Disclosure of Invention
Based on the above, it is necessary to provide a data storage management method and system based on artificial intelligence, which can delete the intelligent similar pictures and further realize efficient storage management.
The technical scheme of the invention is as follows:
an artificial intelligence based data storage management method for data storage management of pictures, the method comprising:
acquiring an intelligent picture data management instruction set by a current picture data owner body on stored current summarized picture data, and acquiring current management picture data according to the intelligent picture data management instruction, wherein the current management picture data is a part of the current summarized picture data, and the current management picture data comprises a plurality of current actual pictures to be managed;
performing similarity calculation on each current actual picture to be managed according to a preset intelligent picture screening algorithm, and obtaining texture feature similarity values between the current actual pictures to be managed;
Selecting actual similar pictures belonging to the same main body from the current management picture data according to the texture feature similarity value, and recording the actual similar pictures as a current similar picture group, wherein the actual similar pictures belonging to the same main body in the current management picture data are multiple, and the actual similar pictures are summarized to be the current similar picture group;
obtaining current actual definition of each actual similar picture in the current similar picture group, removing each actual similar picture according to each current actual definition, generating initial removed picture data after removing, obtaining correction of the initial removed picture data by the current picture data owner body, generating final screened picture data after correcting, and storing the final screened picture data.
Specifically, similarity calculation is carried out on each current actual picture to be managed according to a preset intelligent picture screening algorithm, and texture feature similarity values among the current actual pictures to be managed are obtained; the method specifically comprises the following steps:
generating current actual gray-scale pictures according to each current actual picture to be managed, wherein one current actual picture to be managed corresponds to one current actual gray-scale picture;
Generating a current spectrogram according to each current actual gray picture, and generating a current actual mask image according to each current spectrogram;
generating a current spectrum mask image according to the current actual mask image and the corresponding current spectrum image;
generating a current airspace picture according to the current spectrum mask map;
and respectively generating texture feature similarity values between the two current airspace pictures according to the current airspace pictures.
Specifically, in the step of generating the texture feature similarity values between the two current spatial pictures, the texture feature similarity values between the two current spatial pictures are generated by the following formula:
wherein G is pq Representing a texture feature similarity value between the p-th airspace picture and the q-th airspace picture; F1F 1 p Gradient value frequency vector representing p-th spatial picture, F1 q A gradient value frequency vector representing the q-th airspace picture; F2F 2 p Representing the gradient direction appearance frequency representation vector of the p-th airspace picture, F2 q Representing a gradient direction appearance frequency representation vector of the q-th airspace picture;<F1 p ,F1 q >the cosine similarity of the gradient value frequency vector of the p-th airspace picture and the gradient value frequency vector of the q-th airspace picture is represented; <F2 p ,F2 q >The cosine similarity of the gradient direction appearance frequency representation vector of the p-th airspace picture and the gradient direction appearance frequency representation vector of the q-th airspace picture is represented;<F1 p ,F1 q > 2 representing a distribution characteristic similarity metric value between gradient value distribution of an airspace picture corresponding to the p-th spectrum mask image and gradient value distribution of an airspace picture corresponding to the q-th spectrum mask image;<F2 p ,F2 q > 2 and representing a distribution characteristic similarity metric value between the gradient direction angle distribution of the airspace picture corresponding to the p-th spectrum mask diagram and the gradient direction angle distribution of the airspace picture corresponding to the q-th spectrum mask diagram.
Specifically, selecting actual similar pictures belonging to the same main body from the current management picture data according to the texture feature similarity value, and recording the actual similar pictures as a current similar picture group, wherein the actual similar pictures belonging to the same main body in the current management picture data are multiple, and the actual similar pictures are summarized to form the current similar picture group; the method specifically comprises the following steps:
obtaining a pre-stored standard similarity value according to the texture feature similarity value;
screening out a current actual picture to be managed corresponding to a texture feature similarity value which is greater than or equal to the standard similarity value, and setting the current actual picture to be managed as an initial similar picture;
Acquiring screening and rechecking data of the initial similar pictures according to the initial similar pictures;
and screening the initial similar pictures according to the screening and rechecking data, and generating actual similar pictures, wherein the actual similar pictures are summarized to obtain the current similar picture group.
