Exploring the dynamic relationship between oil painting art and modern social values by combining data mining techniques

European countries are the origin of oil painting, according to historical records, as early as in the Ming and Qing Dynasties, oil painting was imported into China, and the real emergence of the Western academy style in the country originated from the 1950s when domestic students went to France to study oil painting [1–2].

In the 1960s, the Central Academy of Fine Arts opened the first studio studio in China. From the Soviet Union, students returned home to start China’s colleges and universities of art education. In many years of practical exploration, China’s art education model can be formed [3–4].

Since the reform and opening up, social development has stepped into a new stage; oil painting in Chinese colleges and universities opened up the road of diversified development, the emergence of diversified painting styles, presenting the scene of “a hundred schools of thought” [5].

At present, Chinese oil painting is in a new stage of development, and it gradually has the value standard of Chinese cultural characteristics, but from the perspective of the creation of Chinese oil painting, there are still certain problems in its creation, such as oil paintings that are difficult to be understood and appreciated, and the oil painting art lacks a certain sense of direction in its expression [6–7]. Although the performance of art and culture is more abstract, the actual artistic value it has needs to be recognised and understood by the public. With the development and popularisation of network technology, entertainment culture, popular culture and consumer art occupy a dominant cultural position in modern society, so oil painting art with Chinese characteristics gradually loses the form and significance of its existence, and its social value is gradually ignored by the public and society [8–10]. Therefore, in the face of changes in the art environment of the current market and changes in social appreciation of oil painting, we need to reexamine the value of Chinese oil painting art, continue to explore and study the path of oil painting development and its artistic value standards, and reposition oil painting art in a diversified social environment [11–12].

As a branch of human culture, art is important in all aspects. Spiritually speaking, art satisfies people’s aesthetic needs in the form of artistic beauty and gives them spiritual enjoyment and pleasure. From the sociological point of view, it realizes its memory through the ways and means of art. And these ways or methods include “oil painting” [13–14]. Oil painting is a very meaningful carrier of memory, which carries the life condition of the author in the period when he created it. Memory includes suffering but also the celebration of a better life and the implicit inspiration of the work. Each work leaves behind a unique memory of its time. In addition to the memory of the past, many oil paintings also reflect the social contradictions that exist in the present, providing people with warnings and reflections, as well as many warm-hearted places in the present society [15–16]. In addition to the artistic appreciation of the picture, an oil painting also contains a rich humanistic spirit, such as family, friendship, love, faith, responsibility, etc., showing the creator’s awe for the world and people. As a kind of art, the meaning of oil painting is mainly divided into three types. The first is decorative, as the name suggests, which is the most superficial value of the existence of oil painting, followed by the times. The creation of oil painting is closely related to the background of the times in which it is located; it is a true portrayal of the social environment at that time, and it has irreproducible characteristics, and the second is artistry, which is more concerned with the painting skills and artistic conception of the works than decoration, and people have an appreciative eye for this beauty [17–20].

In this paper, the effective colour features of the key regions of the oil painting are extracted, and the Fisher Score algorithm is used to evaluate the degree of correlation between the features of each colour feature region and the class to which it belongs and to describe the relevant features of the oil painting. The grey-scale covariance matrix extraction method is used to extract texture features from the oil painting, and important indexes such as energy and entropy are selected for measurement. On the basis of the Extreme Learning Machine algorithm, the regularisation factor is introduced to construct the RELM decision model to represent the feature vector of the oil painting image. Using a simple Bayesian classifier, the eigenvalues of each oil painting are used to calculate the probability that the oil painting represents the work done by a certain painter. Simulation experiments are used to verify the effectiveness of the proposed algorithm in extracting color and texture features from oil paintings. Based on the analysis results, the dynamic relationship between oil painting art and modern social values is explored.

