Disclosure of Invention
One of the embodiments of the present disclosure provides a wind control method based on AI digital work copyrights, including the steps of: acquiring a historical AI digital image and a current AI digital image, wherein classification labels are preset on the historical AI digital image and the current AI digital image; extracting a plurality of identification features of each historical AI digital image, and extracting a plurality of identification features of the current AI digital image; dividing all the historical AI digital images based on the classification labels to obtain a plurality of digital work sets; determining the stability of each identification feature corresponding to each digital work set, and determining the weight of each identification feature corresponding to the digital work set based on the stability; determining a digital work set to which the current AI digital image belongs based on the classification tag; based on the identification features and the weight of each identification feature, a historical AI digital work with the largest similarity with the current AI digital image and the largest similarity are determined, and based on the largest similarity, an infringement risk is determined. In an embodiment of the present application, extracting a plurality of identification features of each of the historical AI digital images and extracting a plurality of identification features of the current AI digital image includes: and extracting the color features, the outline features and the straight line features of the historical AI digital image, and extracting the color features, the outline features and the straight line features of the current AI digital image.
In some embodiments, the method further comprises extracting color features of the historical AI digital image or the current AI digital image by: determining the color type of each pixel point in the historical AI digital image or the current AI digital image, wherein the color type is determined by comparing the color channel value of the pixel point with the preset color channel value range of a plurality of color types; and counting the color types of all pixel points in the historical AI digital image or the current AI digital image to obtain N target color types before the occurrence frequency of the historical AI digital image or the current AI digital image.
In some embodiments, the method further comprises extracting straight line features of the historical AI digital image or the current AI digital image by: extracting all straight lines in the historical AI digital image or the current AI digital image; and counting the number of straight lines in the historical AI digital image or the current AI digital image and the number of intersection points of the straight lines.
In some embodiments, determining the stability of each identifying feature corresponding to each set of digital works includes: determining a reference digital image in each digital work set, and determining reference color features, reference contour features and reference straight line features of the reference digital image; for each historical AI digital work in each digital work set, comparing the color characteristics with the reference color characteristics to obtain color similarityComparing the profile features with the reference profile features to obtain profile similarity +.>And the straight line characteristic is matched with the reference straight line characteristicThe ratio is used for obtaining the linear similarityThe method comprises the steps of carrying out a first treatment on the surface of the Calculating a first variance of color similarity for each set of digital works>Second variance of contour similarity +.>And a third variance of the linear similarity +.>The method comprises the steps of carrying out a first treatment on the surface of the Based on the first variance->Determining the stability of the color feature based on said second variance +.>Determining the stability of the profile feature and based on said third variance +.>The stability of the straight line feature is determined.
In some embodiments, the method further comprises determining a color similarity between any two digital worksProfile similarity->And straight line similarity->: determining the number of repeating target color types between any two digital worksAnd based on the number of target color types +.>And the number of repeating target color types +.>Calculate color similarity +.>Color similarity->The mathematical expression of (2) is: />Centering of outline features of any two digital worksOverlapping the contour features of the arbitrary two digital works based on the respective centers, rotating and scaling to maximize the number of overlapping pixels of the two contours, and based on the number of overlapping pixels->And the number of pixels of the target contour feature +.>Calculating contour similarity +.>Wherein the target contour feature is one with more pixel points in the contour features of any two digital works, and the contour similarity is +.>The mathematical expression of (2) is: />Based on the number of lines +.>And the number of intersections of straight lines +.>Determining the intersection characteristics of the profile features of any two digital works>,/>,/>When 0, the drug is added>Taking 0 and determining the linear similarity based on the intersection characteristics of the profile features of any two digital works>Straight line similarity->The mathematical expression of (2) is: />, wherein ,/> and />Intersection characteristics of profile features of two digital works, respectively +.>Is the one with the larger value +.>When 0, the drug is added>Taking 0.