Specifically, obtaining current actual definition of each actual similar picture in the current similar picture group, removing each actual similar picture according to each current actual definition, generating initial removed picture data after removing, obtaining correction of the initial removed picture data by the current picture data owner body, generating final screened picture data after correcting, and storing the final screened picture data, wherein the method specifically comprises the following steps:
the current actual sharpness of each of the actual similar pictures is calculated separately based on the following formula,
wherein Y is et Representing the current actual sharpness, alpha (beta) ety ) Representing the frequency of the occurrence of a blurred region in the y-th gradient direction in t detection windows in the e-type detection region, ln () representing the natural logarithmic function, beta ety Representing the y-th gradient direction in t detection windows in the e-type detection region; h represents the number of blurred regions of t in the detection region of the e type;
Screening out an actual similar picture corresponding to the current actual definition which is smaller than or equal to the preset standard definition, and eliminating;
generating initial reject picture data after rejection is completed, acquiring correction of the initial reject picture data by the current picture data owner body, generating final screening picture data after correction is completed, and storing the final screening picture data.
Specifically, a current picture data owner body acquires an intelligent picture data management instruction set in stored current summarized picture data, and acquires current management picture data according to the intelligent picture data management instruction, wherein the current management picture data is a part of the current summarized picture data, the current management picture data comprises a plurality of current actual pictures to be managed, and the method specifically comprises the following steps:
acquiring important picture data marked by the current picture data possession main body from stored current summarized picture data, and generating initial demarcation picture data;
acquiring an initial selected area range of the current picture data owner body for the initial demarcation picture data;
generating a current demarcation correction area according to the initial selected area range;
Generating an intelligent management instruction of the picture data according to the current defined correction area;
and generating current management picture data according to the picture data intelligent management instruction.
Specifically, an artificial intelligence based data storage management system, the system comprising:
the picture management demarcation module is used for acquiring an intelligent picture data management instruction set by a current picture data owner body in stored current summarized picture data, and acquiring current management picture data according to the intelligent picture data management instruction, wherein the current management picture data is a part of the current summarized picture data, and the current management picture data comprises a plurality of current actual pictures to be managed;
the similarity value calculation module is used for calculating the similarity of each current actual picture to be managed according to a preset intelligent picture screening algorithm and obtaining the similarity value of the texture characteristics among the current actual pictures to be managed;
the similar picture eliminating module is used for selecting actual similar pictures belonging to the same main body from the current management picture data according to the texture feature similarity value and marking the actual similar pictures as a current similar picture group, wherein the actual similar pictures belonging to the same main body in the current management picture data are multiple, and the actual similar pictures are summarized to form the current similar picture group;
The intelligent picture management module is used for acquiring the current actual definition of each actual similar picture in the current similar picture group, rejecting each actual similar picture according to each current actual definition, generating initial rejected picture data after rejection is completed, acquiring the correction of the initial rejected picture data by the current picture data owner body, generating final screened picture data after correction is completed, and storing the final screened picture data.
Specifically, the similarity numerical calculation module is further configured to:
generating current actual gray-scale pictures according to each current actual picture to be managed, wherein one current actual picture to be managed corresponds to one current actual gray-scale picture; generating a current spectrogram according to each current actual gray picture, and generating a current actual mask image according to each current spectrogram; generating a current spectrum mask image according to the current actual mask image and the corresponding current spectrum image; generating a current airspace picture according to the current spectrum mask map; respectively generating texture feature similarity between two current airspace pictures according to each current airspace picture;
The similarity value calculation module is further configured to generate a texture feature similarity value between the two current spatial pictures through the following formula:
wherein G is pq Representing a texture feature similarity value between the p-th airspace picture and the q-th airspace picture; F1F 1 p Gradient value frequency vector representing p-th spatial picture, F1 q A gradient value frequency vector representing the q-th airspace picture; F2F 2 p Representing the gradient direction appearance frequency representation vector of the p-th airspace picture, F2 q Representing a gradient direction appearance frequency representation vector of the q-th airspace picture; < F1 p ,F1 q The cosine similarity of the gradient value frequency vector of the p-th airspace picture and the gradient value frequency vector of the q-th airspace picture is represented; < F2 p ,F2 q The cosine similarity of the gradient direction appearance frequency representation vector of the p-th airspace picture and the gradient direction appearance frequency representation vector of the q-th airspace picture is represented; < F1 p ,F1 q > 2 Representing a distribution characteristic similarity metric value between gradient value distribution of an airspace picture corresponding to the p-th spectrum mask image and gradient value distribution of an airspace picture corresponding to the q-th spectrum mask image;<F2 p ,F2 q > 2 and representing a distribution characteristic similarity metric value between the gradient direction angle distribution of the airspace picture corresponding to the p-th spectrum mask diagram and the gradient direction angle distribution of the airspace picture corresponding to the q-th spectrum mask diagram.