2

Oil painting art colour characteristics and texture characteristics mining

2.1

Description of key areas for oil painting

2.1.1

Colour characteristics

Extracting effective features is the key to the description of key regions, and the key regions with high information richness are selected, on the basis of which the effective features of key regions are extracted, that is, the features of an oil painting image can be obtained. Since the oil painting image has rich colours, so this paper extracts the R, G, B, H, S, V colour histogram features of the oil painting image (including colour mean μ variance σ², tilt μ_s, craggy μ_k, energy μ_N) and the texture features formed by the colour changes (including contrast CON, angular direction second-order moments ASM, colour ENT, mean MEAN) and the computational formulas are as follows, respectively: 1 $μ = \sum_{i = 0}^{255} i P (i)$ 2 $σ^{2} = \sum_{i = 0}^{255} {(i - μ)}^{2} P (i)$ 3 $μ_{s} = \frac{1}{σ^{3}} \sum_{i = 0}^{255} {(i - μ)}^{3} P (i)$ 4 $μ_{k} = \frac{1}{σ^{4}} \sum_{i = 0}^{255} {(i - μ)}^{4} P (i)$ 5 $μ_{N} = \sum_{i = 0}^{255} P^{2} (i)$

Where p(i) is the frequency of occurrence of the i nd colour in the image: 6 $C O N = \sum_{i} i^{2} p (i)$ 7 $A S M = \sum_{i} [^{p (i)] 2}$ 8 $E N T = - \sum_{i = 0}^{255} p (i) \log p (i)$ 9 $M E A N = \frac{1}{b} \sum_{i} i p (i)$

Where: P(i) is the frequency of occurrence of grey level difference g_Δ(x,y) at level i, and all possible values of g_Δ(x,y) are at level b. Taking any point (x,y) within the image, the grey level difference between that point and its adjacent point (x + Δx,y + Δy) is calculated as: 10 $g_{Δ} (x, y) = g (x, y) - g (x + Δ x, y + Δ y)$

According to Eqs. (1) to (10), the above 9 features of the 6 colour components H, S, V, R, G, B of the oil painting are calculated respectively, and there are 6×9 = 54 features in total, which are denoted as A_i (i = 1,2,⋯,54) respectively. Then each oil painting can be described as a 54- dimensional vector F = (f₁,f₂,⋯,f_n), where: n = 54, f_i is the measure of feature A_i.

2.1.2

Fisher Score Feature Selection

An oil painting has a great number of colour features, but since some of the colour features of an oil painting are not useful for the classification of oil paintings, they even interfere with the classification results. In order to be able to classify oil paintings more accurately and eliminate the features that interfere with the description of oil paintings, this paper evaluates the extracted features and selects the excellent features. Fisher Score measures the correlation between each feature and its class, and the more correlation there is between the feature and its class, the better it is. The higher the degree of association, the higher the score of the feature, and the lower the degree of association, the lower the score of the feature. So in this paper, the Fisher Score is used to evaluate each feature [21]: 11 $W_{f_{i}} = \sum_{c = 1}^{N} (\bar{T_{f_{i}}} - \bar{T_{f_{i}, c}}) / \sum_{c = 1}^{N} σ_{f_{i}, c}^{2}$

Where: W_fi is the weight score of f_i corresponding to feature A_i, reflecting the influence weight of feature A_i on the classification, W_fi the larger the feature the more important, N is the number of categories (number of painters), $\bar{T_{f i}}$ is the mean value of feature A_i eigenvalues f_i in all the training sets, $\bar{T_{f_{i}}, c}$ is the mean value of feature A_i eigenvalues f_i in the training set of category c, and $σ_{f_{i}, c}^{2}$ is the variance of eigenvalues f_i in the training set of category c. The larger the W_fi of feature A_i, the closer the eigenvalues between different paintings of the same painter are, and the larger the eigenvalue gap between paintings of different painters is.

The features extracted in section 2.1.1 are evaluated by equation (11), and the first n' features with 80% importance share are selected to describe the key regions to get the description $F = ({f^{'}}_{1}, {f^{'}}_{2}, \dots, {f^{'}}_{n'})$ of oil paintings.