In some embodiments, determining the historical AI digital work with the greatest similarity to the current AI digital image, and the greatest similarity, includes: determining the current AI digital image and each historical AI digital image in the set of affiliated digital worksColor similarity of (C)Profile similarity->And straight line similarity->The method comprises the steps of carrying out a first treatment on the surface of the Color similarity->Profile similarity->And straight line similarity->Weighted summation is carried out to obtain the similarity of the current AI digital image and each historical AI digital image in the affiliated digital work set>Similarity->The mathematical expression of (2) is:determining maximum similarity ∈ ->Maximum similarity ∈>Corresponding historical AI digital works.
One of the embodiments of the present specification provides an air control system based on AI digital work copyrights, including: the acquisition module is used for acquiring a historical AI digital image and a current AI digital image, wherein classification labels are preset on the historical AI digital image and the current AI digital image; the characteristic extraction module is used for extracting various identification characteristics of each historical AI digital image and extracting various identification characteristics of the current AI digital image; the dividing module is used for dividing all the historical AI digital images based on the classification labels to obtain a plurality of digital work sets; the weight determining module is used for determining the stability of each identification feature corresponding to each digital work set and determining the weight of each identification feature corresponding to the digital work set based on the stability; the attribution determining module is used for determining an attribution digital work collection of the current AI digital image based on the classification label; and the risk determining module is used for determining historical AI digital works with the maximum similarity with the current AI digital image and the maximum similarity based on the identification characteristics and the weight of each identification characteristic, and determining infringement risk based on the maximum similarity.
The application also provides a computer readable storage medium having stored thereon a computer program which when executed by a processor implements a method as described in any of the above.
One of the embodiments of the present specification provides an electronic terminal, including: a processor and a memory; the memory is configured to store a computer program, and the processor is configured to execute the computer program stored in the memory, so as to cause the terminal to perform the method according to any one of the above.
Compared with the prior art, the intelligent book recommendation method based on data processing for books has the following beneficial effects:
and classifying the historical AI digital images to obtain a plurality of identification characteristics of each type of the historical AI digital images. And then carrying out stability analysis on the identification characteristics of each type of historical AI digital image, and if one of the identification characteristics of the type of historical AI digital image is relatively stable, indicating that the characteristic is a common characteristic of the corresponding type of digital image. Thus, in the subsequent comparison of similarity, it is necessary to compare such features with emphasis. Therefore, the various identification features are weighted based on the stability of the identification features, so that the more stable and important features are weighted more in the subsequent comparison, and the obtained similarity comparison result is more scientific and accurate.
Detailed Description
In order to more clearly illustrate the technical solutions of the embodiments of the present specification, the drawings that are required to be used in the description of the embodiments will be briefly described below.
As shown in fig. 1, the present application provides a wind control method based on AI digital work copyright, comprising the steps of:
s110, acquiring a historical AI digital image and a current AI digital image, wherein classification labels are preset on the historical AI digital image and the current AI digital image;
the classification labels identify the historical AI digital images through a pre-established image identification model to obtain identification labels, and the identification labels are used as the classification labels, so that a large number of historical AI digital images can be marked efficiently. The current AI digital image is the AI digital image which needs to be subjected to copyright risk analysis at the current time.
S120, extracting various identification features of each historical AI digital image, and extracting various identification features of the current AI digital image;
the identification feature may be any feature of the AI digital image, and in this embodiment, color features, contour features, and straight line features are extracted based on the features of most AI digital images at present. The color features are aimed at pictures of types such as art wallpaper and scenery with rich colors, the outline features are pictures with sharp outlines for figures, objects and the like, and the straight line features are pictures with a large number of straight lines for buildings, rooms and the like.
Specifically, in an embodiment of the present application, the method further includes the following steps of extracting color features of the historical AI digital image or the current AI digital image:
determining the color type of each pixel point in a historical AI digital image or a current AI digital image, wherein the color type is determined by comparing the color channel value of the pixel point with the preset color channel value range of a plurality of color types;
and counting the color types of all pixel points in the historical AI digital image or the current AI digital image to obtain N target color types before the occurrence frequency of the historical AI digital image or the current AI digital image.