Specifically, the similar picture eliminating module is further configured to:
obtaining a pre-stored standard similarity value according to the texture feature similarity value; screening out a current actual picture to be managed corresponding to a texture feature similarity value which is greater than or equal to the standard similarity value, and setting the current actual picture to be managed as an initial similar picture; acquiring screening and rechecking data of the initial similar pictures according to the initial similar pictures; screening the initial similar pictures according to the screening and rechecking data, and generating actual similar pictures, wherein the actual similar pictures are summarized to form the current similar picture group;
the intelligent picture management module is also used for:
the current actual sharpness of each of the actual similar pictures is calculated separately based on the following formula,
wherein Y is et Representing the current actual sharpness, alpha (beta) ety ) Representing e-type in the detection zonethe frequency of the fuzzy region in the y-th gradient direction in t detection windows, ln () represents the natural logarithmic function, beta ety Representing the y-th gradient direction in t detection windows in the e-type detection region; h represents the number of blurred regions of t in the detection region of the e type; screening out an actual similar picture corresponding to the current actual definition which is smaller than or equal to the preset standard definition, and eliminating; generating initial reject picture data after rejection is completed, acquiring correction of the initial reject picture data by the current picture data owner body, generating final screening picture data after correction is completed, and storing the final screening picture data.
A computer device comprising a memory storing a computer program and a processor implementing the steps of the artificial intelligence based data storage management method described above when the processor executes the computer program.
A computer readable storage medium having stored thereon a computer program which when executed by a processor performs the steps of the artificial intelligence based data storage management method described above.
The invention has the following technical effects:
according to the data storage management method and system based on artificial intelligence, the intelligent picture data management instruction set in the stored current summarized picture data is acquired sequentially through the acquisition of the current picture data owner body, and the current management picture data is acquired according to the intelligent picture data management instruction, wherein the current management picture data is a part of the current summarized picture data, and the current management picture data comprises a plurality of current actual pictures to be managed; performing similarity calculation on each current actual picture to be managed according to a preset intelligent picture screening algorithm, and obtaining texture feature similarity values between the current actual pictures to be managed; selecting actual similar pictures belonging to the same main body from the current management picture data according to the texture feature similarity value, and recording the actual similar pictures as a current similar picture group, wherein the actual similar pictures belonging to the same main body in the current management picture data are multiple, and the actual similar pictures are summarized to be the current similar picture group; the method comprises the steps of obtaining current actual definition of each actual similar picture in a current similar picture group, removing each actual similar picture according to each current actual definition, generating initial removed picture data after removing, obtaining correction of a main body of the current picture data to the initial removed picture data, generating final screening picture data after correcting, and storing the final screening picture data, namely, in order to achieve intelligent and regional division management, obtaining current management picture data based on an intelligent picture management instruction of the current picture data, wherein the current management picture data is a part of the current summary picture data, in order to achieve intelligent deletion of the same picture or an excessively similar picture in each current actual picture to be managed, performing similarity calculation on each current actual picture to be managed according to a preset intelligent picture screening algorithm, obtaining texture feature similarity value between the current actual pictures, generating final screening pictures after correcting, and storing the final screening picture data, and not performing accurate and further accurate screening on the final screening picture data.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the present application will be further described in detail with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the present application.
