CN114186497A

CN114186497A - Intelligent analysis method, system, equipment and medium for value of art work

Info

Publication number: CN114186497A
Application number: CN202111535378.8A
Authority: CN
Inventors: 朱莉; 蔡翔; 付裕
Original assignee: Hubei University of Technology
Current assignee: Hubei University of Technology
Priority date: 2021-12-15
Filing date: 2021-12-15
Publication date: 2022-03-15
Anticipated expiration: 2041-12-15
Also published as: CN114186497B

Abstract

The invention belongs to the technical field of computer information processing, and discloses an intelligent analysis method, a system, equipment and a medium for the value of an art work, which are used for displaying the user work set as open and multi-angle omnibearing information of a corresponding user; evaluating and scoring the artwork and guiding the comments by using a four-dimensional evaluation mode; meanwhile, different artistic works are automatically evaluated based on computational aesthetics, presented in a visual form and output an evaluation report; forming an evaluation report through full-automatic analysis and summary, and outputting an AI evaluation description of the work, an AI overall comment, an AI growth suggestion and a user growth message; and performing art literacy and capability test and outputting a multi-dimensional objective evaluation result and a learning suggestion. The invention promotes the relearning of individuals and the redisplay of works; and the development closed loop of cultivation, display and evaluation of personal artistic literacy is constructed and realized.

Description

Intelligent analysis method, system, equipment and medium for value of art work

Technical Field

The invention belongs to the technical field of computer information processing, and particularly relates to an intelligent method, a system, equipment and a medium for analyzing the value of an art work.

Background

With the continuous improvement of the quality of life of people, the potential psychological demands on the quality of life and the mental civilization construction drive make people attach more and more importance to the personalized work display and the affirmation of the self value. The artificial intelligence is accompanied with the continuous progress of machine learning and deep learning algorithms, and under the environment that the multimedia network is widely applied, the intelligent algorithms are applied to various fields in life, so that a new man-machine interaction form and a professional skill tool are provided for users.

The first prior art is as follows: CN201710865048.2 artwork selecting method and system. The method and the system for evaluating and selecting the works of art provided by the invention can evaluate and select the works of art for competition for an exhibition on line, increase the popularization range, are convenient and simple, and simultaneously reduce the contribution cost of an author and the evaluation cost of a host; furthermore, the public and transparent evaluation is performed through the evaluation unit, so that the notarization degree and the notarization ability in the evaluation process are increased, and the experience degree of the participants is increased; furthermore, the selecting unit selects the selecting information of the artwork through the receiving client, the selecting mode is fair and fair, the enthusiasm of the author in competition is stimulated, and the system has practicability. Although the selection mode is public and fair, the evaluation of the quality of the work still adopts a manual evaluation mode, so that the subjectivity is strong, and the real fairness and fairness cannot be realized.

The second prior art is: CN201710919334.2 art appreciation ability evaluation method, evaluation server and evaluation system. The invention provides an art appreciation capability assessment method, wherein the assessment method comprises a plurality of assessment periods, and each assessment period comprises the following steps: providing an electronic carrier of an artwork to be practiced to a viewer, wherein the electronic carrier of the artwork to be practiced comprises a picture and/or a video of the artwork to be practiced; receiving an appreciation result made by an appreciator according to the electronic carrier of the artwork to be practiced; and scoring the appreciation ability of the appreciator by using a reference factor to obtain the appreciation ability score of the appreciator, wherein the reference factor comprises the market price of the artwork to be practiced and/or the score of the appreciation result by an expert. The invention also provides an evaluation server and an evaluation system for the art appreciation capability. The method for evaluating the art appreciation ability can be used for evaluating the art appreciation ability of the appreciator, thereby being beneficial to the appreciator to improve the appreciation level. Although the evaluation system is provided, statistics of the works and evaluation results thereof can be provided, the evaluation essence of the quality of the works still adopts a manual evaluation mode, and real fairness and justice cannot be realized.

The prior art is three: the paper analyzes from the actual requirement of a system, an MVC framework under a B/S mode is adopted according to a software development flow, Eclipse is used as a development environment, jsp is adopted to complete the design of a foreground interface, Struts2 is used to complete the logic service processing of the system, a Microsoft SQL Server database is adopted, a hibernate framework is used to realize the data access, a file storage and web Server are tomcat, and virtual reality is carried out through a VR technology, so that a main functional framework of an art student work display platform integrating display and social contact is finally realized. Mainly is the programming realization mode of a system software system, and has insufficient innovation.

At present, a system for displaying, socializing and evaluating related information of artwork does not form a complete system, and the functions are not complete. For example: the simple picture work uploading system has the defects of single form, incapability of realizing personalized display and the like. Because the content of the artistic works has diversity, abstraction and complexity, the work evaluation mostly uses a simple traditional manual evaluation system, and the evaluation result has larger subjective uncertainty due to different personal knowledge experiences, so the evaluation standard of the artistic works is not uniform, and subjective factors dominate. Long period, strong subjective color; the evaluation difficulty of the existing automatic machine for the works is very high due to the lack of objective and timely guidance opinions.

Currently, a manual evaluation mode is generally adopted, subjective factors are very strong, standards are not uniform, and the value of the work is not favorably and objectively evaluated. For example: at present, a Chinese painting style work X is a pair, experts A like Chinese paintings and score 90 is given; expert B likes caricature work, scoring 80. And what are the basis for the scores of 90 and 80? Perhaps by feel alone, theoretical support is difficult. In addition, the manual evaluation mode is time-consuming and labor-consuming, a work is carefully evaluated, specific conclusions and improvement suggestions are given from the beginning of viewing, and the time consumption is averagely different from 5-20 minutes according to different abilities of evaluators.

Through the above analysis, the problems and defects of the prior art are as follows:

(1) a service problem. It is not practical to evaluate the increasingly dramatic works of art entirely by hand. Except for taking related payment courses, most individuals do not have special teachers, cannot get the guidance of the special teachers timely, and are inconvenient to communicate and learn at any time and any place; the time of the expert or teacher is limited, and once too many works are produced, the time and labor are consumed for guidance, and the efficiency is low.

(2) The problem is evaluated. The evaluation result lacks certain objectivity and fairness. The machine automatic evaluation can guarantee objectivity and fairness, but AI artificial intelligence just starts in the aspects of perception and cognition of a human simulator to beauty, and when an evaluation model is established by using an AI algorithm, a massive work training set is needed, the existing work set, particularly the domestic work set, is very limited, and the modeling difficulty is further increased. Making it very difficult to technically implement automated evaluation.

The difficulty in solving the above problems and defects is:

(1) it is unlikely that everyone is equipped with a specialized art teacher and gets learning guidance anytime and anywhere; the chances of multiple people getting together to communicate with each other are also very limited.

(2) The evaluation of the art is not simply a mistake but an evaluation of skill and feelings such as skill and feeling. Recently, AI artificial intelligence technology has made great progress in the fields of object recognition, face recognition, behavior recognition, etc., so that the gap between the ability of computers in recognizing objects, distinguishing people, recognizing affairs, etc. and the human level is gradually reduced, even exceeding the human intelligence level in some specific application scenarios. However, compared with semantic features of object recognition, the human aesthetic rules have not been quantifiable scientific explanation, and the difficulty in selecting aesthetic feature points of the work is high, so that the difficulty in realizing accurate AI evaluation is very high.