2.1.3

Texture characteristics

In order to achieve oil painting recognition, this paper applies the grey scale co-production matrix method to extract the texture features of oil paintings [22]: assuming that Q is the set of pixel pairs with some spatial connection in the target region R, the mathematical expression of the coproduction matrix P is equation (12): 12 $P (l_{1}, l_{2}) = \frac{\sum {[(x_{1}, y_{1}), (x_{2}, y_{2})] \in Q | f (x_{1}, y_{1}) = l_{1} & \cdot f (x_{2}, y_{2}) = l_{2}}}{\sum [(x_{1}, y_{1}), (x_{2}, y_{2}) \in Q]}$

Where the numerator and denominator are the number of pixel pairs with grey value of l₁, l₂ and the total number of pixel pairs are Eq. (13) Eq. (16), respectively: 1)

Energy: 13 $L r = \sum_{l_{1}} \sum_{l_{2}} P^{2} (l_{1}, l_{2})$

2)

Entropy: 14 $L e = \sum_{l_{1}} \sum_{l_{2}} P (l_{1}, l_{2}) \log P (l_{1}, l_{2})$

3)

Contrast: 15 $L d = \sum_{l_{1}} \sum_{l_{2}} {(l_{1} - l_{2})}^{2} P (l_{1}, l_{2})$

4)

Local homogeneity: 16 $L j = \sum_{l_{1}} \sum_{l_{2}} \frac{P (l_{1}, l_{2})}{1 + | l_{1} - l_{2} |}$

2.2

Regularised limit learning machine

The mathematical model of the Extreme Learning Machine (ELM) is equation (17) [23]: 17 $f_{L} (x) = \sum_{i = 1}^{L} β_{i} G (a_{i}, b_{i}, x)$

Where L is the number of hidden layer nodes, β_i is the weight coefficient of the i rd hidden layer node for the output node, where β_i ∈ R, G(a_i,b_i,x) is the output function of the i th hidden layer node, a_i, b_i are the input weight and node bias of the i th hidden layer node, respectively, where a_i ∈ Rⁿ, b_i ∈ R, G(a_i,b_i,x) the output function can be expressed as equation (18): 18 $G (a_{i}, b_{i}, x) = g (a_{i} \cdot x + b_{i})$ where g(·) is the activation function. variables and output variables, where, x_l ∈ Rⁿ, y_l ∈ R^m, then Eq. (19): !19 $y_{l} = \sum_{i = 1}^{L} β_{i} G (a_{i}, b_{i}, x_{l}), l = 1, 2, \dots, N$

The matrix form of equation (19) is equation (20): 20 $H β = Y$

The formula is as in equation (21): 21 $\begin{matrix} H = [\begin{matrix} G (a_{1}, b_{1}, x_{1}) & \dots & G (a_{L}, b_{L}, x_{1}) \\ ⋮ & ⋮ & ⋮ \\ G (a_{1}, b_{1}, x_{N}) & \dots & G (a_{L}, b_{L}, x_{N}) \end{matrix}] = [\begin{matrix} h (x_{1}) \\ ⋮ \\ h (x_{N}) \end{matrix}] \\ β = {[\begin{matrix} β_{1} \\ ⋮ \\ β_{L} \end{matrix}]}_{L \times_{m}}, Y = {[\begin{matrix} y_{1} \\ ⋮ \\ y_{N} \end{matrix}]}_{N \times_{m}} \end{matrix}$

Since a_i,b_i is randomly generated and only β is obtained by ${(x_{l}, y_{l})}_{i = 1}^{N}$ computation, in order to improve the accuracy of the ELM, the ELM introduces a regularisation factor λ, at which point solving β is transformed into the following optimisation problem as equation (22): 22 $\begin{array}{l} \min_{β} : V_{E L M} = \frac{1}{2} {‖ β ‖}^{2} + \frac{λ}{2} \sum_{l = 1}^{N} {‖ ε_{l} ‖}^{2} \\ s . t . : h (x_{l}) β = y_{l} - ε_{l}, l = 1, 2, \dots, N \end{array}$

