristic value to obtain the second intention characteristic value.

Preferably, the calculating the probability that the multimedia file contains the specific content according to the second will characteristic values of all the users includes: respectively comparing the second intention characteristic value of each user with a preset threshold value; and calculating the ratio of the number of the users with the second will characteristic value exceeding a threshold value to the total number of the users to obtain the probability that the multimedia file contains the specific content.

Further, before obtaining the portrait characteristic value and the first intention characteristic value of the audience user, the method further comprises the following steps: and analyzing the behavior data of the user, and determining the portrait characteristic value and the first intention characteristic value of the user.

Preferably, the analyzing the behavior data of the user and determining the portrait characteristic value of the user includes: acquiring behavior data of a user, wherein the behavior data comprises first behavior data for browsing a text related to specific content, second behavior data for browsing a picture related to the specific content, third behavior data for accessing a forum related to the specific content and fourth behavior data for chatting in a chat group related to the specific content; respectively judging whether the first behavior data, the second behavior data, the third behavior data and the fourth behavior data are empty or not, and correspondingly obtaining a first judgment result, a second judgment result, a third judgment result and a fourth judgment result; and according to a preset first weight of the first judgment result, a preset second weight of the second judgment result, a preset third weight of the third judgment result and a preset fourth weight of the fourth judgment result, distributing and integrating the first judgment result, the second judgment result, the third judgment result and the fourth judgment result to obtain the behavior characteristic value of the user.

Preferably, the analyzing the behavior data of the user and determining the first will characteristic value of the user includes: acquiring behavior data of a user in a recent period of time, wherein the behavior data comprises a first time for browsing a text related to specific content, a second time for browsing a picture related to the specific content, a third time for accessing a forum related to the specific content and a fourth time for chatting in a chat group related to the specific content; and giving the first weight to the first time, giving the second weight to the second time, giving the third weight to the third time, giving the fourth weight to the fourth time, and carrying out weighted average on the first time, the second time, the third time and the fourth time to obtain a user intention characteristic value.

Preferably, the multimedia file is a video; the performing feature detection on the multimedia file comprises: extracting a preset number of images at equal time intervals of a video; and performing feature detection on each image, and judging whether the image contains a specific feature, wherein the feature detection comprises sensitive part detection and skin color pixel detection.

Preferably, the determining whether the multimedia file is a multimedia file with specific content according to the feature detection result includes: when the number of the images containing the specific features is judged to be larger than a preset threshold value P, the video is judged to be a specific content video, otherwise, the video is judged to be a normal video; or, determining the ratio of the number of the judged images containing the specific features to the total number of the images obtained by detecting and extracting the video, and judging that the video is the specific content video when the determined ratio is greater than a threshold value Q, or judging that the video is the normal video.

Preferably, the specific content is pornographic content, and the video is an on-demand video or a live video.

In a second aspect, the present invention provides a multimedia file processing method, including:

judging whether the multimedia file is a multimedia file with specific content or not according to the feature detection result;

and processing the multimedia file according to the judgment result.

Preferably, the multimedia file is an on-demand video or a live video, and the specific content is pornographic content; the processing the multimedia file according to the judgment result comprises the following steps: if the multimedia file is the requested pornographic video, quitting the playing of the requested video; and if the multimedia file is the live pornographic video, closing the video live broadcast room for playing the video.

In a third aspect, the present invention provides a multimedia file identification apparatus based on behavior characteristics, including:

the multimedia file playing device comprises an acquisition unit, a display unit and a playing unit, wherein the acquisition unit is used for acquiring a portrait characteristic value and a first intention characteristic value of a viewer user in the playing process of a multimedia file, the portrait characteristic value is used for identifying the preference of the user for specific content, and the first intention characteristic value is used for identifying the intention of the user who wants to watch the specific content within a preset time period;

a calculating unit, configured to calculate, according to the portrait feature value and the first will feature value, a probability that the multimedia file contains a specific content;

the detection unit is used for judging whether the probability exceeds a preset value or not, and if so, performing feature detection on the multimedia file;

and the determining unit is used for judging whether the multimedia file is a multimedia file with specific content according to the characteristic detection result.

Preferably, the calculation unit includes: the first calculation subunit is used for determining a second intention characteristic value of each user according to the portrait characteristic value and the first intention characteristic value; and the second calculating subunit is used for calculating the probability that the multimedia file contains the specific content according to the second intention characteristic values of all the users.

Preferably, the first calculation subunit includes: the first calculation module is used for summing the portrait characteristic value and the first intention characteristic value to obtain a second intention characteristic value; and the second calculation module is used for carrying out weighted average on the portrait characteristic value and the first intention characteristic value according to weights preset for the portrait characteristic value and the first intention characteristic value to obtain a second intention characteristic value.

Preferably, the second calculation subunit includes: the comparison module is used for comparing the second intention characteristic value of each user with a preset threshold value respectively; and the probability calculation module is used for calculating the ratio of the number of the users with the second will characteristic value exceeding a threshold value to the total number of the users to obtain the probability that the multimedia file contains the specific content.

Further, the apparatus further comprises: and the preprocessing unit is used for analyzing the behavior data of the user and determining the portrait characteristic value and the first intention characteristic value of the user.

Preferably, the preprocessing unit includes:

a first processing subunit to: acquiring behavior data of a user, wherein the behavior data comprises first behavior data for browsing a text related to specific content, second behavior data for browsing a picture related to the specific content, third behavior data for accessing a forum related to the specific content and fourth behavior data for chatting in a chat group related to the specific content; respectively judging whether the first behavior data, the second behavior data, the third behavior data and the fourth behavior data are empty or not, and correspondingly obtaining a first judgment result, a second judgment result, a third judgment result and a fourth judgment result; according to a preset first weight of the first judgment result, a preset second weight of the second judgment result, a preset third weight of the third judgment result and a preset fourth weight of the fourth judgment result, distributing and integrating the first judgment result, the second judgment result, the third judgment result and the fourth judgment result to obtain a behavior characteristic value of the user;

the second processing subunit is used for acquiring behavior data of the user in a recent period of time, wherein the behavior data comprises a first time for browsing a text related to the specific content, a second time for browsing a picture related to the specific content, a third time for accessing a forum related to the specific content and a fourth time for chatting in a chat group related to the specific content; and giving the first weight to the first time, giving the second weight to the second time, giving the third weight to the third time, giving the fourth weight to the fourth time, and carrying out weighted average on the first time, the second time, the third time and the fourth time to obtain a user intention characteristic value.