(3) The AI algorithm establishes the accuracy of the model, and one of the key factors is the scale and number of the training data sets. The existing works, particularly domestic works, are very limited, and effective calibration of the works by professionals is lacked, so that the evaluation modeling difficulty is very high.

The significance of solving the problems and the defects is as follows: the system provides a platform for art learning, result display and interaction for everyone, and particularly can obtain objective, fair and efficient result evaluation and learning guidance.

Disclosure of Invention

Aiming at the problems in the prior art, the invention provides an intelligent method, a system, equipment and a medium for analyzing the value of an art work.

The invention is realized in such a way that the method for intelligently analyzing the value of the art works constructs an aesthetic perception calculation model by utilizing a machine learning and deep learning method from different angles of colors, vision, skills, spaciousness, modeling, innovativeness, master matching degree and the potential value of the art works, quantifies and calculates the human aesthetic process, and intelligently analyzes the value of the art works under multiple scenes.

Further, (1) the specific process of constructing the aesthetic perceptual computing model is as follows:

firstly, establishing an image database; the collected over 500 ten thousand domestic and foreign works are subjected to preprocessing processes such as cutting, coding by number, image enhancement and the like as required in an image form to form an image database;

secondly, training a model; firstly, selecting an aesthetic description mode; then, selecting a proper learning model; information such as an aesthetic description mode, psychological rules, a user evaluation framework and the like is fused into a corresponding learning model, then a training data set of an image database in the first step is input into the learning model for training, and finally an aesthetic perception calculation model is formed;

testing the model; inputting a test data set of the image database into the aesthetic perception calculation model, testing the accuracy of the model, analyzing the reason of inaccuracy and quantizing to form a loss function value;

improving the model; and improving the model function according to the loss function value obtained from the third step, thereby improving the evaluation precision.

(2) The intelligent analysis method for the value of the art work comprises the following steps:

sending the works improved by the user to an aesthetic perception calculation model through a system platform of the patent to obtain AI intelligent evaluation; on the basis of AI intelligent evaluation, the method automatically obtains overall evaluation, item evaluation, learning suggestion and value intelligent analysis results of growth mail works by combining expert evaluation, public evaluation and potential evaluation.

Further, the method specifically comprises the following steps:

displaying the works of the user set as open and the multi-angle omnibearing information of the corresponding user;

evaluating and scoring the artwork and guiding the comments by using a four-dimensional evaluation mode; meanwhile, different artistic works are automatically evaluated based on computational aesthetics, presented in a visual form and output an evaluation report;

forming an evaluation report through full-automatic analysis and summary, and outputting an AI evaluation description of the work, an AI overall comment, an AI growth suggestion and a user growth message;

performing art literacy and capability test and outputting a multi-dimensional objective evaluation result and a learning suggestion; meanwhile, information pushed every day is used for assisting in carrying out multi-dimensional objective evaluation results and learning suggestions.

Further, automated assessment of different works of art based on computational aesthetics includes: by means of color, vision, skill, spaciousness, modeling, innovation, master matching degree, work potential value and indexes thereof, machine learning and deep learning algorithms are utilized, and based on corresponding evaluation methods and reference judgment standards, automatic evaluation of art works is carried out.

Further, the method specifically comprises the following steps:

(1) color: comprehensively evaluating the hue, lightness, purity and color richness of the work by using a function;

Color＝H*t₁+V*t₂+S*t₃+F*t₄；

wherein color represents the total evaluation value of color, H represents the evaluation value of hue, and V is shownThe brightness evaluation value, S the purity evaluation value and F the color richness evaluation value; t is t₁～t₄Weight values representing H, V, S and F, respectively;

H. v, S are calculated by opencv library functions;

the evaluation method of FF is as follows: eight base hues were determined to be: r (1, 0, 0) red, G (0, 1, 0) green, B (0, 0, 1) blue, Y (1, 1, 0) yellow, M (1, 0, 1) magenta, C (0, 1, 1) cyan, K (0, 0, 0) black, W (1, 1, 1) white; calculating the Euclidean distance between each pixel of the detected image and each basic tone, wherein the basic tone with the minimum distance is the color of the pixel; counting the pixel number ratio of each basic tone, if the sum of the pixel ratio of the black basic tone and the pixel ratio of the white basic tone exceeds 90%, determining that the picture is a gray tone, otherwise, determining that the picture is a color tone; in the color tone, judging whether the pixel ratio of the remaining 6 basic tones exceeds 5%, if so, judging that the basic tones are prominent colors, and obtaining corresponding F values according to the number of the prominent colors;

(2) and (3) vision: the evaluation value is higher, so that the texture of the work is richer and finer; the evaluation method comprises the following steps: analyzing the texture distribution of the pictures from the visual texture, graying each picture, and performing LBP algorithm processing to obtain a corresponding texture image; the entropy of each texture image is calculated in a quantitative mode, and the larger the entropy value is, the richer the corresponding texture is; normalizing, limiting the numerical value to 0-1, averagely dividing the numerical value into seven grades, wherein 0 is the lowest grade and has the worst visual effect, and 1 is the highest grade and has the best visual effect;

(3) the technique comprises the following steps: a composition capability for characterizing a work; dividing the picture into two types of landscape painting and non-landscape painting for independent detection, wherein the landscape painting mainly comprises a linear and S-shaped composition, and the non-landscape painting mainly comprises line detection; the method specifically comprises the following steps: carrying out edge detection on the picture, and carrying out classified detection and composition; if the scene picture is a landscape picture, checking a straight line or a curve which spans the whole picture, namely 3/4 width or height in the picture; if yes, the picture is a straight-line or S-shaped picture composition; otherwise, judging that the picture has no obvious shape; if it is not a landscape, the picture is examined for a straight line that spans the entire width or height of 3/4, if it is not a landscapeIf yes, the model is in a straight shape; otherwise, detecting four edge points of the top, the bottom, the left and the right on the edge picture, connecting the four edge points clockwise to form a quadrilateral shape, if the length l of a connecting line between two certain points is too short, when the two points are connected with each other, the four edge points are connected with each other in a clockwise direction to form a quadrilateral shape

When the two points are combined, the midpoint is taken to form a triangular shape; outputting corresponding scores based on the classification detection result;

(4) the sense of space: the degree feeling of the real object in the work is represented; evaluating the space sense parameters of the picture of the work from the three aspects of occlusion, contrast and perspective by adopting an automatic evaluation method based on computational aesthetics of deep learning; the method specifically comprises the following steps: preparing a training data set; dividing a training picture set into 5 grades by an expert according to a preset spatial perception basic evaluation criterion for deep learning network training; preprocessing each picture, performing noise reduction processing on the pictures by adopting Gaussian filtering, extracting a gray image from the filtered image, and converting the pixel value of the obtained gray image into a pixel matrix with the size of 224 × 224; performing convolution operation on the whole picture by adopting a convolution core of 3 x 3 to extract an image characteristic value, training by adopting a deep learning network resnet, wherein the number of the convoluted picture per time, namely, batch _ size, is 16, the number of network cycle iteration times, namely, epochs, is 200, an output layer outputs 5 ganglion points corresponding to 5 levels, each neural node value is the probability of the classification to which the picture belongs, and the level corresponding to the highest probability value is selected as a spatial sensation level; after the training is finished, storing the training model, outputting the work to be evaluated from the trained model to judge the belonging grade, and converting the grade into an evaluation result;