Where ε_l is the training bias, which mainly serves to avoid the overfitting problem. According to the Ka-rush-Kuhn-Tucker condition, the optimisation solution problem of Eq. (23) can be converted into: 23 $V_{E L M} = \frac{1}{2} {‖ β ‖}^{2} + \frac{λ}{2} \sum_{l = 1}^{N} {‖ ε_{l} ‖}^{2} - \sum_{l = 1}^{N} \sum_{j = 1}^{m} a_{l, j} (h (x_{l}) β_{j} - y_{l, j} + ε_{l, j})$

The optimal solution of β is calculated from Eq. (23) as Eq. (24): 24 $β = H^{T} {(\frac{I}{λ} + H H^{T})}^{- 1} Y$

From Eq. (24), β in RELM is mainly determined by matrix H, matrix Y and regularisation factor λ in Eq. (21), where the dimension of H is related to the number of nodes of the implicit layer L and the number of training samples N, and Y is the output of the training samples, and the performance of the RELM is affected by the choice of the parameters of the number of nodes of the implicit layer L and the regularisation factor λ, since N and Y have been determined.

In case of binary classification problem, the decision model of RELM is equation (25): 25 $f (x) = s i g n (h (x) H^{T} {(\frac{I}{λ} + H H^{T})}^{- 1} Y)$

In case of a multiclassification problem, the decision model for RELM is equation (26): 26 $l a b e l (x) = \underset{i \in ≺ | (1, 2, \dots, m) | ≻}{\arg \max} f_{i} (x)$

Where f(x) is the value of the first output node and f_i(x) = [f₁(x),f₂(x),⋯f_m(x)]^T.

2.3

Oil painting feature recognition based on colour features and texture features

2.3.1

Oil painting feature extraction vector

For the oil painting image samples, the oil painting colour features and texture features are extracted and the feature vector feature(i,j) of the oil painting image is equation (27): 27 $f e a t u r e (i, j) = {L r, L e, L d, L j, μ, σ, ξ}$

The oil painting recognition process of RELM based on colour features and texture features can be specifically described as follows: 1)

Read the oil painting image sample data.

2)

Extract oil painting image texture features and colour features feature(i,j).

3)

The category coding of the oil painting.

4)

Divide the feature vector feature(i,j) of the oil painting image into training samples and test samples, take the feature vector feature(i,j) of the oil painting image of the training samples as the input of RELM, and the categories of the oil paintings of the training samples as the output of RELM, and establish the RELM model.

5)

For the test samples of oil painting images, the RELM oil painting recognition model is applied for oil painting recognition.

2.3.2

Feature validation

In order to verify the validity of the art style features extracted by the method in this paper, this paper uses the extracted features to classify oil paintings. Park Bayes is widely used in text classification and painting classification, and the results of its classification are usually better than other classifiers, this section briefly introduces the Park Bayes classifier.

A plain Bayesian classifier is one of the classical machine learning algorithms, which is simple in principle, easy to implement, performs well on small-scale data, and is widely used for text classification [24]. The principle of Plain Bayes is to calculate the posterior probability using conditional probability and prior probability. Each oil painting is represented by a n -dimensional feature vector F = (f₁,f₂,⋯,f_n), and the set of painters is C = (c₁,c₂,⋯,c_k). Plain Bayes classifies oil paintings, i.e., using each feature f_i of an oil painting to calculate the probability that an oil painting F is made by a particular painter c_j, calculated as follows: 28 $P (c_{j} | F) = \frac{P (c_{j}) P (F | c_{j})}{P (F)} = \frac{P (c_{j}) P (f_{1}, f_{2}, \dots, f_{n} | c_{j})}{P (f_{1}, f_{2}, \dots, f_{n})}$

Where P(f₁,f₂,⋯,f_n) is constant for all categories.

For F, the greater the posterior probability P(c_jF), the more likely it is that F belongs to c_j. The formula for the category to which F belongs is given below: 29 $c (F) = \underset{c_{j} \in C}{\arg \max} P (f_{1}, f_{2}, \dots, f_{n} | c_{i}) P (c_{j})$

Since each feature is independent of each other, equation (29) can be transformed into: 30 $c (F) = \underset{c_{j} \in G}{\arg \max} P (c_{j}) \prod_{j = 1}^{n} P (f_{i} | c_{j})$

The category of F is obtained by returning the category with the largest a posteriori probability after calculating P(c_j) and P(f_i|c_j) respectively.