In one embodiment, the invention provides an application scenario of a data storage management method based on artificial intelligence, the application scenario comprises an intelligent terminal, the intelligent terminal is used for acquiring an intelligent management instruction of a current picture data owner body on picture data set in stored current summarized picture data, and acquiring current management picture data according to the intelligent management instruction of the picture data, wherein the current management picture data is a part of the current summarized picture data, and the current management picture data comprises a plurality of current actual pictures to be managed; performing similarity calculation on each current actual picture to be managed according to a preset intelligent picture screening algorithm, and obtaining texture feature similarity values between the current actual pictures to be managed; selecting actual similar pictures belonging to the same main body from the current management picture data according to the texture feature similarity value, and recording the actual similar pictures as a current similar picture group, wherein the actual similar pictures belonging to the same main body in the current management picture data are multiple, and the actual similar pictures are summarized to be the current similar picture group; obtaining current actual definition of each actual similar picture in the current similar picture group, removing each actual similar picture according to each current actual definition, generating initial removed picture data after removing, obtaining correction of the initial removed picture data by the current picture data owner body, generating final screened picture data after correcting, and storing the final screened picture data.
In this embodiment, the smart terminal may be, but is not limited to, various personal computers, notebook computers, smart phones, tablet computers, and portable wearable devices.
In one embodiment, as shown in fig. 1, there is provided an artificial intelligence-based data storage management method for data storage management of pictures, the method comprising:
step S100: acquiring an intelligent picture data management instruction set by a current picture data owner body on stored current summarized picture data, and acquiring current management picture data according to the intelligent picture data management instruction, wherein the current management picture data is a part of the current summarized picture data, and the current management picture data comprises a plurality of current actual pictures to be managed;
further, the current picture data owner is a user who has data of all pictures, or it can be understood that the current picture data owner is a user who needs to manage pictures.
In order to achieve intelligent and regional division management, further, an intelligent management instruction of the current picture data is obtained through the current picture data possession main body for the picture data set in the stored current summarized picture data, so that the current management picture data is obtained based on the intelligent management instruction of the picture data, and the current management picture data is a part of the current summarized picture data, therefore, all the current summarized picture data are not directly managed, but are managed based on the requirements of the current picture data possession main body, and further, management efficiency and intelligence are improved.
Step S200: performing similarity calculation on each current actual picture to be managed according to a preset intelligent picture screening algorithm, and obtaining texture feature similarity values between the current actual pictures to be managed;
step S300: selecting actual similar pictures belonging to the same main body from the current management picture data according to the texture feature similarity value, and recording the actual similar pictures as a current similar picture group, wherein the actual similar pictures belonging to the same main body in the current management picture data are multiple, and the actual similar pictures are summarized to be the current similar picture group;
further, in order to intelligently delete the same or too similar pictures in each current actual picture to be managed, similarity calculation is performed on each current actual picture to be managed according to a preset intelligent picture screening algorithm, texture feature similarity values between each current actual picture to be managed are obtained, then actual similar pictures belonging to the same main body are selected from the current management picture data according to the texture feature similarity values, and are recorded as a current similar picture group, so that pictures of the same main body are screened according to the texture feature similarity values, and subsequent deletion is facilitated. The same main body is a main body, for example, the main body can be a sculpture, and then intelligent deletion can be performed on a plurality of pictures of the sculpture, so that efficient storage management is realized, and intelligent similar pictures are deleted.
Step S400: obtaining current actual definition of each actual similar picture in the current similar picture group, removing each actual similar picture according to each current actual definition, generating initial removed picture data after removing, obtaining correction of the initial removed picture data by the current picture data owner body, generating final screened picture data after correcting, and storing the final screened picture data.
In order to further delete useless pictures and perform reliable and efficient management of picture data, further analyze the picture data by acquiring definition, specifically, firstly acquire current actual definition of each actual similar picture in the current similar picture group, then reject each actual similar picture according to each current actual definition, generate initial reject picture data after rejection is completed, acquire a main body possessed by the current picture data to correct the initial reject picture data, generate final screening picture data after correction is completed, store the final screening picture data, further perform accurate acquisition of similarity, then perform primary screening, then reject unclear pictures, and further manage, screen and store picture data.
In one embodiment, step S200: performing similarity calculation on each current actual picture to be managed according to a preset intelligent picture screening algorithm, and obtaining texture feature similarity values between the current actual pictures to be managed; the method specifically comprises the following steps:
step S210: generating current actual gray-scale pictures according to each current actual picture to be managed, wherein one current actual picture to be managed corresponds to one current actual gray-scale picture;
step S220: generating a current spectrogram according to each current actual gray picture, and generating a current actual mask image according to each current spectrogram;
wherein the gray scale map is converted into a spectrogram using fourier transform.