In the present embodiment, the color feature is determined by counting the color type of each pixel in the history AI digital image or the current AI digital image. Where each color type corresponds to a range of color channels, for example, in the RGB color model, the range of color channels for one blue (blue # 3) is { (130-140), (200-210), (230-240) }, and thus any pixel point where an RGB value falls into the range is defined as "blue #3". According to the above process, the complex color channel information is simplified, and the N target color types before the occurrence frequency of the historical AI digital image or the current AI digital image are reserved, so that the information processing efficiency is improved while the important information is reserved.
In one embodiment of the present application, the method further comprises the following steps of extracting straight line characteristics of the historical AI digital image or the current AI digital image:
extracting all straight lines in a historical AI digital image or a current AI digital image;
the number of straight lines in the historical AI digital image or the current AI digital image, and the number of intersections of the straight lines are counted.
The extraction of the straight line can use the existing extraction function, such as HoughLines () function. The extraction of the contours may then use an edge extraction operator, such as a canny operator.
The number of extracted straight lines may be 0 to n, and if the number of straight lines is plural, the state of the plural straight lines is judged by extracting the number of intersections of the straight lines. Of these, there are at most (n (n-1))/2 intersections of already n straight lines. Within this range, the greater the number of intersections, the greater the intersection of the lines, i.e., the bias toward an included angle rather than parallel. Therefore, the straight line shape of the AI digital image is roughly judged by keeping the number of straight lines and the number of intersection points of the straight lines, so that the information is simplified, and meanwhile, important judging properties are kept.
S130, dividing all historical AI digital images based on the classification labels to obtain a plurality of digital work sets;
s140, determining the stability of each identification feature corresponding to each digital work set, and determining the weight of each identification feature corresponding to the digital work set based on the stability;
wherein, since the historical AI digital images are classified in advance, the different classes of historical AI digital images should have more stable one or more identification features. Taking the example of a historical AI digital image of a building class, the historical AI digital image within the building contains a large number of straight line features and has a sharp geometric outline. The straight line features and the contour features are therefore stable for the historical AI digital images of the building class.
In one embodiment of the application, determining the stability of each identification feature corresponding to each set of digital works comprises:
determining a reference digital image in each digital work set, and determining a reference color feature, a reference contour feature and a reference straight line feature of the reference digital image;
for each historical AI digital work in each digital work set, comparing the color characteristics with the reference color characteristics to obtain color similarityComparing the profile features with the reference profile features to obtain profile similarity +.>Comparing the straight line characteristic with the reference straight line characteristic to obtain straight line similarity +.>。
Calculating a first variance of color similarity for each set of digital worksSecond variance of contour similarity +.>And a third variance of the linear similarity +.>;
Based on the first differenceDetermining the stability of the color feature based on the second variance +.>Determining the stability of the profile feature and based on a third variance +.>The stability of the straight line feature is determined.
In this embodiment, by selecting one reference digital image as the reference image for the corresponding digital work set, then comparing all the historical AI digital images with the reference image. The degree of difference obtained can be used to account for its stability. Further in this embodiment, the stability of the identification feature is described using the variance.
In one embodiment of the application, the method further comprises the following steps of determining the color similarity between any two digital worksProfile similarity->And straight line similarity->:
Determining the number of repeating target color types between any two digital worksAnd based on the number of target color types +.>And the number of repeating target color types +.>Calculate color similarity +.>Color similarity->The mathematical expression of (2) is:
the color difference of the two AI digital images is determined by comparing the target color types between the different AI digital images, and the more the repeated target color types exist, the higher the color similarity is.
Centering of outline features of any two digital worksOverlapping the contour features of any two digital works based on the respective centers, and rotating and scaling to maximize the number of overlapping pixels of the two contours, and based on the number of overlapping pixels->And the number of pixels of the target contour feature +.>Calculating contour similarity +.>Wherein the target contour feature is one with more pixel points in the contour features of any two digital works, and the contour similarity is +.>The mathematical expression of (2) is:
since the profile is generally complex, in this embodiment, the center is determined by determining the midpoint of the profile feature. wherein ,/>,/>,/>For the abscissa of the contour feature pixel point of any two digital works, +.>Is the ordinate of the outline characteristic pixel point of any two digital works, +.>The number of contour feature pixel points of any two digital works;
and then, selecting and scaling after overlapping based on the central position, so that the two contour features are fully contacted, and finally, counting the number of overlapped pixels. The more overlapping pixels exist, the higher the profile similarity is explained.
Based on the number of straight linesAnd the number of intersections of straight lines +.>Determining the intersection characteristics of the profile features of any two digital works>,/>,/>When 0, the drug is added>Taking 0 and determining the linear similarity +.>Straight line similarity->The mathematical expression of (2) is:
wherein , and />Intersection characteristics of profile features of two digital works, respectively +.>Is the one with the larger value +.>When 0, the drug is added>Taking 0. Linear similarity is mainly used to compare the morphology of the lines (whether approaching parallelism or intersection).
S150, determining a digital work set to which the current AI digital image belongs based on the classification label;
s160, based on the identification features and the weight of each identification feature, determining the historical AI digital works with the maximum similarity with the current AI digital image and the maximum similarity, and determining the infringement risk based on the maximum similarity.
In one embodiment of the present application, determining a historical AI digital work having a greatest similarity to a current AI digital image, and the greatest similarity, comprises:
determining the color similarity between the current AI digital image and each of the historical AI digital images in the belonging digital work setProfile similarity->And straight line similarity->;
For color similarityProfile similarity->And straight line similarity->Weighted summation is carried out to obtain the similarity +.A of the current AI digital image and each historical AI digital image in the affiliated digital work set>Similarity->The mathematical expression of (2) is:
determining maximum similarityMaximum similarity ∈>Corresponding historical AI digital works.
Wherein the weight of the color features isThe weight of the contour features is +.>The weight of the straight line feature is +.>The three are added to be 1, so that normalization processing is realized. Meanwhile, the smaller the variance, the more stable the description, and the larger the corresponding weight. Therefore, the present embodiment realizes determination of the importance degree of the features of the AI digital image, and determines weights according to the respective importance degrees, and increases the duty ratio of the important features when determining the infringement risk. Thereby making the final analysis result more accurate.
In addition, when the infringement risk is determined based on the maximum similarity, a similarity threshold is set, and the maximum similarity is larger than the similarity threshold, so that the infringement risk is determined. Otherwise, judging that the infringement risk is not existed.
The application provides an air control method based on AI digital work copyright, which obtains various identification characteristics of each type of history AI digital image by classifying the history AI digital image. And then carrying out stability analysis on the identification characteristics of each type of historical AI digital image, and if one of the identification characteristics of the type of historical AI digital image is relatively stable, indicating that the characteristic is a common characteristic of the corresponding type of digital image. Thus, in the subsequent comparison of similarity, it is necessary to compare such features with emphasis. Therefore, the various identification features are weighted based on the stability of the identification features, so that the more stable and important features are weighted more in the subsequent comparison, and the obtained similarity comparison result is more scientific and accurate.
As shown in fig. 2, the present application further provides an air control system based on AI digital work copyrights, including:
the acquisition module is used for acquiring a historical AI digital image and a current AI digital image, wherein classification labels are preset on the historical AI digital image and the current AI digital image;
the characteristic extraction module is used for extracting various identification characteristics of each historical AI digital image and extracting various identification characteristics of the current AI digital image;
the dividing module is used for dividing all the historical AI digital images based on the classification labels to obtain a plurality of digital work sets;
the weight determining module is used for determining the stability of each identification feature corresponding to each digital work set and determining the weight of each identification feature corresponding to the digital work set based on the stability;
the attribution determining module is used for determining an attribution digital work set of the current AI digital image based on the classification label;
and the risk determining module is used for determining historical AI digital works with the maximum similarity with the current AI digital image and the maximum similarity based on the identification characteristics and the weight of each identification characteristic, and determining infringement risk based on the maximum similarity.
The application provides an air control system based on AI digital work copyright, which obtains various identification characteristics of each type of history AI digital image by classifying the history AI digital image. And then carrying out stability analysis on the identification characteristics of each type of historical AI digital image, and if one of the identification characteristics of the type of historical AI digital image is relatively stable, indicating that the characteristic is a common characteristic of the corresponding type of digital image. Thus, in the subsequent comparison of similarity, it is necessary to compare such features with emphasis. Therefore, the various identification features are weighted based on the stability of the identification features, so that the more stable and important features are weighted more in the subsequent comparison, and the obtained similarity comparison result is more scientific and accurate.
It should be noted that, the wind control system based on the AI digital work copyright provided by the foregoing embodiment and the wind control method based on the AI digital work copyright provided by the foregoing embodiment belong to the same concept, and the specific manner in which each module and unit perform the operation has been described in detail in the method embodiment, which is not repeated herein. In practical application, the air control system based on AI digital work copyright provided by the above embodiment can distribute the functions to be completed by different functional modules according to needs, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above, which is not limited herein.
Fig. 3 is a block diagram of an electronic device, as shown in fig. 3, which is an example of a hardware device that may be applied to aspects of the present application, according to some embodiments of the present description. Electronic devices are intended to represent various forms of digital electronic computer devices, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other suitable computers. The electronic device may also represent various forms of mobile devices, such as personal digital processing, cellular telephones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are meant to be exemplary only, and are not meant to limit implementations of the applications described and/or claimed herein.
As shown in fig. 3, the electronic device includes a computing unit that can perform various appropriate actions and processes according to a computer program stored in a Read Only Memory (ROM) or a computer program loaded from a storage unit into a Random Access Memory (RAM). In the RAM, various programs and data required for the operation of the device may also be stored. The computing unit, ROM and RAM are connected to each other by a bus. An input/output (I/O) interface is also connected to the bus.
A plurality of components in an electronic device are connected to an I/O interface, comprising: an input unit, an output unit, a storage unit, and a communication unit. The input unit may be any type of device capable of inputting information to the electronic device, and may receive input numeric or character information and generate key signal inputs related to user settings and/or function controls of the electronic device. The output unit may be any type of device capable of presenting information and may include, but is not limited to, a display, speakers, video/audio output terminals, vibrators, and/or printers. Storage units may include, but are not limited to, magnetic disks, optical disks. The communication unit allows the electronic device to exchange information/data with other devices via a computer network, such as the internet, and/or various telecommunications networks, and may include, but is not limited to, modems, network cards, infrared communication devices, wireless communication transceivers and/or chipsets, such as bluetooth (TM) devices, wiFi devices, wiMax devices, cellular communication devices, and/or the like.
The computing unit may be a variety of general and/or special purpose processing components having processing and computing capabilities. Some examples of computing units include, but are not limited to, central Processing Units (CPUs), graphics Processing Units (GPUs), various specialized Artificial Intelligence (AI) computing chips, various computing units running machine learning model algorithms, digital Signal Processors (DSPs), and any suitable processors, controllers, microcontrollers, and the like. The computing unit performs the various methods and processes described above. For example, in some embodiments, the AI-based digital work copyright management method may be implemented as a computer software program tangibly embodied on a machine-readable medium, such as a storage unit. In some embodiments, part or all of the computer program may be loaded and/or installed onto the electronic device via the ROM and/or the communication unit. In some embodiments, the computing unit may be configured to perform the AI-digital-work-copyright-based wind control method in any other suitable manner (e.g., by means of firmware).
Finally, it should be understood that the embodiments described in this specification are merely illustrative of the principles of the embodiments of this specification. Other variations are possible within the scope of this description. Thus, by way of example, and not limitation, alternative configurations of embodiments of the present specification may be considered as consistent with the teachings of the present specification. Accordingly, the embodiments of the present specification are not limited to only the embodiments explicitly described and depicted in the present specification.