3

Oil Painting Artistic Characteristics Analysis

This paper selects two oil paintings, “Sleeping Venus” and “Unknown Woman”, from many oil paintings, and analyzes them specifically from the aspects of color characteristics and texture characteristics respectively.

3.1

Colour characteristics

3.1.1

Colour histogram

The histogram average method is too rough, while the histogram intersection method and the histogram matching method are too large. A compromise method is to use a combination of image colors and reference colors. This set of reference colors should be able to cover the visual perception of various colors. The number of reference colors should be less than the original image, generally using the clustering method, the number of colors to reduce the number of clusters, and then the clustering of each color through the color distance formula calculated in the color table and its distance from the nearest color [25]. In this way, a simplified histogram can be obtained, combined with the matching eigenvector formula, resulting in a color histogram as shown in Fig. 1, Figs. (a)-(c) are the histograms of Sleeping Venus and Figs. (d)-(f) are the color histograms of Unknown Woman. As can be seen from the figures, the red histogram of “Sleeping Venus” has the highest pixel value of 1880 when the luminance is in the range of 10-15. On the contrary, the red histogram of “Unknown Woman” has the lowest pixel value of 184 when the luminance is in this range. The values of the green and blue histograms in the two paintings are similar.

3.1.2

Colour feature extraction

An oil painting image contains up to 224 kinds of colours. The extraction of image features involves using fewer colors to represent the original image, which can reflect the basic appearance of the image. An image contains tens of thousands of colours. It is not possible to extract the characteristics of these tens of thousands of colors. There can only be extracted by the method of the main colour representing the original image. The main colour extraction is generally used in a clustering method, the quantization of the uniform colour space (quantization of the following are carried out in the HSV space), partitioned into a number of subspaces, and the use of histograms statistics, statistics of each colour space contained in the image to be extracted. The number of pixels in each color space is counted to extract the features of the image, and the peaks in the histogram are extracted to obtain the dominant color of the image.

The extraction of colour features (dominant colour) is generally done using a colour clustering algorithm. Figure 2 and Figure 3 for the basic principles of colour clustering. Figure 2 shows the distribution of all pixels of “Sleeping Venus”in RGB colour space, and Figure 3 shows the distribution of pixels of “Sleeping Venus”in HSV colour space. Here, the quantization level of R, G and B are all taken as 64, and the quantization level of H, S and V are taken as 50. Notice that the H component in HSV space is continuous, i.e., level 1 and level 50 are adjacent to each other in terms of colour (both red), and it can be seen that the pixel points are clustered in both colour spaces, but the clustering patterns are not the same. The distribution of pixel points in the RGB space is more tight and concentrated; the three axes are of equal status, and the anisotropy is different. The three axes are equal and isotropic; G and B are in the interval [0,60], and R is in the interval [0,50]. In contrast, the distribution of pixels in HSV space is somewhat regular, and the projections of pixel points on the H-axis are all concentrated around a few values, which is quite different from the nature of the S-axis and the V-axis, i.e., the distributions are anisotropic.

3.2

Texture characteristics

3.2.1

Grey scale symbiotic features and texture features

The energy (ASM) is a feature that reflects the homogeneity of the image distribution theoretically. Large values indicate more homogeneous and regular texture characteristics, i.e., the image has a coarse texture: the smaller the ASM, the finer the texture. Entropy (ENT) is a measure of the content and information content of an image and is a reflection of the uniformity and complexity of the image texture. The larger the entropy, the finer the texture. On the contrary, the coarser the image texture. Contrast (CON) is the degree of clarity of the image texture. The larger the CON, the finer the texture, and the clearer the image.

On the contrary, the smaller the contrast, the coarser the texture and the less clear the image. Local Uniformity (LOC) measures the similarity of the grey scale co-occurrence matrix elements in the row or column direction. The larger the correlation value, the coarser the image texture.