Further, in this step, firstly, the gray level map is converted into the current spectrogram, circles are made in the spectrogram by taking the center of the spectrogram as the center of the circle to obtain a plurality of different circles, and then, the circle with the smallest area is obtained based on the circle with the smallest area in the spectrogram and the circular ring formed by every two adjacent circles, and the mask image corresponding to the circular ring formed by every two adjacent circles is obtained.
Specifically, the center of the spectrogram is used as a circle center, the spectrograms are divided, firstly, the center of the spectrogram is used as the circle center, a first preset size is used as a radius to form a first circle, the first circle is a circle with the smallest area, then, a second circle is obtained by taking a second preset size as the radius to form a circle, then, the circles drawn in the spectrograms are N circles, a plurality of concentric circles with different radiuses are obtained in the spectrograms respectively, and the difference value of the radius sizes of two adjacent circles is the same. Further, the frequency values of pixels on the same circle in the spectrogram are the same, and the pixel value size represents the amplitude value; finally, a corresponding mask image is obtained based on the circles made on the spectrogram.
Step S230: generating a current spectrum mask image according to the current actual mask image and the corresponding current spectrum image;
further, the different mask images are multiplied with the spectrogram to obtain different spectrogram mask images.
Step S240: generating a current airspace picture according to the current spectrum mask map;
further, in this step, the spectrum mask map is converted into a spatial domain picture by fourier transform.
Step S250: and respectively generating texture feature similarity values between the two current airspace pictures according to the current airspace pictures.
Further, in order to perform accurate similarity calculation, for the purpose of accurately judging whether the similar pictures need to be deleted or not, further, firstly acquiring the pictures, converting the gray level pictures, generating a spectrogram, generating a mask image according to the spectrogram, and then acquiring the texture feature similarity value, specifically, firstly respectively generating current actual gray level pictures according to each current actual picture to be managed, wherein one current actual picture to be managed corresponds to one current actual gray level picture; then, respectively generating a current spectrogram according to each current actual gray picture, and generating a current actual mask image according to each current spectrogram; then, generating a current spectrum mask image according to the current actual mask image and the corresponding current spectrum image; then, generating a current airspace picture according to the current spectrum mask map; and finally, respectively generating texture feature similarity values between the two current airspace pictures according to the current airspace pictures, so that accurate acquisition of data representing similarity is realized, and the accuracy of deleting the data of the desired type later is improved.
Further, the intelligent picture screening algorithm may be understood as all the steps of the above steps S210 to S250. The steps in the embodiment are set, so that intelligent processing is realized, namely the intelligent processing based on the algorithm belongs to the category of artificial intelligence, and the method is named as a data management method based on the artificial intelligence.
In one embodiment, in step S250, in the step of generating the texture feature similarity values between the two current spatial pictures, the texture feature similarity values between the two current spatial pictures are generated, specifically by the following formula:
wherein G is pq Representing a texture feature similarity value between the p-th airspace picture and the q-th airspace picture; F1F 1 p Gradient value frequency vector representing p-th spatial picture, F1 q A gradient value frequency vector representing the q-th airspace picture; F2F 2 p Representing the gradient direction appearance frequency representation vector of the p-th airspace picture, F2 q Representing a gradient direction appearance frequency representation vector of the q-th airspace picture;<F1 p ,F1 q >the cosine similarity of the gradient value frequency vector of the p-th airspace picture and the gradient value frequency vector of the q-th airspace picture is represented;<F2 p ,F2 q >the cosine similarity of the gradient direction appearance frequency representation vector of the p-th airspace picture and the gradient direction appearance frequency representation vector of the q-th airspace picture is represented; <F1 p ,F1 q > 2 Representing a distribution characteristic similarity metric value between gradient value distribution of an airspace picture corresponding to the p-th spectrum mask image and gradient value distribution of an airspace picture corresponding to the q-th spectrum mask image;<F2 p ,F2 q > 2 and representing a distribution characteristic similarity metric value between the gradient direction angle distribution of the airspace picture corresponding to the p-th spectrum mask diagram and the gradient direction angle distribution of the airspace picture corresponding to the q-th spectrum mask diagram.