Preferably, the multimedia file is a video; the detection unit includes: the extraction subunit is used for extracting a preset number of images at equal time intervals of the video; and the detection subunit is used for performing feature detection on each image and judging whether the image contains a specific feature, wherein the feature detection comprises sensitive part detection and skin color pixel detection.

Preferably, the determination unit includes: the first determining subunit is used for determining that the video is a video with specific content when the number of the images containing the specific features is judged to be larger than a preset threshold value P, and otherwise, determining that the video is a normal video; and the second determining subunit is used for determining the ratio of the number of the judged images containing the specific features to the total number of the images obtained by detection and extraction aiming at the video, judging the video to be the specific content video when the determined ratio is greater than a threshold value Q, and otherwise, judging the video to be the normal video.

In a fourth aspect, the present invention provides a multimedia file processing apparatus, comprising:

the determining unit is used for judging whether the multimedia file is a multimedia file with specific content or not according to the characteristic detection result;

and the processing unit is used for processing the multimedia file according to the judgment result.

Preferably, the multimedia file is an on-demand video or a live video, and the specific content is pornographic content; the processing unit is specifically used for quitting playing of the on-demand video when the multimedia file is determined to be the on-demand pornographic video; and when the multimedia file is determined to be the live pornographic video, closing the video live broadcast room for playing the video.

The invention has the following beneficial effects:

the invention analyzes the correlation between the internet surfing behavior of a user and watching specific content, and proposes that in the process of playing a multimedia file, the probability of the multimedia file containing the specific content is calculated according to the portrait characteristic value of each user and the first intention characteristic value by acquiring the portrait characteristic value of the audience user and the first intention characteristic value for expressing that the user wants to watch the specific content within the preset time, and the probability is compared with the preset value to determine whether the multimedia file needs to be further detected, so that the multimedia file to be analyzed is obtained by the auxiliary screening of the user behavior characteristic, the specific content of the screened multimedia file is detected, and the identification efficiency and the accuracy of the multimedia file with the specific content are improved. The method and the device are used for detecting the bad contents of the multimedia files such as pornography, horror and the like, can greatly improve the detection efficiency and reliability, and are convenient for controlling the multimedia files.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions and advantages of the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present invention, and other drawings can be obtained by those skilled in the art without creative efforts.

Fig. 1 is a block diagram of a hardware configuration of a computer terminal that can be used to implement the behavior feature-based multimedia file recognition method according to an embodiment of the present invention;

FIG. 2 is a flowchart of a behavior feature-based multimedia file identification method disclosed in embodiment 1 of the present invention;

FIG. 3 is a flowchart of a behavior feature-based multimedia file identification method disclosed in embodiment 2 of the present invention;

FIG. 4 is a flowchart illustrating a multimedia file processing method according to embodiment 3 of the present invention;

FIG. 5 is a diagram of an apparatus for identifying multimedia files based on behavior features according to embodiment 4 of the present invention;

FIG. 6 is a diagram of an apparatus for identifying multimedia files based on behavior features according to embodiment 4 of the present invention;

FIG. 7 is a diagram of a multimedia file processing apparatus according to embodiment 5 of the present invention;

fig. 8 is a block diagram of a structure of a computer terminal according to an embodiment of the present invention.

Detailed Description

In order to make the technical solutions of the present invention better understood, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

It should be noted that the terms "first," "second," and the like in the description and claims of the present invention and in the drawings described above are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments of the invention described herein are capable of operation in sequences other than those illustrated or described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.

Example 1

The present embodiment provides an embodiment of a method for multimedia file identification based on behavioral characteristics, it should be noted that the steps shown in the flowchart of the figure can be executed in a computer system such as a set of computer executable instructions, and that although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that here.

The method embodiments provided in the present application may be executed in a mobile terminal, a computer terminal or a similar computing device. Taking the example of running on a computer terminal, fig. 1 is a hardware structure block diagram of a computer terminal that can be used to implement the behavior feature-based multimedia file identification method of the present invention. As shown in fig. 1, the computer terminal 100 may include one or more (only one shown) processors 102 (the processors 102 may include, but are not limited to, a processing device such as a microprocessor MCU or a programmable logic device FPGA), a memory 104 for storing data, and a transmission device 106 for communication functions. It will be understood by those skilled in the art that the structure shown in fig. 1 is only an illustration and is not intended to limit the structure of the electronic device. For example, computer terminal 100 may also include more or fewer components than shown in FIG. 1, or have a different configuration than shown in FIG. 1.

The memory 104 may be used to store software programs and modules of application software, such as program instructions/modules corresponding to the multimedia file identification method based on behavioral characteristics in the embodiment of the present invention, and the processor 102 executes various functional applications and data processing by running the software programs and modules stored in the memory 104, that is, implements the above-mentioned multimedia file identification method based on behavioral characteristics. The memory 104 may include high speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include memory located remotely from the processor 102, which may be connected to the computer terminal 10 via a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.

The transmission device 106 is used for receiving or transmitting data via a network. Specific examples of the network described above may include a wireless network provided by a communication provider of the computer terminal 100. In one example, the transmission device 106 includes a Network adapter (NIC), which can be connected to other Network devices through a base station so as to communicate with the internet. In one example, the transmission device 106 may be a Radio Frequency (RF) module, which is used for communicating with the internet in a wireless manner.

Under the operating environment, the application provides a multimedia file identification method based on behavior characteristics as shown in fig. 2. The method can be applied to intelligent terminal equipment, and is executed by a processor in the intelligent terminal equipment, and the intelligent terminal equipment can be an intelligent mobile phone, a tablet personal computer and the like. The intelligent terminal device is provided with at least one application program, and the embodiment of the invention does not limit the types of the application programs, and can be a system application program or a software application program.