Conversely, the finer the image texture. Table 1 shows the greyscale symbiotic features and texture features of the two oil paintings, and three parts of the two oil paintings are randomly selected for the calculation of the eigenvalues, F_crs, F_con, F_dlr, and F_reg are roughness, contrast, directionality and regularity, respectively. The results of the calculated data show that only the entropy and contrast values of “Sleeping Venus” are more consistent with the subjective judgement of human vision, and the entropy and contrast of part 3 are 2.065 and 0.317 respectively. The energy and correlation values are less consistent with subjective judgments of human vision.

Table 1.

The characteristics and texture characteristics of the grey scale of the two paintings

The Venus of falling asleep
Site	Energy	Entropy	Contrast ratio	Local uniformity
1	0.03486	1.34948	0.09156	0.00496
2	0.38654	1.42654	0.17962	0.00612
3	0.26526	2.06534	0.31654	0.00935
Site	F_crs	F_con	F_dlr	F_reg
1	15.26546	69.48863	32.78952	0.91566
2	14.49663	55.34969	33.45669	0.96163
3	11.26544	57.15693	34.04626	0.99154
Anonymous girl
Site	Energy	Entropy	Contrast ratio	Local uniformity
1	0.26896	1.76563	0.07965	0.48622
2	0.22645	1.46236	0.07636	0.03486
3	0.27956	1.52364	0.09356	0.14689
Site	F_crs	F_con	F_dlr	F_reg
1	17.48863	31.56548	31.76453	0.94856
2	16.15369	23.49587	40.19569	0.98463
3	16.05645	30.48648	26.48698	0.97565

The texture roughness value characterises the texture roughness characteristics of oil paintings, and the actual texture characteristics perceived by human subjective vision are extremely consistent, i.e., the calculated texture roughness value of oil paintings with visually significantly larger sensory characteristics is significantly larger. In other words, the calculated texture roughness values of oil paintings with visually large sensory characteristics are significantly larger than those of images with visually small sensory characteristics. The results of the calculation of the textural features of The Unknown Woman show that Part 1 has the highest roughness and contrast, 17.489 and 31.565, respectively, and Part 2 has the highest directionality and regularity, 40.196 and 0.985.

3.2.2

Texture feature extraction accuracy

Figure 4 shows the accuracy curve for texture feature extraction through experimental results. It can be judged that, whether it is artificial texture or natural texture, with the change of the distance parameter, the extraction accuracy also changes together and conforms to a certain law. However, the difference is that in the artificial texture, the extraction accuracy shows a periodic change every 6 feature points for a cycle to reach the lowest value of the correct rate. The two lowest values are 0.822 and 0.793. Similar to the grey scale distribution law of the artificial texture and in the natural texture, the classification accuracy of the overall trend of the decreasing trend, in the fifteenth feature point, to maintained at about 0.7. And the classification accuracy reflects the strength of the algorithm’s ability to describe the texture under different parameter values to a certain extent. Therefore, considering the above factors, the distance parameter is finally selected as 1.

3.2.3

Grey-scale-gradient covariance matrices

Table 2 shows the four types of grey gradient covariance matrix eigenmeans, listing the changes in the image to be eigenvalues before and after the rotation. The number of grey levels of the image is generally 256 grey levels. The calculation, in order to reduce the amount of computation, the image texture interval compression, and the experimental results have a small impact on the case. This thesis took 16 grey levels. The specific process of the algorithm is: (1) Get the grey matrix of the image N×M, calculate the gradient matrix of the image by using the 3×3 Sobel template, and get the gradient matrix of N×M by calculation. (2) The grey matrix and grey gradient matrix are divided into many 16×16 grey-gradient covariance matrices, so that the maximum and minimum difference of grey scale and the minimum and maximum difference of gradient are not more than 16. Quantification with equal probability greatly reduces the amount of computation. The eigenmeans of the eigenquantities ASM, CON, CORRLN, and IDM after rotating them by 90° are 0.047, 28.766, 0.125, and 0.18.

Table 2.