Specifically, the frequency of occurrence of different gradient values of pixel points in the airspace picture forms a gradient value frequency vector of the airspace picture; the frequency of occurrence of different gradient direction angles of pixel points in the airspace picture forms a gradient direction occurrence frequency representation vector of the airspace picture; classifying based on the differences of gradient value frequency vectors and gradient direction occurrence frequency characterization vectors among different airspace pictures to obtain airspace pictures of different categories.
Further, G pq Representing the distribution similarity between the texture distribution of the airspace picture corresponding to the p-th spectrum mask image and the texture distribution of the airspace picture corresponding to the q-th spectrum mask image,
G pq the larger the pattern of two airspace pictures isThe more similar the texture distribution, the more likely the two spatial pictures are of one type, so that the similarity comparison between the two spatial pictures is obtained based on the similarity comparison of the texture distribution, and further, whether the two spatial pictures are of the same type or not is accurately judged, so that whether deletion is needed or not is judged.
In one embodiment, step S300: selecting actual similar pictures belonging to the same main body from the current management picture data according to the texture feature similarity value, and recording the actual similar pictures as a current similar picture group, wherein the actual similar pictures belonging to the same main body in the current management picture data are multiple, and the actual similar pictures are summarized to be the current similar picture group; the method specifically comprises the following steps:
step S310: obtaining a pre-stored standard similarity value according to the texture feature similarity value;
step S320: screening out a current actual picture to be managed corresponding to a texture feature similarity value which is greater than or equal to the standard similarity value, and setting the current actual picture to be managed as an initial similar picture;
step S330: acquiring screening and rechecking data of the initial similar pictures according to the initial similar pictures;
step S340: and screening the initial similar pictures according to the screening and rechecking data, and generating actual similar pictures, wherein the actual similar pictures are summarized to obtain the current similar picture group.
Further, in order to further delete the same picture, it is further required to determine whether the pictures are similar, for example, assuming that there are three total pictures A, B, C and D, comparing a with B, comparing a with C, comparing D with C, and obtaining one texture feature similarity value respectively, thus, 3 texture feature similarity values are summed up, assuming that the values of the texture feature similarity values are 70, 91, 92 and 92.1 respectively and the standard similarity value is 90, the current actual picture to be managed corresponding to the texture feature similarity values are 91, 92 and 92.1, namely the initial similar picture, and then implementing further accurate rejection of the similar picture by manually screening and reviewing data of the initial similar picture, specifically, obtaining a pre-stored standard similarity value according to the texture feature similarity value; then, screening out the current actual picture to be managed corresponding to the texture feature similarity value which is larger than or equal to the standard similarity value, and setting the current actual picture to be managed as an initial similar picture; then, according to the initial similar picture, screening and rechecking data of the initial similar picture are obtained; and finally, screening the initial similar pictures according to the screening and rechecking data, and generating actual similar pictures, wherein the current similar picture group is obtained after the actual similar pictures are summarized, so that the screening and rejecting of the similar pictures are accurately realized.
In one embodiment, step S400: obtaining current actual definition of each actual similar picture in the current similar picture group, removing each actual similar picture according to each current actual definition, generating initial removed picture data after removing, obtaining correction of the initial removed picture data by a main body owned by the current picture data, generating final screened picture data after correcting, and storing the final screened picture data, wherein the method specifically comprises the following steps of:
step S410: the current actual sharpness of each of the actual similar pictures is calculated separately based on the following formula,
wherein Y is et Representing the current actual sharpness, alpha (beta) ety ) Representing the frequency of the occurrence of a blurred region in the y-th gradient direction in t detection windows in the e-type detection region, ln () representing the natural logarithmic function, beta ety Representing the y-th gradient direction in t detection windows in the e-type detection region; h represents the number of blurred regions of t in the detection region of the e type;
further, in order to achieve accurate acquisition of definition, a plurality of types are respectively set, and the upper side, the lower side, the left side, the right side and the middle part of the picture are respectively set, each area corresponds to one type, each area is divided into more tiny areas, each area is likely to be a fuzzy area, the occurrence of the fuzzy area indicates that insufficient definition exists, and in order to accurately detect, a plurality of gradient directions are set, in order to conduct fine detection, t detection windows are set, and accordingly reliability and accuracy of the acquired current actual definition are further achieved.