(5) modeling: the shape and the volume of the content of the representation work; carrying out edge detection on the picture, and detecting and modeling in a classified manner; detecting whether the area of the middle point and the dot circle of the whole picture exceeds 2/3, if not, judging that the picture has no obvious shape, and if so, judging that the picture has a dot shape or other shapes; then detecting the straight line of the whole picture in the picture, if the straight line is in a line type shape, detecting whether all the lines are connected into a plane to form the line type shape, and if not, the straight line is in a point-line-surface type shape;

(6) the innovation is as follows: evaluating from subject innovation, expression technique innovation, composition innovation, modeling innovation and other dimensions by adopting a deep learning automatic evaluation method based on computational aesthetics; the method specifically comprises the following steps: dividing the images into 5 grades according to the innovation degree, manually evaluating 20000 work images, classifying the 5 grades respectively, and averaging 5000 images in each grade; performing model training and training learning of RGB three-channel colors by adopting an Alex Net network, and outputting a trained model to evaluate the work to be evaluated;

(7) master matching degree: the method is used for measuring which side of the master or famous family is closest to the work, and the painting style representing the author of the work is most similar to the painting style or the painting method of the master or famous family; a deep learning method is adopted, and the similarity parameter values of the picture of the work and the work of which teacher are obtained comprehensively from the aspects of composition, color, texture, space and the like; the evaluation function is as follows:

Simila＝SSIM*t₁+CSIM*t₂+TSIM*t₃+PSIM*t₄；

wherein, Simila represents the total evaluation value of the master matching degree, SSIM represents the structure similarity evaluation value, CSIM represents the color similarity evaluation value, TSIM represents the texture similarity evaluation value, and PSIM represents the space sense similarity evaluation value; t 1-t 4 respectively represent the weight values of CSIM, V, S and F; the method comprises the following specific steps: dividing the master work set into four types, namely, people, animals, landscapes and the like, and selecting the corresponding master work set according to the type of the evaluated picture; calculating the structural similarity CSIM between the evaluated picture and each picture in the master work set_iThe maximum value is taken as CSIM ═ CSIM_max(ii) a The master work corresponding to the maximum value is the final matching work, and the master is the best matching master.

It is a further object of the present invention to provide a computer apparatus comprising a memory and a processor, the memory storing a computer program which, when executed by the processor, causes the processor to perform the steps of the method of intelligent value analysis of an art work.

It is another object of the present invention to provide a computer-readable storage medium storing a computer program which, when executed by a processor, causes the processor to execute the steps of the method for intelligently analyzing the value of an art work.

Another object of the present invention is to provide an art value information processing system for implementing the intelligent analysis method for the value of an art, the art value information processing system including:

the information management submodule is used for managing the user information and the account information;

the work display submodule is used for uploading and displaying management of works;

the work evaluation submodule is used for evaluating the works and outputting corresponding suggestions;

the personal learning submodule is used for storing a professional art knowledge database and art literacy and capability tests;

and the communication interaction submodule is used for providing a work communication and art activity development platform.

Further, the information management subsystem includes:

the account information management module is used for managing a user name, a mobile phone number, a head portrait, a login password and other account information;

the user information management module is used for managing the real name, sex, identity, birth date, authority and other information of the user;

and the copyright registration module is used for performing chain linking and copyright registration on the content of the work to obtain a unique ID (identity) which can not be tampered and is accurately proved by original creation and simultaneously recording the unique ID on the chain.

Further, the work presentation subsystem comprises:

the work uploading module is used for uploading the work by inputting the type of the work and inputting work information according to a certain rule; meanwhile, the system is used for storing uploaded works by utilizing a cloud storage server;

the work display module is used for displaying the basic information of the user and displaying the works disclosed by the user according to the comprehensive ordering of the gifts clicked, praised, commented and sent out; the user basic information includes: user nickname, age, and number of praise obtained;

the works application module is used for selecting different picture frames to decorate the works, and the decorated works are placed indoors or outdoors, on the wall of a living room or on a display wall of a painting exhibition.

Further, the work evaluation subsystem includes:

the expert online commenting module is used for manually evaluating and scoring the user works by an expert or a teacher and giving learning guidance;

the AI intelligent evaluation module is used for carrying out automatic evaluation of works based on computational aesthetics aiming at different artistic works and outputting an evaluation report in a visual form;

and the suggestion output module is used for outputting the AI evaluation description of the work, the AI overall comment and the AI growth suggestion by using a recommendation algorithm aiming at the evaluation report and the suggestion aiming at the user.

Further, the professional learning subsystem comprises:

the professional art knowledge database is used for storing the birth level, the genre, the creation characteristics and other information of thousands of famous art creators in the ancient and present and the drawing of most of the creators; meanwhile, the system is used for automatically pushing knowledge to the user every day;

the art literacy and competence test module is used for carrying out art literacy and competence test by utilizing multi-form and multi-class test questions and outputting a multi-dimensional objective evaluation result and a learning suggestion;

the communication interaction subsystem comprises:

the work communication module is used for creating hearts and feelings of works and expressing the attitude of the works through the modes of leaving messages, commenting, giving gifts and interacting the works by utilizing the social network platform;

the art activity developing platform is used for developing art activities by utilizing expert lectures, online lectures, art competitions, researches, practices and other art activity forms combining online and offline.

By combining all the technical schemes, the invention has the advantages and positive effects that: the method utilizes the technologies of machine learning, cloud computing, big data and the like to construct an individualized cloud exhibition hall of the personal artistic works in multimedia forms including images, videos and the like, and the artistic ability is visualized; through a multi-dimensional evaluation mechanism, a comprehensive evaluation form of mutual combination, complementation and complementation of subjective evaluation and objective evaluation, manual evaluation and machine evaluation is carried out, an authoritative and objective personal learning suggestion is given automatically, and personal relearning and product redisplay are promoted; the development closed loop of the personal artistic literacy learning, displaying and evaluation is carried out, an integrated solution scheme of complete work displaying, interacting and evaluating is provided, and a thought is provided for the learning of other fields of the person; the method is beneficial to the growth of the comprehensive strength of personal art care and creates an Internet ecological circle of the artwork.

The invention provides an integrated comprehensive social platform combining personal learning, work display, communication interaction, manual evaluation and automatic AI evaluation in the art field, realizes a closed loop for intelligently evaluating the artistic learning output result to obtain a suggestion to continue learning, provides a personalized online personal work exhibition hall, a personal work database and a matched application service, and has no similar system in the market.