FIG. 2 is a flowchart illustrating a behavior feature-based multimedia file recognition method according to an embodiment of the present invention. As shown in fig. 2, an alternative version of the method comprises the steps of:

step S201, in the process of playing a multimedia file, obtaining a portrait characteristic value and a first intention characteristic value of a viewer user, wherein the portrait characteristic value is used for identifying the preference of the user for specific content, and the first intention characteristic value is used for identifying the intention of the user who wants to watch the specific content within a preset time period;

step S202, calculating the probability that the multimedia file contains specific content according to the portrait characteristic value and the first intention characteristic value;

step S203, judging whether the probability exceeds a preset value, if so, performing feature detection on the multimedia file;

step S204, judging whether the multimedia file is a multimedia file with specific content according to the characteristic detection result.

As a preferred implementation manner of step S202, the calculating a probability that the multimedia file contains specific content according to the portrait characteristic value and the first will characteristic value includes:

Example 2

The embodiment provides a video content identification method based on behavior characteristics. Under the operating environment as in embodiment 1, the present application provides a multimedia file identification method based on behavioral characteristics as shown in fig. 3. As shown in fig. 3, fig. 3 is a flowchart of a behavior feature-based multimedia file identification method according to an embodiment of the present invention, where an alternative scheme of the method includes the following steps:

the method comprises the following steps: and analyzing the behavior data of the user, and determining the portrait characteristic value and the first intention characteristic value of the user.

The user's online behavior can reflect the user's preference, and the user portrait can be determined by analyzing the user's behaviors such as searching, browsing, clicking recommendation information, etc., for example, the user portrait is a preferred pornographic video, and accordingly, the user portrait can also assist in judging the user's current or future online behavior, for example, the probability that the user who prefers pornographic video watches pornographic video at present or in the future is greater than that of the user who does not prefer pornographic video. The user portrait can reflect various preferences of the user, so that the current or future behavior of the user is not accurate enough to be judged only by the user portrait, the internet surfing behavior of the user is continuous, the search or browsing of a certain content is continued for a certain time, for example, the user pays attention to pornographic content in the last few minutes, the probability of continuously browsing pornographic-related content in the current or future time is high, and on the basis, the behavior characteristics of the user in the time period before the current time can be referred to assist in judging the current or future behavior of the user.

Analyzing the user behavior data may identify a user's preference for particular content using profile feature values and identify a user's intent with respect to particular content within a period of time prior to a current time using a first intent feature value.

Analyzing the behavior data of the user and determining the portrait characteristic value of the user comprises the following steps: acquiring behavior data of a user, wherein the behavior data comprises first behavior data for browsing a text related to specific content, second behavior data for browsing a picture related to the specific content, third behavior data for accessing a forum related to the specific content and fourth behavior data for chatting in a chat group related to the specific content; respectively judging whether the first behavior data, the second behavior data, the third behavior data and the fourth behavior data are null, if so, marking as 0, and if not, marking as 1, and correspondingly obtaining a first judgment result R1, a second judgment result R2, a third judgment result R3 and a fourth judgment result R4;

and according to a preset first weight W1 of the first judgment result, a preset second weight W3 of the second judgment result, a preset third weight W3 of the third judgment result and a preset fourth weight W4 of the fourth judgment result, distributing and integrating the first judgment result, the second judgment result, the third judgment result and the fourth judgment result to obtain the behavior characteristic value of the user. In one preferred embodiment, the behavior characteristic value B is W1 × R1+ W2 × R2+ W3 × R3+ W4 × R4, and in another preferred embodiment, the behavior characteristic value B is (W1 × R1+ W2 × R2+ W3 × R3+ W4R 4)/4.

Analyzing the behavior data of the user, determining the first will characteristic value of the user, and can be realized by two ways: (1) acquiring screen display content of a user for judgment through similar software such as a computer housekeeper and the like operated on a user terminal; (2) the traffic of the user can be captured on the network, such as a packet capturing on a router, so as to analyze the operation performed by the user. The method comprises the following specific steps: acquiring behavior data of a user in a recent period of time, wherein the behavior data comprises a first time for browsing a text related to specific content, a second time for browsing a picture related to the specific content, a third time for accessing a forum related to the specific content and a fourth time for chatting in a chat group related to the specific content; and giving the first weight W1 to the first time, the second weight W2 to the second time, the third weight W3 to the third time and the fourth weight W4 to the fourth time, and carrying out weighted average on the first time, the second time, the third time and the fourth time to obtain a first will characteristic value of the user.

For example, assuming that the specific content is pornographic content, the portrait feature value represents the preference degree of the user for the pornographic content, the first will feature value represents the desire of the user to watch pornographic video in a period of time before the moment, the online behavior of the user is analyzed, the online behavior mainly comprises whether the user browses pornographic-related characters and pictures in the last period of time, whether the user visits pornographic-related forums or not, and whether the user talks in pornographic chat groups, wherein browsing pornographic novels, pornographic-related segments or microblogs and the like can be regarded as browsing pornographic-related characters, browsing pictures marked as pornographic, pictures on pornographic websites and various beautiful girl pictures on normal websites can be regarded as browsing pornographic-related pictures; and then, calculating an image characteristic value of the user according to the weights of the behavioral characteristics, wherein if the weight corresponding to browsing the pornographic related characters is 0.4, the weight corresponding to browsing the pornographic related pictures is 0.3, the weight for visiting the pornographic forum is 0.6, and the weight for speaking in the pornographic chat group is 0.5, if the user browses the pornographic related pictures in the latest period of time, visits the pornographic related forum and also speaks in the pornographic chat group, the behavioral characteristic value B of the user is 0.4 +0.6 + 1+0.3 + 1+0.5 + 1.4, and according to the analysis of historical data, more attention of the user to the pornographic content is indicated by more than 1, and the user can be marked as the pornographic user. If the user spends 10 minutes to watch the pornograph, and 20 minutes to access the pornograph forum within 40 minutes before the current time, the first will characteristic value is (0.4 × 10+0.3 × 10+0.6 × 20)/40 ═ 0.475.

Step two: in the process of playing the multimedia file, the portrait characteristic value and the first intention characteristic value of the audience user are obtained.