Four mean gray gradient symbiotic matrix characteristics mean

Characteristic quantity	Original drawing	Rotations 90°	Rotations 180°	Rotations 270°	σ / μ/(%)
ASM	0.05489	0.04653	0.05649	0.05696	0.91652
CON	28.46665	28.76565	28.34856	28.48632	0.34896
CORRLN	0.17895	0.12549	0.15796	0.24896	18.56486
ENT	3.52948	3.51896	3.58654	3.51686	0.25486
IDM	0.16563	0.17969	0.17952	0.17896	0.24988

3.3

Oil painting style identification

Fig. 5 shows the prediction results of oil painting styles. The classification results are plotted in a manner where the predicted labels overlap with the actual labels to indicate the correct classification, while the highlighted nodes indicate the images that have been mistakenly classified. The horizontal coordinate indicates the style painting number, where 1-50 is realistic, 51-100 is light intense, 101-150 is flat colour block, 151-200 is strong brushstroke, 201-250 is colourful, 251-300 is abstract, and 301-350 is thick paint stack. The vertical coordinate indicates the classification of style painting. From the prediction results, it can be seen that among the 350 oil paintings of different styles, 305 were correctly classified and 45 were incorrectly classified, with a prediction accuracy of 87.143%.

3.4

The dynamic relationship between oil painting art and modern social values

3.4.1

Embodiment of the value dimension and the construction of artistic images

Wealth and strength refer to the wealth of the State and the strength of the people, and it is only on the basis of the realization of wealth and strength that the Chinese nation can maintain the prosperity of the State and the happiness and well-being of the people. As the song goes, “Without a strong country, there can be no rich family.” The wealth of the country and the strength of the people are some of the important and enduring goals of the Chinese nation and are like a guiding light pointing out the direction for the construction of socialism. Democracy is the people’s aspiration for a better life. The goal of democracy in China is to pursue the people’s democracy and realize that the people are the masters of their own house, which is also the source of life of socialist construction, providing a guarantee for the people to create and share a happy life. Civilization is a state that should be present in the construction of socialism, and it is also strong support for the Chinese nation’s adherence to the road of rejuvenation. Harmony is a value demand in socialist construction, which has an important impact on maintaining social harmony and stability, while also ensuring the sustainable development of society.

The oil paintings record a series of important events of the Chinese nation, which are the symbols of the Chinese nation’s road to success, symbolizing the wealth, strength and democracy of the Chinese nation. Chinese Memory and Chinese Dream deeply expresses the strong desire of the Chinese nation for a hundred years for national wealth and strength, national rejuvenation, and people’s happiness, which can attract a sense of empathy from the audience. Based on this, oil paintings can give the children and grandchildren of the Chinese nation a deep memory and achieve the purpose of revitalizing China, which is conducive to the construction of the national artistic image.

3.4.2

Embodiment of the social dimension and the construction of an artistic identity

Freedom refers mainly to the people’s freedom, including freedom of will, development and other aspects; at the same time, freedom is also the best aspiration of human beings and is the value goal of building a socialist society. Equality is a priority for all citizens, with a focus on the equality of citizens before law, respect for human rights, protection of human rights, and granting equal rights to individuals in society. Justice refers to the maintenance of social fairness and justice, in the series of social life, only the full maintenance of justice, in order to maximize the satisfaction of the construction of a harmonious socialist society. The rule of law is to adhere to the principle and strategy of ruling the country according to law, safeguarding the legitimate rights and interests of citizens and maintaining social harmony and stability through the construction of the legal system.

For example, the oil painting “Love Devotion” by Jiang Yuanqiang was created in 2014. The characters in work include Peng Liyuan, the “Mother of the Nation”, who has zero-distance contact with children from special groups and feeds them, which is worthy of learning by the public. Just like the lyrics of the song “Love Devotion”: “If everyone gives a little love, the world will become a better place”. Through this work, the concept of “equality” in the socialist core values can be manifested, and it has the function of ideological guidance and behavioral guidance for the public. Based on this, modern oil paintings can help to construct a national artistic image.