The invention realizes the innovation of the display mode. The existing art work display system mainly uploads and displays works and introduces the name, age and identity of an author. The artistic work display realized by the invention is presented from 5 dimensions such as a work exhibition hall, author information, work content, work value, work potential value and the like, the personal value of an author, the artistic value and the commercial value of a work are reflected, and a constructive opinion is given for market valuation of the work, and the method specifically comprises the following steps:

(1) the work exhibition hall: the method comprises the steps of total number of works, representative works, classification presentation of the works and historical presentation of the works, and is a direct display of a work set. According to the time advancing sequence of a large number of works, the development condition of an author in the art field is analyzed, the personal art literacy cultivation process is embodied, and on the basis, a recommendation algorithm is utilized to automatically push personal learning suggestions.

(2) Author information: including author name, grade, age, academic degree, educational experience, work experience, etc., embodying the individual learning ability, professional ability and artistic potential of the author. And then the mutual influence relationship between the personal ability and the artistic literacy is analyzed by combining the 'work exhibition hall' information, thereby giving personal growth suggestions.

(3) The content of the work is as follows: including work name, creation time, creation location, work story, block chain copyright, etc., and embodies the artistic value of a single artistic work. The influence of the creation environment on the psychology of the author and the content of the work is analyzed by combining the creation time and the creation place, and the relationship between psychology and art is researched.

(4) Evaluation of the work: the method comprises the steps of evaluating values of experts, masses and AI intelligence, analyzing the artistic value and the direct commercial value of the work by combining subjective evaluation and objective evaluation, and giving a constructive opinion for market evaluation of the work.

(5) The potential value of the work is as follows: the application effect of the work in different scenes, different industries or different markets is shown, and the potential commercial value of the work in the future is analyzed.

The invention realizes innovation of a work evaluation mode. Traditional work evaluation is generally manual comment; the method is applied to the teaching field at most and is used for evaluating the teaching quality of teachers or the learning effect of students; and secondly, the auction industry, the basis for the valuation of the works. Because of the evaluation of a single person or a few persons, the evaluation is subjective and even depends on the ability level of the evaluator. The work evaluation mode provided by the invention is developed from 4 dimensions such as expert comment, popular evaluation, AI intelligent evaluation and potential value, and the like, and aims to analyze the artistic value, the commercial value and the potential value of the work in a manner of combining subjectivity and objectivity and combining manpower with machines, and provide constructive opinions for market evaluation of the work, and the work evaluation method specifically comprises the following steps:

(1) and (4) evaluating by an expert: experts in the experienced field evaluate the works by using professional knowledge and emotional colors, and the image of the works in the eyes of the experts is reflected.

(2) And (3) public evaluation: the impression of the works in the public is reflected through the forms of the number of praise, the number of comments, the content description, the number of received gifts, the value of the gifts and the like.

(3) AI Intelligent assessment: the method adopts artificial intelligence theory and technology, realizes automatic evaluation by using methods such as big data, cloud computing, machine vision, deep learning and the like, and objectively, fairly and efficiently realizes evaluation which is not good or difficult to make manually.

(4) Potential evaluation: by displaying the artistic effect of the work applied to a specific field, a specific industry or a specific scene, the method helps to mine the commercial application value of the work, and further realizes value conversion.

The invention also realizes AI intelligent evaluation algorithm innovation. For the field of picture evaluation, manual evaluation is mainly used, and automatic evaluation based on computational aesthetics only stays in parameters such as color and composition, and is mostly realized by a traditional image processing mode. The invention realizes the automatic evaluation of parameters of the pictures such as color, composition, vision, spaciousness, modeling, innovation, master matching degree and the like. For the parameters of 'spaciousness' and 'innovativeness', human eyes can easily make judgment according to experience and feeling, but a machine is difficult to describe and evaluate formally. Meanwhile, according to the evaluation result, the guidance suggestion can be given after intelligent analysis.

The invention also realizes the improvement of AI intelligent evaluation performance. The conventional automatic evaluation based on the computational aesthetics consumes a long time which is about 10-30 s, and the evaluation speed is optimized by using big data and cloud computing technology, so that the time consumption is shortened to about 3-8 s.

Drawings

FIG. 1 is a flow chart of an intelligent analyzing method for the value of an art work provided by an embodiment of the invention.

FIG. 2 is a schematic diagram of an architecture of an art value information processing system provided by an embodiment of the invention;

in fig. 2: 1. a facility layer; 2. a network layer; 3. a data layer; 4. an application layer; 5. and displaying the layer.

FIG. 3 is a schematic diagram of an artwork value information processing system provided by an embodiment of the present invention.

FIG. 4 is a schematic diagram of a work presentation subsystem interface provided by an embodiment of the present invention.

Fig. 5 is a schematic diagram of a learning subsystem interface provided in an embodiment of the present invention.

Fig. 6 is a diagram of AI evaluation author information provided by an embodiment of the present invention.

Fig. 7 is a diagram of AI evaluation parameter information provided by an embodiment of the present invention.

Fig. 8 is an AI-evaluation guidance information diagram provided by an embodiment of the present invention.

FIG. 9 shows AI evaluation color parameter result 1 according to an embodiment of the invention.

Fig. 10 shows the AI-evaluated color parameter result 2 according to an embodiment of the present invention.

Fig. 11 is a picture of a work according to an embodiment of the present invention.

FIG. 12 is an AI evaluation triangle schema provided by an embodiment of the invention.

Fig. 13 is a picture of a work according to an embodiment of the present invention.

FIG. 14 is an AI evaluation texture profile provided by an embodiment of the invention.

Detailed Description

In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail with reference to the following embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

Aiming at the problems in the prior art, the invention provides an intelligent analysis method, a system, equipment and a medium for the value of an art work, and the invention is described in detail below by combining the attached drawings.

As shown in FIG. 1, the intelligent analysis method for the value of the art provided by the invention comprises the following steps:

s101: user registration, namely, logging in by using a registered user name or a registered mobile phone number; inputting and uploading the name of the work, the name and age of the creator, the description of the work and the type of the work according to certain rules;

s102: displaying the works of the user set as open and the multi-angle omnibearing information of the corresponding user;

s103: evaluating and scoring the artwork and guiding the comments by using a four-dimensional evaluation mode; meanwhile, different artistic works are automatically evaluated based on computational aesthetics, presented in a visual form and output an evaluation report;

s104: forming an evaluation report through full-automatic analysis and summary, and outputting an AI evaluation description of the work, an AI overall comment, an AI growth suggestion and a user growth message;

s105: performing art literacy and capability test and outputting a multi-dimensional objective evaluation result and a learning suggestion; meanwhile, information pushed every day is used for assisting in carrying out multi-dimensional objective evaluation results and learning suggestions.

As a preferred embodiment, the core method in each step is specifically implemented as follows:

(1) and constructing an aesthetic perceptual calculation model. The specific process is as follows:

creating an image database. And (3) forming an image database by carrying out preprocessing processes such as cutting, coding by number, image enhancement and the like on the collected over 500 thousands of domestic and foreign works in an image form according to needs. Wherein a part of the image is used as a training data set, a part of the image is used as a testing data set, and the proportion of the two is approximately 6: 4.