Step S420: screening out an actual similar picture corresponding to the current actual definition which is smaller than or equal to the preset standard definition, and eliminating;
step S430: generating initial reject picture data after rejection is completed, acquiring correction of the initial reject picture data by the current picture data owner body, generating final screening picture data after correction is completed, and storing the final screening picture data.
In order to perform accurate rejection, firstly intelligently screening pictures with poor definition, then manually performing data correction of the pictures to realize accurate and efficient rejection of redundant invalid picture data, specifically, firstly screening out actual similar pictures corresponding to the current actual definition which is smaller than or equal to the preset standard definition, performing rejection, then generating initial rejection picture data after the rejection is completed, acquiring correction of a main body of the current picture data on the initial rejection picture data, generating final screening picture data after the correction is completed, and storing the final screening picture data.
In one embodiment, step S100: acquiring an intelligent picture data management instruction set by a current picture data owner body on stored current summarized picture data, and acquiring current management picture data according to the intelligent picture data management instruction, wherein the current management picture data is a part of the current summarized picture data, and the current management picture data comprises a plurality of current actual pictures to be managed, and specifically comprises:
Step S110: acquiring important picture data marked by the current picture data possession main body from stored current summarized picture data, and generating initial demarcation picture data;
step S120: acquiring an initial selected area range of the current picture data owner body for the initial demarcation picture data;
step S130: generating a current demarcation correction area according to the initial selected area range;
step S140: generating an intelligent management instruction of the picture data according to the current defined correction area;
step S150: and generating current management picture data according to the picture data intelligent management instruction.
Further, in order to accurately generate the region, considering that the obtained current picture data possession body is not accurate for initially defining the initially selected region range of the picture data, since the current picture data possession body is present to use a brush to perform region division, the accurately defining of the selected picture is performed by correction, specifically, the initially selected region range of the initially defined picture data is obtained by the current picture data possession body; then, generating a current demarcation correction area according to the initial selected area range; then, generating an intelligent management instruction of the picture data according to the current defined correction area; and finally, generating current management picture data according to the picture data intelligent management instruction.
In one embodiment, as shown in FIG. 2, the present invention also provides an artificial intelligence based data storage management system, the system comprising:
the picture management demarcation module is used for acquiring an intelligent picture data management instruction set by a current picture data owner body in stored current summarized picture data, and acquiring current management picture data according to the intelligent picture data management instruction, wherein the current management picture data is a part of the current summarized picture data, and the current management picture data comprises a plurality of current actual pictures to be managed;
the similarity value calculation module is used for calculating the similarity of each current actual picture to be managed according to a preset intelligent picture screening algorithm and obtaining the similarity value of the texture characteristics among the current actual pictures to be managed;
the similar picture eliminating module is used for selecting actual similar pictures belonging to the same main body from the current management picture data according to the texture feature similarity value and marking the actual similar pictures as a current similar picture group, wherein the actual similar pictures belonging to the same main body in the current management picture data are multiple, and the actual similar pictures are summarized to form the current similar picture group;
The intelligent picture management module is used for acquiring the current actual definition of each actual similar picture in the current similar picture group, rejecting each actual similar picture according to each current actual definition, generating initial rejected picture data after rejection is completed, acquiring the correction of the initial rejected picture data by the current picture data owner body, generating final screened picture data after correction is completed, and storing the final screened picture data.