3.4.3

Embodiment of the Personal Dimension and the Construction of an Artistic Identity

Patriotism is a relationship of dependence and an emotional attitude of citizens towards their country, and an important indicator of the motherland and the individual’s code of conduct. In the construction of a socialist society, citizens should take the revitalization of China as their responsibility, express strong patriotic feelings, adhere to the unity of the motherland, and maintain national unity. Dedication is the embodiment of a citizen’s behavioral attitude toward their profession, which requires citizens to be able to be conscientious and dedicated to their duties at their workplaces and to actively contribute to the development of society. The term integrity refers to honesty and trustworthiness, which are requirements for citizens’ morality. It emphasizes the keeping of promises and sincerity towards others, which is conducive to the creation of a harmonious society. Friendship focuses on emphasizing mutual respect, care, and help among social citizens and is a norm for interpersonal relationships in society.

Taking Wu Houxin’s oil painting “Chinese Memory and Chinese Dream” as an example, it involves relatively more content, and also highlights the theme of patriotism more. First of all, the oil painting describes the historical events of China’s ban on smoking and the elimination of tobacco in 1840. Under the leadership of Lin Zexu, the “Humen Smoking Cessation” was carried out to formally challenge the Western powers, which demonstrated the deep patriotic feelings in the hearts of Chinese children. Secondly, the Eight-Power Allied Forces invaded China and burned the Yuanmingyuan, leaving the remnants of the stone pillars standing in the Yuanmingyuan. Mr. Wu took the remnants of the stone pillars as the material for his oil paintings, which is not only the iron evidence of the invasion of Chinese territory by the great powers, but also the monument to the shame of the Chinese nation left by history. Through the vivid painting of the ruins of the stone pillars in the oil paintings, the audience will have a strong sense of empathy and inspire patriotic fervor in their hearts.

4

Conclusion

This paper uses the Fisher score to calculate the degree of correlation between each color feature of an oil painting and its class, as well as to evaluate each feature. Using the grayscale covariance matrix, texture features of the oil painting are extracted, and mathematical expressions are used to represent their energy, entropy, contrast, and other indexes. Add an extreme learning machine to optimize the feature extraction model of the oil painting, and use a plain Bayesian classifier to verify the effectiveness of feature extraction.

This paper specifically analyzes the color, texture, and stylistic features of the oil paintings extracted through simulation experiments. The difference between the red histograms of the two oil paintings is large, and the pixel value of “Sleeping Venus” is the highest in the luminance range of 10-15, which is 1880; on the contrary, the red histogram of Unknown Woman has the lowest pixel value in this luminance range, which is only 184. The values of the green and blue histograms of the two oil paintings are close to each other.

The calculation results of the texture characteristics of The Unknown Woman show that the roughness and contrast of part 1 of this oil painting are the highest, which are 17.489 and 31.565, respectively, and the directionality and regularity of part 2 are the highest, which are 40.196 and 0.985.

Using the model presented in this paper to identify the styles of 350 oil paintings, 305 were correctly classified, and the prediction accuracy was 87.143%.

Based on the characteristics of oil painting, we study its embodiment in modern social values and analyze the dynamic development law of oil painting art in the modern age.

Langue:: Anglais

Périodicité:: 1 fois par an
Sujets de la revue:: Sciences de la vie, Sciences de la vie, autres, Mathématiques, Mathématiques appliquées, Mathématiques générales, Physique, Physique, autres

RSS Feed de la revue

Exploring the dynamic relationship between oil painting art and modern social values by combining data mining techniques

Maoquan Zeng

Publié en ligne: 03 févr. 2025

Reçu: 09 sept. 2024

Accepté: 15 déc. 2024

DOI: https://doi.org/10.2478/amns-2025-0024

Mots clésFisher Score, Grey scale covariance matrix, Regularized limit learning machine, Plain Bayes, Oil painting art

© 2025 Maoquan Zeng, published by Sciendo

This work is licensed under the Creative Commons Attribution 4.0 International License.

Mots clés
Fisher Score, Grey scale covariance matrix, Regularized limit learning machine, Plain Bayes, Oil painting art