And secondly, training a model. First, an aesthetic description is chosen, such as: color, vision, technique, spaciousness, modeling, innovation, master matching degree, potential value of art works and other parameter forms. Then, an appropriate learning model is selected, for example: the 'color' parameter can select a statistical mathematical model, the 'vision' parameter can select a local binary pattern algorithm (LBP) model, the 'technical' parameter adopts a classical machine learning classification composition detection algorithm model, the 'spacial' parameter adopts a Resnet deep learning network model, the 'innovative' parameter adopts a mixed model combining an Alex Net deep learning network and a classical machine vision method, and the like. Information such as an aesthetic description mode, psychological rules, a user evaluation framework and the like is fused into a corresponding learning model, then a training data set of an image database in the first step is input into the learning model for training, and finally an aesthetic perception calculation model is formed.

And testing the model. And inputting the test data set of the image database into the aesthetic perception calculation model, testing the accuracy of the model, analyzing the reason of inaccuracy and quantizing to form a loss function value.

And fourthly, improving the model. And improving the model function according to the loss function value obtained from the third step, thereby improving the evaluation precision.

(2) Intelligent analysis of artistic work value

And sending the works improved by the user to an aesthetic perception calculation model through the system platform of the patent to obtain AI intelligent evaluation. On the basis of AI intelligent evaluation, the intelligent analysis results of the values of the works such as general evaluation, item evaluation, learning suggestion, growth words and the like are automatically obtained by combining expert evaluation, public evaluation and potential evaluation.

One skilled in the art can also use other steps to implement the method for intelligently analyzing the value of an art work provided by the present invention, and the method for intelligently analyzing the value of an art work provided by the present invention shown in fig. 1 is only one specific embodiment.

As shown in fig. 2, the art value information processing system provided by the present invention includes:

the facility layer 1 is a hardware device layer for providing service for the system, and the local device can also be a cloud device;

a network layer 2, which realizes data transmission by using various existing networks;

the data layer 3 is a technical core layer and is used for realizing AI intelligent analysis and processing of the value of the artwork;

the application layer 4 realizes the core function of the system through an information management submodule, a work display submodule, a work evaluation submodule, a personal learning submodule and an exchange interaction submodule;

and the display layer 5 provides system interaction functions such as uploading, downloading, recording, watching and the like for the user through various terminal devices.

In an embodiment of the present invention, the information management sub-module includes an account information management module, a user information management module, and a copyright registration module, and is configured to manage user information and account information.

The work display submodule comprises a work uploading module, a display module and a work application module and is used for uploading and displaying management of works.

And the work evaluation submodule comprises an expert online comment module, an AI intelligent comment module and a suggestion output module and is used for evaluating the work and outputting corresponding suggestions.

And the personal learning submodule comprises a professional art knowledge database and an art literacy and competence test module, and is used for storing the professional art knowledge database and the art literacy and competence test.

The technical solution of the present invention is further described below with reference to the accompanying drawings.

Example 1:

as shown in fig. 3, the system for displaying and evaluating the artwork under the machine learning algorithm of the present invention mainly comprises five core portions, namely, an information management sub-module, a work displaying sub-module, a work evaluating sub-module, a personal learning sub-module, and an interaction sub-module.

(1) Information management submodule

The user information management mainly comprises: account information management, user information management, and the like.

1) Account information management, such as: and managing information such as user name, mobile phone number, head portrait, login password and the like.

The first login needs to be registered, and the registration has two modes, namely a user name mode and a mobile phone number mode. The 'user name mode' needs to provide a user name, and set a password according to a certain rule, wherein the user name can be a real name, a nickname or a mailbox. The 'mobile phone number mode' needs to provide a personal mobile phone number and completes registration in a short message verification mode. And logging in again, and selecting a user name mode or a mobile phone number mode. The 'user name mode' requires inputting a user name and a password, and the system can be logged in after matching; the 'mobile phone number mode' needs to input the mobile phone number, and the system can be logged in after the short message is successfully verified. After login is successful, the user name, head portrait, mobile phone number, password and other information can be modified, but re-authentication is required.

2) User information management, such as: real name, gender, identity (individual or teacher identity registration), date of birth (distinguishing the creation age of the user), rights, etc. Wherein: "identity" is divided into individuals, teachers or specialists, institutions or institutions. "personal" may be a student, an hobbyist, etc.; the teacher may be a college professional teacher or a training institution teacher, and the expert is a famous person in the art field. The teacher can become a high-level 'instructor' after the assessment is passed in a certain way; an organization or institution acts as a B-end, group-form user.

3) The electronic copyright and block chain technology of works has the advantages of decentralization, time stamp marking, non-tampering property and low right maintenance cost, and can effectively solve the problems of difficult right infringement, difficult evidence collection, difficult tracing, difficult compensation, difficult authorization transaction and the like of the traditional digital copyright. The system provided by the invention adds the works managed by the system into a block chain network system, realizes content chaining and copyright registration, obtains a unique ID which can not be tampered and is accurately proved by originality, proves the attribution and integrity of the copyright, and simultaneously records the unique ID on the chain.

(2) The work shows submodule piece, mainly include: uploading works, applying the works and the like.

Uploading works, such as: the writing of information such as the information and the type of the work, the storage of the work and the like. Uploading a work first needs to select the type of the uploaded work (including but not limited to pictures and videos), and "work information" needs to be filled according to certain rules: the name of the work is required to highlight the core of the work, the age and the name of the creator need to fill in real information, and the description of the work explains the idea of the expression of the work as much as possible. The corresponding works can be uploaded after the information is filled, the filling of the information is convenient to manage, the copyright is protected, and an important judgment basis is provided for the evaluation of the works. The cloud storage server is adopted for storing the works, the uploading speed of the works is high, the storage space is large, massive work data and character information can be stored, the evaluation algorithm is stored in the cloud server, the calling speed is high, the operation speed is high, and the evaluation result can be given in the fastest time.

1) Personalized exhibition halls, such as: user basic information, and works displayed by the user. The basic information of the user comprises information such as a nickname, age and the number of obtained praise. "works shown by the user" only show works set as public, and do not include personal works set as private, and the works are comprehensively sorted according to clicks, praise, comments, and gifts sent out.

2) Work applications, such as: and (5) decorating the works. The works decoration means that a plurality of different picture frames are selected for decorating the works, a client can adjust and select the decoration most suitable for the works of art, and the decorated works can be placed indoors or outdoors, on the wall of a living room or on a display wall of a painting exhibition. The selection is various, and the real application environment can be designed according to the selection.

(3) The work evaluation submodule mainly comprises functions of expert online comment, system AI intelligent comment, growth suggestion and the like.

1) The user and the teacher end can evaluate and score the artistic works in a four-dimensional evaluation mode, the only difference between the user end and the teacher end is that the expert can comment and guide the works only by the teacher end, the function is that the expert or the teacher can evaluate and give professional guidance to the works of the user on line, and the system can provide a one-to-one mode for communication between the user and the teacher.