In one embodiment, the similarity value calculation module is further configured to:
generating current actual gray-scale pictures according to each current actual picture to be managed, wherein one current actual picture to be managed corresponds to one current actual gray-scale picture; generating a current spectrogram according to each current actual gray picture, and generating a current actual mask image according to each current spectrogram; generating a current spectrum mask image according to the current actual mask image and the corresponding current spectrum image; generating a current airspace picture according to the current spectrum mask map; respectively generating texture feature similarity between two current airspace pictures according to each current airspace picture;
The similarity value calculation module is further configured to generate a texture feature similarity value between the two current spatial pictures through the following formula:
wherein G is pq Representing a texture feature similarity value between the p-th airspace picture and the q-th airspace picture; F1F 1 p Gradient value frequency vector representing p-th spatial picture, F1 q A gradient value frequency vector representing the q-th airspace picture; F2F 2 p Representing the gradient direction appearance frequency representation vector of the p-th airspace picture, F2 q Representing a gradient direction appearance frequency representation vector of the q-th airspace picture;<F1 p ,F1 q >the cosine similarity of the gradient value frequency vector of the p-th airspace picture and the gradient value frequency vector of the q-th airspace picture is represented;<F2 p ,F2 q >the cosine similarity of the gradient direction appearance frequency representation vector of the p-th airspace picture and the gradient direction appearance frequency representation vector of the q-th airspace picture is represented;<F1 p ,F1 q > 2 representing a distribution characteristic similarity metric value between gradient value distribution of an airspace picture corresponding to the p-th spectrum mask image and gradient value distribution of an airspace picture corresponding to the q-th spectrum mask image;<F2 p ,F2 q > 2 and representing a distribution characteristic similarity metric value between the gradient direction angle distribution of the airspace picture corresponding to the p-th spectrum mask diagram and the gradient direction angle distribution of the airspace picture corresponding to the q-th spectrum mask diagram.
In one embodiment, the homogeneous picture rejection module is further configured to:
obtaining a pre-stored standard similarity value according to the texture feature similarity value; screening out a current actual picture to be managed corresponding to a texture feature similarity value which is greater than or equal to the standard similarity value, and setting the current actual picture to be managed as an initial similar picture; acquiring screening and rechecking data of the initial similar pictures according to the initial similar pictures; screening the initial similar pictures according to the screening and rechecking data, and generating actual similar pictures, wherein the actual similar pictures are summarized to form the current similar picture group;
the intelligent picture management module is also used for:
the current actual sharpness of each of the actual similar pictures is calculated separately based on the following formula,
wherein T is et Representing the current actual sharpness, alpha (beta) ety ) Representing the frequency of the occurrence of a blurred region in the y-th gradient direction in t detection windows in the e-type detection region, ln () representing the natural logarithmic function, beta ety Representing the y-th gradient direction in t detection windows in the e-type detection region; h represents the number of blurred regions of t in the detection region of the e type; screening out an actual similar picture corresponding to the current actual definition which is smaller than or equal to the preset standard definition, and eliminating; generating initial reject picture data after rejection is completed, acquiring correction of the initial reject picture data by the current picture data owner body, generating final screening picture data after correction is completed, and storing the final screening picture data.
In one embodiment, the picture management demarcation module is further for: acquiring important picture data marked by the current picture data possession main body from stored current summarized picture data, and generating initial demarcation picture data; acquiring an initial selected area range of the current picture data owner body for the initial demarcation picture data; generating a current demarcation correction area according to the initial selected area range; generating an intelligent management instruction of the picture data according to the current defined correction area; and generating current management picture data according to the picture data intelligent management instruction.
In one embodiment, as shown in FIG. 3, a computer device includes a memory storing a computer program and a processor implementing the steps described above for an artificial intelligence based data storage management method when executing the computer program.
A computer readable storage medium having stored thereon a computer program which when executed by a processor performs the steps of the artificial intelligence based data storage management method described above.
Those skilled in the art will appreciate that implementing all or part of the above described methods may be accomplished by way of a computer program stored on a non-transitory computer readable storage medium, which when executed, may comprise the steps of the embodiments of the methods described above. Any reference to memory, storage, database, or other medium used in the various embodiments provided herein may include non-volatile and/or volatile memory. The nonvolatile memory can include Read Only Memory (ROM), programmable ROM (PROM), electrically Programmable ROM (EPROM), electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double Data Rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous Link DRAM (SLDRAM), memory bus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), among others.
The technical features of the above embodiments may be arbitrarily combined, and all possible combinations of the technical features in the above embodiments are not described for brevity of description, however, as long as there is no contradiction between the combinations of the technical features, they should be considered as the scope of the description.
The above examples merely represent a few embodiments of the present application, which are described in more detail and are not to be construed as limiting the scope of the invention. It should be noted that it would be apparent to those skilled in the art that various modifications and improvements could be made without departing from the spirit of the present application, which would be within the scope of the present application. Accordingly, the scope of protection of the present application is to be determined by the claims appended hereto.