The multimedia files comprise text, pictures, videos and audio files, the scheme can be used for identifying whether the files contain specific content, the specific content can be horror and/or pornographic content, and for example, the scheme of the invention is adopted for identifying whether the text is pornographic text, whether the pictures are pornographic pictures and whether the videos are pornographic videos.

When the multimedia file is a video file, the video can be an on-demand video or a live video, and the live video comprises a video played in a live broadcast room. In the video playing process, a portrait characteristic value and a first intention characteristic value of a viewer user are obtained, the portrait characteristic value is used for identifying the preference degree of the user for specific content, the first intention characteristic value is used for identifying the intention of the user for watching the specific content within a preset time period, and the preset time period generally refers to a time period which is advanced from the current time, for example, 40 minutes before the current time.

Step three: and calculating the probability that the multimedia file contains specific content according to the portrait characteristic value and the first intention characteristic value.

The calculating the probability that the multimedia file contains the specific content according to the portrait characteristic value and the first will characteristic value specifically includes: determining a second intention characteristic value of each user according to the portrait characteristic value and the first intention characteristic value; and calculating the probability that the multimedia file contains the specific content according to the second intention characteristic values of all the users. By combining the portrait characteristic value and the first intention characteristic value, the accuracy of judging whether the multimedia file contains specific content can be improved.

In one embodiment, the second intent feature value may be obtained by summing the portrait feature value and the first intent feature value; and respectively comparing the second intention characteristic value of each user with a preset threshold value, and calculating the ratio of the number of the users with the second intention characteristic values exceeding the threshold value to the total number of the users to obtain the probability that the multimedia file contains the specific content.

In another embodiment, the portrait characteristic value and the first intention characteristic value may be weighted and averaged according to weights preset for the portrait characteristic value and the first intention characteristic value to obtain a second intention characteristic value; and respectively comparing the second intention characteristic value of each user with a preset threshold value, and calculating the ratio of the number of the users with the second intention characteristic values exceeding the threshold value to the total number of the users to obtain the probability that the multimedia file contains the specific content.

Step four: and judging whether the probability exceeds a preset value, and if so, carrying out feature detection on the multimedia file.

The preset value can be set manually, and the preset value can be adjusted by combining the judgment result of whether the multimedia file is the multimedia file with specific content or not, so that the accuracy of the final judgment result is improved. If the probability does not exceed the preset value, the probability that the multimedia file being played contains the specific content is low, and further detection of the multimedia file can be abandoned without any processing for improving the detection efficiency and accuracy. If the probability exceeds a preset value, the video is more likely to contain specific content, and the content of the multimedia file needs to be further detected.

For text files, further detection includes character detection of the text content. A character feature library may be pre-established for storing feature characters extracted from a specific content file (e.g., pornography novels, pornography pictures, etc.), and then the feature characters in the character feature library are used to match with the text content, when a matching result exceeds a preset matching threshold, the text file is said to contain more feature characters, and it may be determined that the text file is a text of a specific content.

For the picture file, the further detection comprises character detection, sensitive part detection, skin color pixel detection, blood color pixel detection and the like on the picture. The character detection utilizes a character feature library to carry out feature character matching for detection, the sensitive part detection utilizes a sensitive part feature library to carry out sensitive part matching for detection, and the blood color pixel detection and the skin color pixel detection can firstly establish a blood color model and a skin color model and then carry out the blood color pixel detection and the skin color pixel detection on the picture according to the blood color model and the skin color model. The construction method of the blood color model and the skin color model is the prior art and is not described herein again.

For the audio file, during further detection, an audio detection model can be trained, and the audio file to be detected is input into the audio detection model to obtain a detection result of whether the audio file contains specific content. The construction method of the audio detection model is the prior art, and is not described herein again.

For the video file, further detection comprises audio detection and image detection; the audio detection can adopt an audio detection model for detection; the image detection comprises the steps of extracting images of the video and carrying out feature detection on the images. Specifically, the extracting an image of the video and performing feature detection on the image includes: extracting a preset number of images at equal time intervals of the video, for example, extracting images by intermittently capturing the video for 10 s; and then, carrying out feature detection on each image to judge whether the image contains specific features, wherein the feature detection comprises motion detection, character detection, sensitive part detection, skin color pixel detection, blood color pixel detection and the like.

Step five: and judging whether the multimedia file is a multimedia file with specific content or not according to the characteristic detection result.

In one embodiment, whether the number of pictures containing the specific features is greater than a preset threshold value P can be judged by counting the number of the pictures containing the specific features, when the number of the pictures containing the specific features is judged to be greater than the preset threshold value P, the video is determined to be a specific content video, and otherwise, the video is judged to be a normal video. In another embodiment, the number of pictures containing the specific feature may be counted, a ratio of the number of the judged images containing the specific feature to the total number of the images detected and extracted for the video is determined, when the determined ratio is greater than a threshold Q, the video is determined to be a specific content video, otherwise, the video is determined to be a normal video.

For videos judged to be of a particular content, further processing may be performed, such as sorting, ranking, or quitting playback of the videos.

The method can be used for identifying pornographic videos, wherein the characteristic detection of the video images comprises sensitive part detection and skin color pixel detection.

One possible method of sensitive site detection includes:

and S11, searching the pre-stored characteristic data corresponding to the human sensitive part picture matched with the image to be identified in the human sensitive part index. The human sensitive part index can organize and store the characteristic data of the human sensitive part picture in order according to a certain mode, and is convenient to search. The human sensitive part picture can be obtained by marking the human sensitive part in the pornographic picture and generating a picture. The feature data may be a vector feature, which may be any feature in the existing image recognition method, such as description texture, HOG (Histogram of oriented gradients, image Gradient direction Histogram), LBP (Local Binary Patterns), or the like. The characteristic data of the image to be recognized can be extracted, and the distance between the characteristic data of the image to be recognized and the characteristic data of the human sensitive part picture is calculated, so that whether the image to be recognized is matched with the human sensitive part picture or not is judged according to the distance. For example, the distance may be represented by using a euclidean distance, and if the euclidean distance between the feature data of the image to be recognized and the feature data of one of the human sensitive portion pictures is the shortest, and the euclidean distance is smaller than a euclidean distance threshold, the image to be recognized and the human sensitive portion picture are matched. It is understood that whether the matching is performed may also be determined by other similarity measures, such as correlation coefficients, and so on, which are not listed here.