2) The AI intelligent evaluation function can realize the automatic evaluation of the works based on the calculation aesthetics aiming at different artistic works and present the works in a visual form. Taking a picture work as an example, by using indexes such as color, vision, technique, spaciousness, modeling, innovation, master matching degree, work potential value and the like, a machine learning algorithm is utilized to design a corresponding evaluation method and a reference judgment standard, and finally an evaluation report is obtained. The specific method for evaluating the AI intelligent evaluation parameters comprises the following steps:

a. the Color and the Color parameter are the comprehensive evaluation of the tone, brightness, purity and Color richness of the picture, and the evaluation function is shown in formula (1). Wherein color is a color total evaluation value, H is a hue evaluation value, V is a lightness evaluation value, S is a purity evaluation value, and F is a color richness evaluation value; t is t₁～t₄The weight values of H, V, S and F, respectively, can be adjusted:

Color＝H*t₁+V*t₂+S*t₃+F*t₄ (1)

H. v, S can be calculated by opencv library functions, and F is the parameter of the most difficult to evaluate metric and cannot be directly calculated by the existing algorithm. The evaluation method for design F is as follows: let the eight basic hues be: r (1, 0, 0) red, G (0, 1, 0) green, B (0, 0, 1) blue, Y (1, 1, 0) yellow, M (1, 0, 1) magenta, C (0, 1, 1) cyan, K (0, 0, 0) black, W (1, 1, 1) white. Firstly, the Euclidean distance between each pixel of the detected image and each basic tone is calculated, and the basic tone with the minimum distance is the color of the pixel. Then, the pixel number ratio of each basic tone is counted, if the sum of the pixel ratio of the black and white two basic tones exceeds 90%, the picture is determined to be gray tone (black and white tone), otherwise, the picture is determined to be color tone. In the color tone, whether the pixel ratio of the remaining 6 basic tones exceeds 5% is judged, if so, the basic tone is judged to be a prominent color, and a corresponding F value is obtained according to the number of the prominent colors. The basic decision criteria are shown in table 1.

TABLE 1 basic evaluation criteria for F values

b. The Vision Vision 'Vision' parameter embodies the grain distribution condition of the picture, and the higher the evaluation value is, the richer and finer the grain of the picture is. The evaluation method is as follows: analyzing the texture distribution of the pictures from the visual texture, graying each picture, and performing LBP algorithm processing to obtain a corresponding texture image; then, the entropy of each texture image is calculated in a quantitative mode, and the larger the entropy value is, the richer the corresponding texture is. Experiments show that the approximate distribution range of the texture image entropy values of all pictures is between 0 and 7. There is also a rule for the distribution of texture image entropy values: when the picture is a pure color picture, for example, a pure white or pure black picture, no texture exists, and the entropy value corresponding to the texture map is 0. With the continuous improvement of the richness of the image texture, the entropy value of the corresponding texture map is continuously increased and continuously approaches to 7, and finally, normalization operation is performed to limit the numerical value between 0 and 1, wherein the numerical value is divided into seven grades on average, 0 is the lowest grade (namely the visual effect is the worst), and 1 is the highest grade (the visual effect is the best).

c. Technical Skill, the "technical" parameter embodies the composition capabilities of a picture composition. The common picture composition patterns include a straight line type, a triangle type, a polygon type, a circle type, an S type and the like, and pictures with different types of attributes have different picture composition modes. The picture is divided into two types of landscape painting and non-landscape painting for independent detection, the landscape painting mainly comprises a straight line type composition and an S type composition, the non-landscape painting has richer composition modes, and the detection method mainly comprises line detection. First, edge detection is performed on the picture, and then composition is detected in a classified manner. If the scene is a landscape picture, the whole picture is traversed in the inspection picture (3)4 width or height). If yes, the picture is a straight-line or S-shaped picture composition; otherwise, judging that the picture has no obvious shape. If the picture is a non-landscape picture, a straight line crossing the whole picture (3/4 width or height) in the picture is checked, and if the straight line exists, the picture is in a straight line shape. Otherwise, detecting four edge points of the top, the bottom, the left and the right on the edge picture, and connecting the four edge points clockwise to form a quadrilateral shape, if the length l of a connecting line between two points is too short, if

When the two points are combined, the midpoint is taken to form a triangular shape. The relevant methods and reference criteria are shown in table 2.

TABLE 2 basic evaluation criteria for "composition" parameters

d. The Space sense Space and the Space sense parameter embody the feeling of the real object in a picture approaching to the reality. The common types include single-layer shielding, multi-layer shielding, virtual-real (convergence, divergence, shape, brightness, cold and warm, pure gray) contrast, size perspective and the like. And evaluating the 'spatial perception' parameters of the work picture from the three aspects of the existence of occlusion, the existence of contrast and the existence of perspective. The picture is tiled without shielding and without contrast, so that the space sense is weak; the picture has single-layer shielding and no other contrast, namely the space sense is weak; the picture is shielded by multiple layers to have moderate spatial sense; the picture has multilayer shielding, and the contrast of virtual and real is obviously strong space sense; the picture has multilayer shielding, obvious contrast between virtual and real images and obvious size perspective, which means that the space sense is stronger, and the evaluation standard is shown in table 3.

TABLE 3 basic evaluation criteria for spatial impression

Spatial perception grade	Evaluation criteria	Evaluation results (points)
			Is weaker	No space collocation and single-side tiling	<60
Weak point of weakness	Single layer shielding and comfortable space	60-70
			Is moderate	Multiple layers of shielding, basically conforming to the size of the near part and the distance part	70-80
Is stronger	Multi-layer shielding and obvious perspective	80-90
			Is very strong	Obvious level contrast, virtual-real contrast and large-small perspective	90-100

Due to the large evaluation difficulty and various types, an experienced expert can directly judge the 'spatial sensation' condition of one picture by naked eyes, so that the manual evaluation can realize the evaluation of the 'spatial sensation' parameter, but the degree of the 'spatial sensation' is different from person to person, and the manual evaluation cannot realize 'fairness'. In addition, human eyes can easily judge the 'spatial sense' parameter by experience and sense, but the machine is difficult to describe formally and cannot be realized by a traditional machine learning algorithm. Therefore, the evaluation of the 'spatial sense' parameter is realized by adopting an automatic evaluation method based on calculation aesthetics of deep learning.

First, a training data set is prepared. The training picture set is divided into 5 levels by experts according to the basic evaluation criterion of the spatial perception in the table 3 for deep learning network training. Then, preprocessing each picture, carrying out noise reduction processing on the pictures by adopting Gaussian filtering to remove noisy redundant information in the pictures, extracting a gray image from the filtered images, and converting the pixel value of the obtained gray image into a pixel matrix with the size of 224 × 224. And performing convolution operation on the whole picture by adopting a convolution core of 3 x 3 to extract an image characteristic value, training by adopting a deep learning network resnet, wherein the number of the convoluted pictures, namely, batch _ size, is 16, the number of network cycle iterations, namely, epochs, is 200, 5 ganglion points corresponding to 5 grades are output by an output layer, each nerve node value is the probability of the classification to which the picture belongs, and the grade corresponding to the highest probability value is selected as a space sense grade. And after the training is finished, the training model is stored, when the system inputs a test picture, the test picture is sent into the model to judge the affiliated grade, and finally the grade is converted into an evaluation result.

e. The Model' parameter reflects the shape and volume of the picture content, and is commonly a point Model, a line Model, a surface Model and a point-line-surface Model. And carrying out edge detection on the picture, and then detecting the shape in a classified manner. Firstly, whether the area of the middle point and the circle of the whole picture exceeds 2/3 is detected, if not, the picture is judged to have no obvious shape, and if so, the picture is in a point shape or other shapes. And detecting the straight line of the whole picture in the picture, if the straight line is in a line type shape, then detecting whether all the lines are connected into a plane to form the line type shape, and if not, the straight line is in a point-line-surface type shape. The basic evaluation criteria of the "shape" are shown in table 4, and the "shape" parameters only have differences in category, and have no good or bad score, so the evaluation result is the detection result, and no score difference exists.

TABLE 4 basic evaluation criteria for the build

Class number	Detection standard	Evaluation results
			1	The area of the middle point and the circle of the whole picture is not enough 2/3	Without moulding
2	The area of the middle point and the circle of the whole picture exceeds 2/3	Point type modeling
			3	Straight line of picture	Line model
4	Stored online energy connected noodles	Surface type modeling
			5	2. 3 and 4 are both present	Dotted line surface type modeling

f. Innovativeness, the 'innovation' parameter is embodied in uniqueness of the picture in the aspects of texture, shape, color and the like. The method can be divided into dimensions of theme innovation, expression technique innovation, composition innovation, modeling innovation and the like. The parameter is similar to the 'spatial perception', the judgment of the picture 'innovation' parameter is easy to be made by the personal experience, but the machine is difficult to describe formally and cannot be realized by the traditional machine learning algorithm. Therefore, the evaluation of the 'innovative' parameters is also realized by adopting an automatic evaluation method based on calculation aesthetics of deep learning. The evaluation criteria are shown in table 5, with 5 levels according to the degree of "novelty". 20000 pictures of the works are manually evaluated and classified into the 5 grades respectively, and 5000 pictures are averaged for each grade. The training process is basically the same as the "sense of space" parameter, and the difference is two points: firstly, performing model training by adopting an Alex Net network; and secondly, training and learning the RGB three-channel colors by considering the colors of the images.

TABLE 5 basic evaluation criteria for innovations

Grade	Degree of innovation	Evaluation results (points)
			1	Is weaker	<60
2	Weak point of weakness	60-70
			3	Is moderate	70-80
4	Is stronger	80-90
			5	Is very strong	90-100

g. The master matching degree is evaluated by the master matching degree, the picture of the work is closest to the master or the famous person, the picture style of the author of the work is most similar to the picture style or the drawing method of the master or the famous person, the evaluation is abstract, and the evaluation result cannot be obtained by the traditional evaluation method. The invention adopts a 'deep learning' method, comprehensively obtains the similarity parameter values of the picture of the work and the work of which teacher from the aspects of composition, color, texture, space sense and the like, and the evaluation function is shown in a formula (2). Wherein, Simila is a general evaluation value of 'master matching degree', SSIM is a structure similarity evaluation value, CSIM is a color similarity evaluation value, TSIM is a texture similarity evaluation value, and PSIM is a space sense similarity evaluation value; t 1-t 4 are the weighting values of CSIM, V, S and F, respectively, and are adjustable, and the default values are all 0.25.

Simila＝SSIM*t₁+CSIM*t₂+TSIM*t₃+PSIM*t₄ (2)

Taking the composition similarity evaluation value as an example, the specific steps are as follows: firstly, dividing master works into four categories, namely people, animals, landscapes and the like, and then selecting a corresponding master works set according to the type of an evaluated picture; then, the structural similarity CSIM between the evaluated picture and each picture in the master work set is calculated_iThe maximum value is taken as CSIM ═ CSIM_maxThe master work corresponding to the maximum value is the final matching work, and the master is the best matching master.

In conclusion, under the combined action of the parameter algorithms, the evaluation result is presented in a visual form.

3) According to the evaluation result of each work, the invention provides an AI subentry evaluation description, an AI overall evaluation and an AI growth suggestion by using a recommendation algorithm, and finally provides a growth message of an individual. In the database designed by the invention, the explanatory sentences of each parameter in the 'AI subentry evaluation explanation' reach 1 ten thousand, and the explanatory words of 'AI overall comment', 'AI growth suggestion' and 'growth message' reach 2 ten thousand.

(4) The personal learning submodule:

professional learning mainly comprises: professional art knowledge expansion, art literacy and capability test and the like.

The professional art knowledge database is used for inputting the detailed information of the birth level, the genre, the creation characteristics and the like of thousands of famous art creators in the ancient and modern countries and drawing of most of the creators. Meanwhile, the system automatically pushes knowledge to the user every day through a design program, so that the user can know the creation style and the representative works of the author, and the knowledge plane and the visual field of the user are widened.

The artistic literacy and competence test presents the test questions in a simple and interesting manner in a classified manner, and multi-dimensional objective evaluation results and learning suggestions are given after the test is finished so as to promote the popularization of mass artistic knowledge.

(5) The communication interaction submodule comprises:

the communication interaction submodule mainly comprises: works communication, artistic activities and the like.

Work communication, such as: the social network platform is adopted in modes of 'praise, comment and encouragement', users can express the love of works of other people in interactive modes of leaving messages, commenting, praise, delivering gifts and the like, and the social network platform can communicate with creative mood and creative feeling. Artistic activities, such as: the combination of on-line and off-line of expert lectures, on-line lectures, art competitions, research and practice. The forms of expert lectures and online teaching are mainly effective and reliable ways for improving the user's artistic literacy, painting skill and the like. The artistic work competition and research and study practice create good creation opportunities for users, and achieve the purpose of improving personal artistic literacy.

The technical effects of the present invention will be described in detail with reference to experiments.

1. The system core function interface implementation condition is as follows:

2. AI automated assessment algorithm illustrates:

(1) color:

(2) The technique comprises the following steps:

(3) And (3) vision:

FIG. 14 is a diagram illustrating an estimated texture distribution according to an embodiment of the present invention.

It should be noted that the embodiments of the present invention can be realized by hardware, software, or a combination of software and hardware. The hardware portion may be implemented using dedicated logic; the software portions may be stored in a memory and executed by a suitable instruction execution system, such as a microprocessor or specially designed hardware. Those skilled in the art will appreciate that the apparatus and methods described above may be implemented using computer executable instructions and/or embodied in processor control code, such code being provided on a carrier medium such as a disk, CD-or DVD-ROM, programmable memory such as read only memory (firmware), or a data carrier such as an optical or electronic signal carrier, for example. The apparatus and its modules of the present invention may be implemented by hardware circuits such as very large scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, or programmable hardware devices such as field programmable gate arrays, programmable logic devices, etc., or by software executed by various types of processors, or by a combination of hardware circuits and software, e.g., firmware.

The above description is only for the purpose of illustrating the present invention and the appended claims are not to be construed as limiting the scope of the invention, which is intended to cover all modifications, equivalents and improvements that are within the spirit and scope of the invention as defined by the appended claims.

Claims

1. An intelligent analysis method for the value of an art work is characterized by comprising the following steps:

from different angles of color, vision, skill, spaciousness, modeling, innovation, master matching degree and artistic work potential values, an aesthetic perception calculation model is constructed by utilizing a machine learning and deep learning method, the aesthetic process is quantized and calculated, and the value of the artistic work under multiple scenes is intelligently analyzed.