And S12, calculating the confidence corresponding to the image to be recognized according to the matched feature data. The confidence measure is a function of how well a decision matches the actual observation. The higher the confidence coefficient is, the higher the matching degree of the image to be recognized and the image of the sensitive part of the human body is. In one embodiment, the euclidean distance between the feature data of the image to be recognized and the matched feature data and the confidence are in a negative correlation relationship, and a function of the negative correlation may be used to represent the relationship therebetween, such as c ═ e-x, where x is the euclidean distance between the feature data of the image to be recognized and the matched feature data and c is the confidence.

And S13, judging whether the image to be recognized is a pornographic image or not according to the confidence corresponding to the image to be recognized. When the confidence coefficient is higher than the first confidence coefficient threshold value, the matching degree of the image to be recognized and the matched human sensitive part picture is high, and the image to be recognized is a pornographic image.

One achievable method of flesh tone pixel detection includes:

and S21, detecting the human body area pixels and the human head area pixels in the video image.

The human body detection generally adopts an Adaboost (an iterative algorithm) human body detection algorithm (of course, other algorithms can also be adopted), whether a human body exists in an image is judged by the Adaboost human body detection algorithm based on edge histogram features, an integral graph of a video image is firstly calculated, the edge histogram features are extracted, and a cascading method is operated to search a human body region in the image according to a set classifier feature library. The training method of the classifier feature library comprises the following steps: calculating an integral image of the sample image, and extracting the quasi-rectangular characteristics of the sample image; screening effective characteristics according to an Adaboost algorithm to form a weak classifier; forming a strong classifier by combining a plurality of weak classifiers; and cascading a plurality of strong classifiers to form a classifier feature library for human body detection. When the human body detection unit detects that a human body exists, the video image is detected again, and whether the human head exists is judged.

The human head detection adopts an Adaboost human head detection algorithm, whether a human head exists in an image is judged through the Adaboost human head detection algorithm based on the rectangle-like features, an integral graph of the image is firstly calculated, edge histogram features are extracted, and a cascade method is operated to search a human head area in the image according to a trained classifier feature library. The training method of the classifier feature library comprises the following steps: calculating an integral image of the sample image, and extracting the quasi-rectangular characteristics of the sample image; screening effective characteristics according to an Adaboost algorithm to form a weak classifier; forming a strong classifier by combining a plurality of weak classifiers; and cascading a plurality of strong classifiers to form a classifier feature library for human head detection.

And S22, counting the proportion of skin color pixels to image pixels, the proportion of skin color pixels to human body area pixels and the proportion of head area pixels to skin color pixels in each video image.

And S23, judging whether the video image is a pornographic image or not according to a preset first proportional threshold of the skin color pixel and the image pixel, a second proportional threshold of the skin color pixel and the human body area pixel, a third proportional threshold of the head area pixel and the skin color pixel and a preset judgment strategy.

Firstly, judging whether the ratio of the skin color pixels to the image pixels is greater than a first proportional threshold, whether the ratio of the skin color pixels to the human body area pixels is greater than a second proportional threshold, and whether the ratio of the human head area pixels to the skin color pixels is greater than a third proportional threshold, and respectively obtaining a first result, a second result and a third result; and then judging whether the first result, the second result and the third result meet the judgment strategy, if so, indicating that the skin color pixels of the video image accord with the characteristics of the pornographic image, and determining that the video image is the pornographic image. The judgment strategy can meet at least two conditions that the proportion of skin color pixels to image pixels is larger than a first proportion threshold value, the proportion of skin color pixels to human body area pixels is larger than a second proportion threshold value, and the proportion of head area pixels to skin color pixels is larger than a third proportion threshold value.

It should be noted that the foregoing method embodiments are described as a series of acts or combinations for simplicity in explanation, but it should be understood by those skilled in the art that the present invention is not limited by the order of acts or acts described, as some steps may occur in other orders or concurrently in accordance with the invention. Further, those skilled in the art should also appreciate that the embodiments described in the specification are preferred embodiments and that the acts and modules referred to are not necessarily required by the invention.

Through the above description of the embodiments, those skilled in the art can clearly understand that the method according to the above embodiments can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware, but the former is a better implementation mode in many cases. Based on such understanding, the technical solutions of the present invention may be embodied in the form of a software product, which is stored in a storage medium (e.g., ROM/RAM, magnetic disk, optical disk) and includes instructions for enabling a terminal device (e.g., a mobile phone, a computer, a server, or a network device) to execute the method according to the embodiments of the present invention.

Example 3

The embodiment provides a multimedia file processing method. In the operating environment as in embodiment 1, the present application provides a multimedia file processing method as shown in fig. 4. As shown in fig. 4, fig. 4 is a flowchart of a multimedia file processing method according to an embodiment of the present invention, and an alternative scheme of the method includes the following steps:

s401: in the multimedia file playing process, obtaining a portrait characteristic value and a first intention characteristic value of a viewer user, wherein the portrait characteristic value is used for identifying the preference of the user for specific content, and the first intention characteristic value is used for identifying the intention of the user wishing to watch the specific content within a preset time period;

s402: calculating the probability that the multimedia file contains specific content according to the portrait characteristic value and the first intention characteristic value;

s403: judging whether the probability exceeds a preset value, if so, performing feature detection on the multimedia file;

s404: judging whether the multimedia file is a multimedia file with specific content or not according to the feature detection result;

s405: and processing the multimedia file according to the judgment result.

According to the embodiment, the multimedia files are preliminarily screened through the user behavior characteristics, and then the screened multimedia files are subjected to specific content detection, so that the identification efficiency and accuracy of specific content are improved. The invention is used for pornographic video detection, can greatly improve the detection efficiency and reliability, and is convenient for the management and control of multimedia videos.

Example 4

The embodiment provides a multimedia file identification device based on behavior characteristics. As shown in fig. 5, the apparatus includes an acquisition unit 20, a calculation unit 30, a detection unit 40, and a determination unit 50.

An obtaining unit 20, configured to obtain, during a playing process of a multimedia file, a portrait feature value and a first intention feature value of a viewer user, where the portrait feature value is used to identify a preference of the user for a specific content, and the first intention feature value is used to identify an intention of the user that the user wishes to view the specific content within a preset time period;

a calculating unit 30, configured to calculate a probability that the multimedia file contains a specific content according to the portrait feature value and the first will feature value;

the detection unit 40 is configured to determine whether the probability exceeds a preset value, and if so, perform feature detection on the multimedia file;

and the determining unit 50 is configured to determine whether the multimedia file is a multimedia file with specific content according to the feature detection result.

In the behavior feature-based multimedia file identification apparatus of this embodiment, the obtaining unit 20 is configured to perform step S201 in embodiment 1 of the present invention, the calculating unit 30 is configured to perform step S202 in embodiment 1 of the present invention, the detecting unit 40 is configured to perform step S203 in embodiment 1 of the present invention, and the determining unit 50 is configured to perform step S204 in embodiment 1 of the present invention.

Referring to fig. 6, as an alternative embodiment, the calculation unit 30 includes:

a first calculating subunit 301, configured to determine a second intention characteristic value of each user according to the portrait characteristic value and the first intention characteristic value;

a second calculating subunit 302, configured to calculate, according to the second will characteristic values of all users, a probability that the multimedia file contains the specific content.

As an alternative implementation, the first computing subunit 301 includes:

a first calculating module 3011, configured to sum the portrait feature value and the first will feature value to obtain the second will feature value;

the second calculating module 3012 is configured to perform weighted average on the image feature value and the first intention feature value according to weights preset for the image feature value and the first intention feature value in advance, so as to obtain a second intention feature value.

Further, the second calculating subunit 302 includes:

a comparison module 3021, configured to compare the second intention characteristic value of each user with a preset threshold value respectively;

a probability calculating module 3022, configured to calculate a ratio of the number of users whose second will characteristic value exceeds a threshold to the total number of users, so as to obtain a probability that the multimedia file contains a specific content.

Further, the device further comprises a preprocessing unit 10, wherein the preprocessing unit 10 is used for analyzing the behavior data of the user and determining the portrait characteristic value and the first intention characteristic value of the user. The pre-processing unit 10 comprises a first pre-processing sub-unit 101 and a second pre-processing sub-unit 102.

A first processing subunit 101, configured to: acquiring behavior data of a user, wherein the behavior data comprises first behavior data for browsing a text related to specific content, second behavior data for browsing a picture related to the specific content, third behavior data for accessing a forum related to the specific content and fourth behavior data for chatting in a chat group related to the specific content; respectively judging whether the first behavior data, the second behavior data, the third behavior data and the fourth behavior data are empty or not, and correspondingly obtaining a first judgment result, a second judgment result, a third judgment result and a fourth judgment result; according to a preset first weight of the first judgment result, a preset second weight of the second judgment result, a preset third weight of the third judgment result and a preset fourth weight of the fourth judgment result, distributing and integrating the first judgment result, the second judgment result, the third judgment result and the fourth judgment result to obtain a behavior characteristic value of the user;

the second processing subunit 102 is configured to obtain behavior data of the user in a recent period of time, where the behavior data includes a first time for browsing a text related to the specific content, a second time for browsing a picture related to the specific content, a third time for accessing a forum related to the specific content, and a fourth time for chatting in a chat group related to the specific content; and giving the first weight to the first time, giving the second weight to the second time, giving the third weight to the third time, giving the fourth weight to the fourth time, and carrying out weighted average on the first time, the second time, the third time and the fourth time to obtain a user intention characteristic value.

As an optional implementation manner, the multimedia file is a video, and the detecting unit 40 includes:

an extracting subunit 401, configured to extract a preset number of images at equal time intervals of the video;

a detecting subunit 402, configured to perform feature detection on each image, and determine whether the image includes a specific feature, where the feature detection includes sensitive portion detection and skin color pixel detection.

As an alternative embodiment, the determining unit 50 includes:

a first determining subunit 501, configured to determine that the video is a video with a specific content when it is determined that the number of images with a specific feature is greater than a preset threshold P, and otherwise determine that the video is a normal video; or

A second determining subunit 502, configured to determine a ratio of the number of the determined images including the specific feature to the total number of the images extracted for the video detection, and determine that the video is a specific content video when the determined ratio is greater than a threshold Q, otherwise determine that the video is a normal video.

As a preferable mode of this embodiment, the specific content is a pornographic content, and the video is an on-demand video or a live video.

Example 5

The embodiment provides a multimedia file processing device. As shown in fig. 7, the apparatus includes an acquisition unit 20, a calculation unit 30, a detection unit 40, a determination unit 50, and a processing unit 60.

a determining unit 50, configured to determine whether the multimedia file is a multimedia file with specific content according to a feature detection result;

and the processing unit 60 is configured to process the multimedia file according to the determination result.

In the multimedia file processing apparatus of this embodiment, the obtaining unit 20 is configured to execute step S401 in embodiment 3 of the present invention, the calculating unit 30 is configured to execute step S402 in embodiment 3 of the present invention, the detecting unit 40 is configured to execute step S403 in embodiment 3 of the present invention, the determining unit 50 is configured to execute step S404 in embodiment 3 of the present invention, and the processing unit 60 is configured to execute step S405 in embodiment 3 of the present invention.

Preferably, the multimedia file is an on-demand video or a live video, and the specific content is pornographic content. The processing unit 60 is specifically configured to: when the multimedia file is determined to be the requested pornographic video, quitting the playing of the requested video; and when the multimedia file is determined to be the live pornographic video, closing the video live broadcast room for playing the video.

Example 6

The embodiment of the invention also provides a storage medium. Optionally, in this embodiment, the storage medium may be configured to store program codes executed by a behavior feature-based multimedia file identification method according to the above embodiment.

Optionally, in this embodiment, the storage medium may be located in at least one network device of a plurality of network devices of a computer network.

Optionally, in this embodiment, the storage medium is configured to store program code for performing the following steps:

the method comprises the steps that firstly, in the process of playing a multimedia file, a portrait characteristic value and a first intention characteristic value of a viewer user are obtained, wherein the portrait characteristic value is used for identifying the preference of the user for specific content, and the first intention characteristic value is used for identifying the intention of the user who wants to watch the specific content within a preset time period;

secondly, calculating the probability that the multimedia file contains specific content according to the portrait characteristic value and the first intention characteristic value;

thirdly, judging whether the probability exceeds a preset value, if so, performing feature detection on the multimedia file;

and fourthly, judging whether the multimedia file is the multimedia file with specific content or not according to the characteristic detection result.

Optionally, the storage medium is further arranged to store program code for performing the steps of: determining a second intention characteristic value of each user according to the portrait characteristic value and the first intention characteristic value; and calculating the probability that the multimedia file contains the specific content according to the second intention characteristic values of all the users.

Optionally, the storage medium is further arranged to store program code for performing the steps of: and summing the portrait characteristic value and the first intention characteristic value to obtain the second intention characteristic value, or carrying out weighted average on the portrait characteristic value and the first intention characteristic value according to weights preset for the portrait characteristic value and the first intention characteristic value to obtain the second intention characteristic value.

Optionally, the storage medium is further arranged to store program code for performing the steps of: respectively comparing the second intention characteristic value of each user with a preset threshold value; and calculating the ratio of the number of the users with the second will characteristic value exceeding a threshold value to the total number of the users to obtain the probability that the multimedia file contains the specific content.

Optionally, the storage medium is further arranged to store program code for performing the steps of: and analyzing the behavior data of the user, and determining the portrait characteristic value and the first intention characteristic value of the user.

Optionally, the storage medium is further arranged to store program code for performing the steps of: when the multimedia file is a video, extracting a preset number of images at equal time intervals from the video; and performing feature detection on each image, and judging whether the image contains a specific feature, wherein the feature detection comprises sensitive part detection and skin color pixel detection.

Optionally, the storage medium is further arranged to store program code for performing the steps of: when the number of the images containing the specific features is judged to be larger than a preset threshold value P, the video is judged to be a specific content video, otherwise, the video is judged to be a normal video; or, determining the ratio of the number of the judged images containing the specific features to the total number of the images obtained by detecting and extracting the video, and judging that the video is the specific content video when the determined ratio is greater than a threshold value Q, or judging that the video is the normal video.

Optionally, in this embodiment, the storage medium may include, but is not limited to: a U-disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a removable hard disk, a magnetic or optical disk, and other various media capable of storing program codes.

Example 7

The embodiment of the invention also provides a storage medium. Alternatively, in this embodiment, the storage medium may be configured to store program codes executed by a video processing method of the foregoing embodiment.

fourthly, judging whether the multimedia file is a multimedia file with specific content or not according to the characteristic detection result;

and fifthly, processing the multimedia file according to the judgment result.

Optionally, the storage medium is further arranged to store program code for performing the steps of: when the multimedia file is the requested pornographic video, quitting the playing of the requested video; and when the multimedia file is the live pornographic video, closing the video live broadcast room for playing the video.

Example 8

The embodiment of the invention also provides a computer terminal, which can be any computer terminal device in a computer terminal group. Optionally, in this embodiment, the computer terminal may also be replaced with a terminal device such as a mobile terminal.

Optionally, in this embodiment, the computer terminal may be located in at least one network device of a plurality of network devices of a computer network.

Alternatively, fig. 8 is a block diagram of a structure of a computer terminal according to an embodiment of the present invention. As shown in fig. 8, the computer terminal a may include: one or more processors 801 (only one of which is shown), a memory 803, and a transmission device 805.

The memory 803 may be used to store software programs and modules, such as program instructions/modules corresponding to the multimedia file identification method and apparatus based on behavior characteristics in the embodiment of the present invention, and the processor 801 executes various functional applications and data processing by running the software programs and modules stored in the memory 803, so as to implement the above-mentioned multimedia file identification method. The memory 803 may include high-speed random access memory, and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 803 may further include memory located remotely from the processor 801, which may be connected to the computer terminal a via a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.

The transmission device 805 is used for receiving or transmitting data via a network. Examples of the network may include a wired network and a wireless network. In one example, the transmission device 805 includes a network adapter that can be connected to a router via a network cable to communicate with the internet or a local area network. In one example, the transmission device 805 is a radio frequency module, which is used for communicating with the internet in a wireless manner.

Specifically, the memory 803 is used for storing preset action conditions, information of preset authorized users, and application programs.

The processor 801 may call the information and applications stored in the memory 803 via the transmission device to perform the following steps:

For specific examples in this embodiment, reference may be made to the examples described in embodiment 1 and embodiment 2, which are not described herein again.

The above-mentioned serial numbers of the embodiments of the present invention are merely for description and do not represent the merits of the embodiments.

The integrated unit in the above embodiments, if implemented in the form of a software functional unit and sold or used as a separate product, may be stored in the above computer-readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for causing one or more computer devices (which may be personal computers, servers, network devices, etc.) to execute all or part of the steps of the method according to the embodiments of the present invention.

In the above embodiments of the present invention, the descriptions of the respective embodiments have respective emphasis, and for parts that are not described in detail in a certain embodiment, reference may be made to related descriptions of other embodiments.

In the several embodiments provided in the present application, it should be understood that the disclosed client may be implemented in other manners. The above-described embodiments of the apparatus are merely illustrative, and for example, the division of the units is only one type of division of logical functions, and there may be other divisions when actually implemented, for example, a plurality of units or components may be combined or may be integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, units or modules, and may be in an electrical or other form.

The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.

In addition, functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit. The integrated unit can be realized in a form of hardware, and can also be realized in a form of a software functional unit.

Claims

1. A multimedia file identification method based on behavior characteristics is characterized by comprising the following steps:

2. The method of claim 1, wherein calculating the probability that the multimedia file contains the specific content according to the portrait characteristic value and the first will characteristic value comprises:

3. The method of claim 2, wherein determining a second intent feature value for each user based on the portrait feature value and the first intent feature value comprises:

summing the portrait feature value and the first intent feature value to obtain the second intent feature value, or,

and carrying out weighted average on the image characteristic value and the first intention characteristic value according to weights preset for the image characteristic value and the first intention characteristic value to obtain a second intention characteristic value.

4. The method of claim 2, wherein the calculating the probability that the multimedia file contains the specific content according to the second will characteristic values of all users comprises:

respectively comparing the second intention characteristic value of each user with a preset threshold value;

and calculating the ratio of the number of the users with the second will characteristic value exceeding a threshold value to the total number of the users to obtain the probability that the multimedia file contains the specific content.

5. The method of claim 1, further comprising, prior to obtaining the representation characteristic value and the first intent characteristic value of the viewer user:

and analyzing the behavior data of the user, and determining the portrait characteristic value and the first intention characteristic value of the user.

6. The method of claim 5, wherein analyzing the behavioral data of the user to determine a portrait characteristic value of the user comprises:

acquiring behavior data of a user, wherein the behavior data comprises first behavior data for browsing a text related to specific content, second behavior data for browsing a picture related to the specific content, third behavior data for accessing a forum related to the specific content and fourth behavior data for chatting in a chat group related to the specific content;

respectively judging whether the first behavior data, the second behavior data, the third behavior data and the fourth behavior data are empty or not, and correspondingly obtaining a first judgment result, a second judgment result, a third judgment result and a fourth judgment result;

and according to a preset first weight of the first judgment result, a preset second weight of the second judgment result, a preset third weight of the third judgment result and a preset fourth weight of the fourth judgment result, distributing and integrating the first judgment result, the second judgment result, the third judgment result and the fourth judgment result to obtain the behavior characteristic value of the user.

7. The method of claim 6, wherein analyzing the behavior data of the user to determine a first willingness characteristic value of the user comprises:

acquiring behavior data of a user in a recent period of time, wherein the behavior data comprises a first time for browsing a text related to specific content, a second time for browsing a picture related to the specific content, a third time for accessing a forum related to the specific content and a fourth time for chatting in a chat group related to the specific content;

and giving the first weight to the first time, giving the second weight to the second time, giving the third weight to the third time, giving the fourth weight to the fourth time, and carrying out weighted average on the first time, the second time, the third time and the fourth time to obtain a user intention characteristic value.

8. The method of claim 1, wherein the multimedia file is a video;

the performing feature detection on the multimedia file comprises:

extracting a preset number of images at equal time intervals of a video;

and performing feature detection on each image, and judging whether the image contains a specific feature, wherein the feature detection comprises sensitive part detection and skin color pixel detection.

9. The method of claim 8, wherein the determining whether the multimedia file is a multimedia file with specific content according to the feature detection result comprises:

when the number of the images containing the specific features is judged to be larger than a preset threshold value P, the video is judged to be a specific content video, otherwise, the video is judged to be a normal video; or

And determining the ratio of the number of the judged images containing the specific features to the total number of the images extracted by the video detection, judging the video to be a specific content video when the determined ratio is greater than a threshold value Q, and otherwise, judging the video to be a normal video.

10. The method according to claim 8 or 9, wherein the specific content is pornographic content, and the video is video on demand or live.

11. A method for processing a multimedia file, comprising:

and processing the multimedia file according to the judgment result.

12. The method of claim 11, wherein the multimedia file is an on-demand video or a live video, and the specific content is a pornographic content;

the processing the multimedia file according to the judgment result comprises the following steps: if the multimedia file is the requested pornographic video, quitting the playing of the requested video; and if the multimedia file is the live pornographic video, closing the video live broadcast room for playing the video.

13. A multimedia file recognition apparatus based on behavior characteristics, comprising:

14. The apparatus of claim 13, wherein the computing unit comprises:

the first calculation subunit is used for determining a second intention characteristic value of each user according to the portrait characteristic value and the first intention characteristic value;

and the second calculating subunit is used for calculating the probability that the multimedia file contains the specific content according to the second intention characteristic values of all the users.

15. The apparatus of claim 14, wherein the first computing subunit comprises:

the first calculation module is used for summing the portrait characteristic value and the first intention characteristic value to obtain a second intention characteristic value;

and the second calculation module is used for carrying out weighted average on the portrait characteristic value and the first intention characteristic value according to weights preset for the portrait characteristic value and the first intention characteristic value to obtain a second intention characteristic value.

16. The apparatus of claim 14, wherein the second computing subunit comprises:

the comparison module is used for comparing the second intention characteristic value of each user with a preset threshold value respectively;

and the probability calculation module is used for calculating the ratio of the number of the users with the second will characteristic value exceeding a threshold value to the total number of the users to obtain the probability that the multimedia file contains the specific content.

17. The apparatus of claim 13, further comprising:

and the preprocessing unit is used for analyzing the behavior data of the user and determining the portrait characteristic value and the first intention characteristic value of the user.

18. The apparatus of claim 17, wherein the pre-processing unit comprises:

19. The apparatus of claim 13, wherein the multimedia file is a video;

the detection unit includes:

the extraction subunit is used for extracting a preset number of images at equal time intervals of the video;

and the detection subunit is used for performing feature detection on each image and judging whether the image contains a specific feature, wherein the feature detection comprises sensitive part detection and skin color pixel detection.

20. The apparatus of claim 19, wherein the determining unit comprises:

the first determining subunit is used for determining that the video is a video with specific content when the number of the images containing the specific features is judged to be larger than a preset threshold value P, and otherwise, determining that the video is a normal video;

and the second determining subunit is used for determining the ratio of the number of the judged images containing the specific features to the total number of the images obtained by detection and extraction aiming at the video, judging the video to be the specific content video when the determined ratio is greater than a threshold value Q, and otherwise, judging the video to be the normal video.

21. The apparatus according to claim 19 or 20, wherein the specific content is pornographic content, and the video is video on demand or live.

22. A multimedia file processing apparatus, characterized in that the apparatus comprises:

23. The apparatus of claim 22, wherein the multimedia file is an on-demand video or a live video, and the specific content is a pornographic content;

the processing unit is specifically used for quitting playing of the on-demand video when the multimedia file is determined to be the on-demand pornographic video; and when the multimedia file is determined to be the live pornographic video, closing the video live broadcast room for playing the video.

24. A storage medium, characterized in that the storage medium is configured to store program code for executing the behavior feature based multimedia file recognition method according to any one of claims 1 to 10.