2. The intelligent method of analyzing the value of an artwork of claim 1,

(1) the specific process of constructing the aesthetic perception calculation model is as follows:

3. The intelligent method of analyzing the value of an art work of claim 1, wherein the method comprises:

4. The intelligent work of art value parsing method of claim 3, wherein said intelligent parsing method further comprises:

(1) the colors are used for showing the expressive force of the works, and the functions are used for comprehensively evaluating the tone, the lightness, the purity and the color richness of the works;

Color＝H*t₁+V*t₂+S*t₃+F*t₄；

wherein color represents a color total evaluation value, H represents a hue evaluation value, V represents a lightness evaluation value, S represents a purity evaluation value, and F represents a color richness evaluation value; t is t₁～t₄Weight values representing H, V, S and F, respectively; H. v, S are calculated by opencv library functions;

eight base hues were determined to be: r (1, 0, 0) red, G (0, 1, 0) green, B (0, 0, 1) blue, Y (1, 1, 0) yellow, M (1, 0, 1) magenta, C (0, 1, 1) cyan, K (0, 0, 0) black, W (1, 1, 1) white; calculating the Euclidean distance between each pixel of the detected image and each basic tone, wherein the basic tone with the minimum distance is the color of the pixel; counting the pixel number ratio of each basic tone, if the sum of the pixel ratio of the black basic tone and the pixel ratio of the white basic tone exceeds 90%, determining that the picture is a gray tone, otherwise, determining that the picture is a color tone; in the color tone, judging whether the pixel ratio of the remaining 6 basic tones exceeds 5%, if so, judging that the basic tones are prominent colors, and obtaining corresponding F values according to the number of the prominent colors;

(2) the vision is used for representing the texture distribution condition of the work, and the higher the evaluation value is, the richer and finer the texture of the work is; analyzing the texture distribution of the pictures from the visual texture, graying each picture, and performing LBP algorithm processing to obtain a corresponding texture image; the entropy of each texture image is calculated in a quantitative mode, and the larger the entropy value is, the richer the corresponding texture is; normalizing, limiting the numerical value to 0-1, averagely dividing the numerical value into seven grades, wherein 0 is the lowest grade and has the worst visual effect, and 1 is the highest grade and has the best visual effect;

(3) the technique is used for representing the composition capability of the work, and dividing the picture into two types of landscape painting and non-landscape painting for independent detection, wherein the landscape painting mainly comprises a linear and S-shaped composition, and the non-landscape painting mainly comprises line detection; the method specifically comprises the following steps: carrying out edge detection on the picture, and carrying out classified detection and composition; if the scene is a landscape painting, the whole part of the picture is checked to spanA straight line or a curve of the width or the height of the picture 3/4; if yes, the picture is a straight-line or S-shaped picture composition; otherwise, judging that the picture has no obvious shape; if the picture is a non-landscape picture, checking a straight line which spans the whole picture, namely 3/4 width or height in the picture, and if the straight line exists, forming a straight line shape; otherwise, detecting four edge points of the top, the bottom, the left and the right on the edge picture, connecting the four edge points clockwise to form a quadrilateral shape, and if the length of a connecting line between two certain points is too short, when the length of the connecting line is too short

(4) the spatial impression is used for representing the degree impression of the real object in the work approaching to reality, and the spatial impression parameters of the picture of the work are evaluated from the aspects of occlusion, contrast and perspective by adopting an automatic evaluation method based on the calculation aesthetics of deep learning; preparing a training data set; dividing a training picture set into 5 grades by an expert according to a preset spatial perception basic evaluation criterion for deep learning network training; preprocessing each picture, performing noise reduction processing on the pictures by adopting Gaussian filtering, extracting a gray image from the filtered image, and converting the pixel value of the obtained gray image into a pixel matrix with the size of 224 × 224; performing convolution operation on the whole picture by adopting a convolution core of 3 x 3 to extract an image characteristic value, training by adopting a deep learning network resnet, wherein the number of the convoluted picture per time, namely, batch _ size, is 16, the number of network cycle iteration times, namely, epochs, is 200, an output layer outputs 5 ganglion points corresponding to 5 levels, each neural node value is the probability of the classification to which the picture belongs, and the level corresponding to the highest probability value is selected as a spatial sensation level; after the training is finished, storing the training model, outputting the work to be evaluated from the trained model to judge the belonging grade, and converting the grade into an evaluation result;

(5) the shape is used for representing the shape and the volume of the content of the work; carrying out edge detection on the picture, and detecting and modeling in a classified manner; detecting whether the area of the middle point and the dot circle of the whole picture exceeds 2/3, if not, judging that the picture has no obvious shape, and if so, judging that the picture has a dot shape or other shapes; then detecting the straight line of the whole picture in the picture, if the straight line is in a line type shape, detecting whether all the lines are connected into a plane to form the line type shape, and if not, the straight line is in a point-line-surface type shape;

(6) innovativeness is used for reflecting novelty of the content of a work, and an automatic evaluation method based on computational aesthetics for deep learning is adopted for evaluating theme innovation, expression technique innovation, composition innovation, modeling innovation and other dimensions; dividing the images into 5 grades according to the innovation degree, manually evaluating 20000 work images, classifying the 5 grades respectively, and averaging 5000 images in each grade; performing model training and training learning of RGB three-channel colors by adopting an Alex Net network, and outputting a trained model to evaluate the work to be evaluated;

(7) the master matching degree is used for measuring which master or famous family works are closest to which master or famous family works, and the painting style representing the author of the works is most similar to the painting style or the painting method of which master or famous family works; a deep learning method is adopted, and the similarity parameter values of the picture of the work and the work of which teacher are obtained comprehensively from the aspects of composition, color, texture, space and the like; the evaluation function is as follows:

Simila＝SSIM*t₁+CSIM*t₂+TSIM*t₃+PSIM*t₄；

wherein, Simila represents the total evaluation value of the master matching degree, SSIM represents the structure similarity evaluation value, CSIM represents the color similarity evaluation value, TSIM represents the texture similarity evaluation value, and PSIM represents the space sense similarity evaluation value; t 1-t 4 respectively represent the weight values of CSIM, V, S and F;

dividing the master work set into four types, namely, people, animals, landscapes and the like, and selecting the corresponding master work set according to the type of the evaluated picture; calculating the structural similarity CSIMi between the evaluated picture and each picture in the master work set, and taking the maximum value as CSIM ═ CSIM_max(ii) a The master work corresponding to the maximum value is the final matching work, and the master is the best matching master.

5. A computer-readable storage medium storing a computer program which, when executed by a processor, causes the processor to perform the steps of the method of intelligently analyzing the value of a work of art of any one of claims 1 to 3.

6. An artwork value information processing system for implementing the artwork value intelligent analysis method of any one of claims 1 to 3, wherein the artwork value information processing system comprises:

7. The artwork value information processing system of claim 6, wherein said information management subsystem comprises:

8. The artwork value information processing system of claim 6, wherein said work presentation subsystem comprises:

9. The work of art value information processing system of claim 6 wherein the work evaluation subsystem comprises:

10. The artwork value information processing system of claim 6, wherein said professional learning subsystem comprises:

the communication interaction subsystem comprises: