Deep Learning Modeling and Visual Aesthetics Integration Path in Cultural and Creative Designs

Further re-examination and introspection, and then the use of modern design methods, the cultural factors in a modern appearance, so that people in the spiritual level to explore the use of the process of satisfaction, which is the difference between cultural and creative products and general merchandise [1-2]. In the era of homogenization, designers need to create different psychological feelings for consumers, highlight the brand image of the product, enhance brand identity and loyalty, and form brand equity [3-4].

Cultural and creative products are quite diversified and basically utility-oriented, and all kinds of cultural and creative products are almost all frequently used items in daily life [5-6]. In terms of the shape of the products, they include household goods, stationery, clothing and accessories, 3C goods, tableware, decorations, toys and others [7-8]. However, the contemporary society is an era of material proliferation, a mug with images of famous paintings or a handbag with beautiful patterns is no longer very attractive to modern people who already have too many mugs and handbags, so how to link the information and cultural products in a skillful and creative way in order to attract the users’ favor is the real challenge of the cultural and creative product design [9-10].

Culture can move people’s hearts, and cultural creative products should be taken from tangible and intangible cultural assets, and users’ personal preferences and so on should be integrated into them, so that the cultural creative products will not be empty shells with only appearances, but will have storytelling, legend, and allusion, thus shaking people’s emotions [11-12]. With the development of computer hardware, especially the rise of GPU parallel computing, coupled with the explosion of big data, the explosion of deep learning was brought about around 2012 [13-14]. Deep learning is an important implementation of Artificial Intelligence (AI) today. Initially, deep learning was used in the field of speech recognition and the recognition rate was significantly improved. Then, convolutional neural networks achieved significant success in image recognition [15-16]. When AlphaGo won the big Go battle against Lee Sedol, it again proved the effectiveness of deep learning, which in turn was gradually and successfully applied to many fields. Then, Generative Adversarial Networks gained great success in areas such as image generation and style migration [17-18]. Therefore, it is necessary to study the personalized AI painting generation algorithm, integrate the user’s personal preferences and so on into the content of cultural and creative products, realize the personalized cultural and creative product design, and provide ideas and feasible solutions for the technology integration and innovation [19-20].

At the same time, based on the perspective of cultural and creative product design, the application of aesthetic visual elements not only needs to be compatible with the requirements of product design to produce a certain visual impact, but also should be integrated with interdisciplinary methods and theories, reflecting the commonality of art [21-22]. On this basis, the aesthetic visual elements of cultural and creative products should meet the requirements of modern people for the construction of spiritual civilization, and draw on the connotation of art cross-media theory. The only way to bring out the artistic value of cultural and creative products is to continuously improve the designers’ design ability and enhance their sense of innovation [23-24].

This paper firstly explains the basic concept of cultural creative design. Secondly, this paper introduces the multiple roles of color in culturally creative product design.On this basis, this paper constructs a type of image primary color adaptive extraction model based on a contour coefficient method.The model calculates the contour coefficient values of different primary colors during pixel clustering and completes the adaptive primary color extraction according to the largest contour coefficient value.Further, an intelligent color matching algorithm that incorporates visual aesthetics is proposed. The algorithm first utilizes eye tracking technology to obtain the visual aesthetics data stream. Then, the data stream is used to drive the Pix2Pix image translation model and quantitatively analyzed using the above color matching evaluation method. The effectiveness of the proposed method in this paper is finally verified by using examples of color design and cultural creative design products.

2

Method

2.1

Cultural Creative Design

Cultural creative design is a form of fusion of various cultural elements, which is transformed into products or services with practicality, aesthetics, and cultural connotations through creative means and design thinking.Its meaning mainly encompasses the following aspects: firstly, cultural creative design must include certain cultural elements, such as traditional culture, historical culture, regional culture, folk culture, and so on.These cultural elements are the basis for designers to design and the soul of their work.Secondly, the core of cultural creative design lies in creativity, as it allows us to break the routine, break stereotypes, and create design works with novelty, uniqueness, and personalization. Designers need to have keen observation and creativity, and be able to skillfully integrate cultural elements into the design to create creative works. Thirdly, cultural creative design needs to use advanced technical means and tools, such as computer-aided design and 3D printing, in order to realize the accuracy and efficiency of design. At the same time, designers also need to have certain handcrafting abilities and craft skills to be able to transform design drawings into actual products or services. Finally, the ultimate goal of cultural and creative design is to create commercial value, and it is necessary to consider commercial factors such as market demand for products, consumer psychology, and business models.Designers need to understand market demand and consumer psychology, and grasp the commercial nature of their designs to become competitive in the market.

Cultural creative design is mainly characterized by the following aspects: firstly, it is diverse. The elements of cultural creative design come from a wide range of sources and can integrate elements from many fields such as culture, art, science and technology, and commerce, and are characterized by diversity. Designers need to select and integrate elements from different fields to create works with uniqueness and diversity. Second is innovation. Cultural and creative design emphasizes innovation, and it should be able to break rules and stereotypes to create design works that are novel and unique. Designers need to constantly innovate and try to find new design ideas and methods to create more excellent works. The third is personalization. Cultural and creative design emphasizes personalization, enabling it to meet the needs and preferences of different consumers and create products or services with personalized characteristics. Designers need to deeply understand the psychology and needs of consumers and design according to different market demands. The last thing is sustainability.Cultural and creative design should focus on sustainability, which means considering the environmental protection, social responsibility, and economic benefits of the design work in order to realize sustainable development.Designers need to consider the life cycle of the product during the design process, minimize its impact on the environment, and improve its sustainability.

2.2

The role of color in cultural creative product design

2.2.1

The influence of color on product image

1)

Emotion and mood

Color can elicit emotional and moody responses from people, thus having a direct impact on product image. For example, red is often associated with passion, energy, and vitality, blue with calmness, steadiness, and tranquility, and yellow with happiness, warmth, and liveliness. Therefore, by choosing different colors, designers can convey specific emotions and moods in their products, thus influencing the product image.

2)

Cultural and symbolic meanings

Different colors may have different symbolic meanings in different cultures. Different colors can represent different cultures. For example, in Chinese culture, red usually symbolizes happiness, wealth, and success, but in the eyes of Westerners, red may symbolize danger. Therefore, in cultural and creative product design, the right choice of color can help the product resonate with the target culture and convey the right symbolism, thus shaping the product’s image.

3)

Brand Identity

Color plays a key role in brand image. Many brands use specific colors to establish their unique brand identity. For example, Coca-Cola uses a combination of red and white to form its iconic brand image. By using colors from a brand’s logo in product design, designers can strengthen the brand identity and align the product with the brand image, thereby positively impacting the product image.

4)

Attracting the target audience

Color can be used in product design to attract the target audience. Different age, gender, culture, and interest groups may have different preferences for different colors. Therefore, by studying the preferences and trends of the target audience, designers can choose colors that are suitable for the target audience, thus increasing the attractiveness and competitiveness of the product in the target market.

5)

Innovation and differentiation

The use of unique and innovative colors in cultural and creative product design can help products stand out in the market and differentiate themselves from competitors. By choosing unusual color combinations or using non-traditional colors, designers can create compelling product images that will attract attention in the marketplace, stimulate consumer interest and curiosity, and thus positively impact the product image.

6)

Sense of space and visual hierarchy

Color can also be used to create a sense of space and visual hierarchy in product design [25]. By using different shades, such as light and dark, and contrasting colors, designers can create a sense of hierarchy and depth in a product, thus giving it more depth and three-dimensionality. This helps to create a rich visual experience and enhance the three-dimensionality and texture of the product image.

7)

User experience and ease of use

Color can be utilized in product design to enhance user experience and ease of use. Through the rational use of color, the product interface or operation interface can be more intuitive, easy to identify, and use.For example, using different colors for different functions or operations can help users find the required functions more quickly, improving ease of use and user satisfaction.

2.2.2

The Impact of Color on User Experience

1)

Emotional experience

Color can evoke emotional resonance in users and trigger emotional experiences. Different colors have specific emotional associations in psychology, such as red and passion, blue and calmness, and yellow and vitality.Through the clever use of color, the product’s emotional experience and evaluation can be affected by the user.

2)

Cognitive impact

Color also has an impact on people’s perception. Differences in color produce different cognitive effects, such as brightness, contrast, saturation, etc., which all affect the user’s visual experience. For example, highly saturated colors are more likely to attract the user’s attention, while low-brightness colors are more likely to create a soft and warm atmosphere. Appropriate use of color allows users to make different understandings and evaluations of goods from a cognitive perspective.

3)

Behavioral orientation

Color can also have an impact on user behavior. Through the clever use of color, users can be guided to specific behaviors, such as buying, clicking, sharing, and so on. For example, some cultural and creative products attract users’ attention by using bright colors, thus prompting them to purchase or interact with the product.

4)

Cultural identity

Color in culturally creative product design can also affect the product’s cultural identity of the user. Different cultures have different symbols and symbolic meanings for colors, and the reasonable use of symbolic colors in specific cultures can cause users to identify and resonate with the products and specific cultures, thus affecting users’ experience and emotional connection to the products.

2.2.3

The relationship between color and cultural symbols

1)

Cultural heritage

As part of culture, colors and cultural symbols carry rich cultural connotations and symbolism. Through the reasonable use of colors and cultural symbols in a specific culture, the values, traditions and history of a specific culture can be inherited and displayed in product design [26].

2)

Personalized expression

Different cultures have different interpretations and evaluations of colors and cultural symbols, therefore, in cultural creative product design, the clever use of colors and cultural symbols with symbolic meanings in specific cultures can express the designer’s personality and unique perspective. By integrating colors and cultural symbols into product design, the product can be different in colors and symbols, thus highlighting its uniqueness and creativity.

3)

User recognition

Color and cultural symbols in product design can also contribute to user recognition and resonance. When users see familiar colors and cultural symbols in the product, they will feel friendly and familiar, thus enhancing their sense of identity with the product. This sense of identity can prompt users to like and accept the product more, and produce emotional connection and emotional experiences for the product.

4)

Cross-cultural communication

In cultural creative product design, the reasonable use of color and cultural symbols can cross different cultural groups and achieve cross-cultural communication. Different cultures have different interpretations and evaluations of colors and cultural symbols, so by using colors and cultural symbols with symbolic meanings in different cultures, we can arouse the resonance and interest of users in different cultures, and promote the dissemination and inheritance of products in different cultures.

2.3

Adaptive primary color extraction methods and evaluation indexes

2.3.1

K-Means primary color extraction method

The K-Means algorithm is a clustering algorithm based on Euclidean distance, the algorithm considers that the closer the Euclidean distance between two target data, the greater the similarity [27]. The average value of the data objects in each cluster is calculated as the new clustering center and the iteration is started until the clustering center reaches the maximum number of iterations. The spatial Euclidean distance calculation formula is shown in equation (1): (1) $d (x, C_{i}) = \sqrt{\sum_{j = 1}^{n} {(x_{j} - C_{i j})}^{2}}$

Eq:

x - data object.

C_i - The ird clustering center.

n - Dimension of the data.

x_j,C_ij - The jth attribute value representing x and C_i respectively.

The time complexity of the K-Means algorithm is shown in equation (2), which shows that the time complexity of the algorithm is close to linear, which also represents that the K-Means algorithm is more efficient in practical applications.

(2)

O (K N T)

Eq:

K - Total number of clusters.

N - amount of data.

T - Number of iterations.

2.3.2

Adaptive extraction algorithm for image primary color based on contour coefficient method

In the field of image processing and pattern recognition, the use of K-Means algorithm to extract the main color of the image can quickly extract the main color of the image in the complex color space, the core idea is that each pixel point within the class to the minimum of the sum of the squares of the distances from it to the center of the corresponding clusters, the K-Means algorithm is applied to the extraction of the main color of the image of the objective function is defined as follows: (3) $J = \sum_{n = 1}^{N} \sum_{k = 1}^{K} r_{n k} | | C (n) - μ_{k} | |^{2}$

Eq:

J - the total distance of all divided categories.

n, C(n) - Color values representing each pixel index of the image with the current pixel, respectively.

N, K - Total number of sample colors with the number of color classifications.

k - The kth category color.

r_nk - Dichotomy to determine if the current color is a class k color.

μ_k - Cluster center for the kth class color.

Since the color information of each image is different, its main color information and the number of main colors are also different. How to adaptively extract the main color of the image is the focus and difficulty of this section, for the K-Means clustering algorithm can only extract a fixed number of main colors of the problem, this section proposes to use the outline coefficient method to improve the K-Means clustering algorithm with a view to completing the task of adaptive main color extraction [28]. Specifically, for objective i there are i ∈ C_i, whose intra-cluster distance and inter-cluster distance are calculated as defined below: (4) $a (i) = \frac{1}{| C_{i} - 1 |} \sum_{j \in C_{i}, i \neq j} d (i, j)$ (5) $b (i) = \min_{k \neq i} \frac{1}{| C_{k} |} \sum_{j \in C_{k}} d (i, j)$

Eq:

a(i) - intra-cluster distance, the average Euclidean distance between the current pixel point and all points in the same cluster.

b(i) - Inter-cluster distance, the average Euclidean distance between the current pixel point and all points in the next cluster.

d(i,j) - Spatial Euclidean distance between target i and target j in cluster C_i.

C_k - Cluster k.

The formula for the contour coefficient value of target i is defined in equation (6): (6) $s (i) = {\begin{array}{l} 1 - \frac{a (i)}{b (i)}, a (i) < b (i) \\ 0, a (i) = b (i) \\ \frac{b (i)}{a (i)} - 1, a (i) > b (i) \end{array}$

Eq:

s(i) - contour coefficient value, which takes values between -1 and 1, representing the best and the worst clustering effect, respectively.

The overall contour coefficient value s of the clusters for this number of clusters is calculated as defined in equation (7): (7) $s = \frac{\sum_{i = 1}^{N} s (i)}{N}$

Eq:

N - the total number of target samples.

s(i) - the value of the contour coefficient for each sample in the sample set.

2.3.3

Evaluation metrics of primary color extraction algorithm

This section proposes a comprehensive evaluation of the primary color extraction effect using the similarity between the features of the primary color image and the paired images. The objective quantitative evaluation metrics are defined as follows: (8) ${\begin{array}{l} E_{1} = P S N R (I_{0}, I_{1}) \\ E_{2} = S S I M (I_{0}, I_{1}) \end{array}$

In Eq:

E₁, E₂ - represent the two types of quantitative evaluation indicators, respectively.

I₀, I₁ - represents the source image and the main color image, respectively.

Adaptive extraction of the main color of the image based on the contour coefficient method is shown in Fig. 1.

2.4

Design of color matching algorithms incorporating visual aesthetics

2.4.1

Visual aesthetics

The three eye movement behavior indicators are average gaze time, average number of gaze points, and first gaze time, which reflect the visual comfort, visual attractiveness, and visual impact of the test image or video, respectively. Taking the average gaze time as an example, its calculation formula in the eye tracking experiment is shown in equation (9).

(9)

T = \frac{\sum_{i = 1}^{n} T (A O I)}{n}

Eq:

n - number of test images or videos.

T(AOI) - test time for dividing the region of interest in the eye movement experiment.

Use eye movement behavior indicators to construct visual aesthetics data flow, normalized pre-processing of the above three eye movement behavior indicators, in which the average gaze time and the average number of gaze points are positively correlated with the degree of visual aesthetics fondness, taking the average gaze time as an example, the processing method is defined as follows: (10) $h^{'} = \frac{h - h_{-} \min}{h_{-} \max - h_{-} \min}$

Eq:

h - the current data to be processed before normalization.

h_min, h_max - the minimum and maximum values of the average gaze time, respectively.

The first gaze time measure is negatively correlated with visual aesthetics favoritism in the interaction task and is processed as defined below: (11) $j^{'} = 1 - \frac{j - j_{-} \min}{j_{-} \max - j_{-} \min}$

Eq:

j - the current data to be processed before normalization.

j_min,j_max - the minimum and maximum values of the first gaze time, respectively.

The visual aesthetic parameter W is shown in equation (12): (12) $W = (α h^{'} + β i^{'} + γ j^{'}) \times 10$

Eq:

W - the visual aesthetic parameter for fusing three different eye movement behavior data, taking values between [0, 10].

α, β, γ - the weights of the three different eye movement behavior data.

Where the three weights are set differently according to different visual tasks, α = β = γ = 1/3 is used in this chapter to calculate the visual aesthetics score W.

2.4.2

Image Translation Model

The network model diagram of Pix2Pix is shown in Fig. 2. The image translation model is a conditional generative adversarial network with U-Net and Markov discriminator (PatchGAN) as generator and discriminator respectively, whose generator input is a real sample image x and output is a generated image G(x). Since the discriminator needs to judge the authenticity of the output image of the generator, the input of the discriminator is a pair of images consisting of the generated image G(x) and the real image.

In addition, the Pix2Pix network model introduces L₁Loss to judge the image globally on the basis of CGAN, as shown in equation (13).

(13)

G^{*} = \arg \min_{G} \max_{D} L_{c G 4 N} (G, D) + λ L_{L_{1}} (G)

Eq:

G, D - represents the generator and discriminator respectively.

L_l1(G), λ - represents the weights of L₁Loss & L₁Loss respectively.

Where L₁Loss is defined as defined below: (14) $L_{L_{1}} (G) = E_{x, y, z} [| | y - G (x, z) | |_{1}]$

Eq:

G - generator.

x,y,z - denotes the true sample, conditional probability, and random noise, respectively.

The loss function of Pix2Pix is shown in equation (15): (15) $\begin{array}{l} L (G_{\min}, D_{\max}) = E_{x \sim p_{d a x} (x)} [\log D (x | y)] \\ + E_{z \sim p_{z} (z)} [\log (1 - D (G (z | y)))] + λ L_{L_{1}} (G) \end{array}$

2.4.3

Intelligent color matching model incorporating visual aesthetics

The palette visual aesthetic scoring model is shown in Figure 3. The evaluation model in this section uses the SE-InceptionV3 network model, which compresses the image features by global average pooling and scores the probability distribution of the image aesthetic quality scores, and trains the scoring model using the real visual aesthetic parameters of the palette as the labels of the corresponding palette.

The loss function of the backbone network Pix2Pix is updated using the aesthetic scores of the color palette, and the aesthetic loss function S(G) and the total loss function of the intelligent color matching algorithm incorporating visual aesthetics established in this chapter are shown in Eq. (16) and Eq. (17), respectively: (16) $S (G) = 10 - S c o r e$ (17) $\begin{array}{l} L (G_{\min}, D_{\max}) = E_{x \sim p_{d a t a} (x)} [\log D (x | y)] \\ + E_{z \sim p_{s} (z)} [\log (1 - D (G (z | y)))] \\ + λ L_{L_{i}} (G) + γ S (G) \end{array}$

In Eq:

Score, 10 - denotes the score of the visual aesthetic scoring model of the color palette and the full score of the aesthetic evaluation, respectively.

γ - the weight of S(G).

The schematic structure of the intelligent color matching algorithm network model incorporating visual aesthetics is shown in Fig. 4.

The proposed intelligent color matching algorithm incorporating visual aesthetics in this section, the framework of the intelligent color matching algorithm incorporating visual aesthetics is shown in Fig. 5.

3

Results and discussion

3.1

Calculation of Color Dynamic Harmonization Model Effect Values

The experiments in this section use the color matching algorithm proposed in this paper for the design of six color schemes and validate the effectiveness of the color matching algorithm proposed in this paper in several dimensions, such as color aesthetics, visual aesthetics and color intention.

3.1.1

Calculation of color beauty

The color aesthetics calculation results of each color scheme are shown in Table 1. As can be seen from the results of the aesthetics calculations of each scheme, all six schemes comply with the aesthetics principle, in which the color aesthetics values of Scheme 1 and Scheme 5 reach 1.24 and 1.03, respectively, and can be used as the preferred color schemes for further refinement.

Table 1.

Color beauty calculation results

Scheme	Color value(HSV)	Color design order sense	Total color number	Chromatic aberration	Brightness difference	Purity difference	Color beauty
1	(8,16,20),(187,98,50)	5.73	1.15	1	1	1	2
2	(198,100,91),(252,11,90),(15,5,28),(14,84,91)	17.28	0.96	6	6	5	1
3	(316,75,67),(0,83,90),(63,99,87),(240,53,60)	19.17	1.01	6	6	6	1
4	(146,89,55),(51,97,100),(19,18,54),(198,100,91)	16.43	0.86	6	6	6	1
5	(358,71,64),(149,87,50),(15,5,28)	14.17	1.77	3	1	3	1
6	(39,94,96),(88,72,75),(286,58,56)	9.29	0.93	3	3	3	1

3.1.2

Calculation of visual aesthetics

The results of the visual aesthetics calculation for each color scheme are shown in Table 2. Since the color schemes have the same color matching area and only differ in the number of colors, schemes 2 to 4 have the same visual aesthetics, and schemes 5 and 6 have the same visual aesthetics. Since the design scheme basically maintains horizontal symmetry, it has a good degree of symmetry, balance, and proportion in the horizontal direction, while the difference is larger in the vertical direction.

Table 2.

The vision of the color scheme is calculated

Scheme number	Symmetry	Equalization	Proportionality	Mean
1	0.8658	0.8537	0.9345	0.8876
2	0.7942	0.761	0.8257	0.7903
3	0.7942	0.761	0.8257	0.7903
4	0.7942	0.761	0.8257	0.7903
5	0.8967	0.8882	0.9493	0.9093
6	0.8967	0.8882	0.9493	0.9093

3.1.3

Color Imagery Calculations

In this section, “atmospheric” and “subtle” are used as representatives of imagery vocabulary for the evaluation of cultural imagery. A Likert seven-level scale was established to collect users and investigate their perception and evaluation of these six color schemes. And then the anti-fuzzy processing of user evaluation is carried out to investigate the user’s demand for cultural imagery, and the anti-fuzzy processing of user evaluation data is shown in Table 3. Accordingly, the results of the comprehensive cultural imagery ranking of the six color schemes are as follows: Scheme 1>Scheme 5>Scheme 2>Scheme 6>Scheme 3=Scheme 4.

Table 3.

User evaluation data anti-fuzzy processing

Imagery	1	2	3	4	5	6
Generous	0.802	0.651	0.615	0.558	0.824	0.599
Connotation	0.827	0.694	0.612	0.665	0.752	0.636
Comprehensive evaluation	0.815	0.673	0.614	0.612	0.788	0.618

3.1.4

Comprehensive Evaluation of Color Imagery Effect Values

The above results were regressed in MATLAB software. The dependent variable is the user’s evaluation value of the “atmospheric” image, the independent variables are the color beauty and visual beauty, and the regression analysis results of the “atmospheric” and “implicit” image evaluation are shown in Table 4 and Table 5. Based on the results of regression analysis, it can be seen that the correlation coefficient tends to be close to 1, and the linear fit is better. p < 0.05, which represents that the regression equation is valid. The results of the regression analysis are in good agreement with the original data, and the second set of data is odd data, which should be removed. The “atmospheric” regression equation is: E=0.089+0.63x1+0.25x2. Similarly, the “implicit” regression equation is: E=0.39+0.3x1+0.05x2.

Table 4.

“Generous” image evaluation regression analysis results

Scheme	Regression coefficient	The regression coefficient confidence interval	Residual error	Residual confidence interval	Correlation coefficient R²	F value	The probability of F corresponds to p	Error variance
1	0.089	[-0.45,0.59]	0.0032	[-0.13,0.15]	0.92	36.51	0.005	8.36x10^-4
2	0.089	[-0.45,0.59]	0.0395	[0.05,0.09]
3	0.63	[0.22,0.69]	-0.0239	[-0.15,0.08]
4	0.63	[0.22,0.69]	-0.0165	[-0.15,0.09]
5	0.25	[-0.65,0.89]	-0.0031	[-0.1,0.2]
6	0.25	[-0.65,0.89]	2.86 x10^-4	[-0.05,0.05]

Table 5.

“Implicit” image evaluation regression analysis results

Scheme	Regression coefficient	The regression coefficient confidence interval	Residual error	Residual confidence interval	Correlation coefficient R²	F value	The probability of F corresponds to p	Error variance
1	0.39	[-0.89,1.5]	0.059	[-0.16,0.25]	0.78	23.25	0.012	0.006
2	0.39	[-0.89,1.5]	0.032	[-0.22,0.31]
3	0.3	[-0.25,0.85]	-0.06	[-0.24,0.06]
4	0.3	[-0.25,0.85]	0.01	[-0.24,0.24]
5	0.05	[-1.68,1.85]	-0.06	[-0.26,0.18]
6	0.05	[-1.68,1.85]	-0.02	[-0.05,0.05]

3.2

Evaluation of visual aesthetic effect of cultural creative design products

This section of the experiment continues to evaluate the overall visual aesthetic effect of culturally creative design products using the method of this paper.The test objects are 12 culturally creative products designed using the method of this paper.

3.2.1

Score of visual aesthetics evaluation elements of cultural creative design products

The scores of visual aesthetics evaluation elements are shown in Table 6. According to the evaluation results, the average visual quality (AVQ) scores were low for Figure 11 (5.75) and high for Figure 1 (7.15).The results of the scores for the eight semantic variables: naturalness was highest for Figure 12 (8.6) and lowest for Figure 5 (2.92). The highest diversity was Figure 8 (6.18) and the lowest was Figure 9 (3.39). The Kruskal-Wallis test was conducted by grouping the 12 photos according to the semantic variables (eight groups in total), and the test results showed that there was no statistically significant difference between the eight semantic variables (P=0.447>0.05), which indicates that the naturalness, diversity, coordination, curiosity, orderliness, vividness, culture, and innovation shown by the cultural and creative design products are equalized of.

Table 6.

Visual evaluation factor score

	AVQ	Naturality	Diversity	Coordination	Peculiarity	Ordinal	Animality	Culture	Innovation
Figure 1	7.15	6.14	4.61	6.85	3.71	5.29	6.48	5.38	5.8
Figure 2	6.34	6.52	5.49	6.73	4.71	5	6.72	7.41	7.77
Figure 3	6.05	3.81	5.41	6.14	5.23	4.71	6.41	3.6	7.26
Figure 4	5.83	5.92	6.1	7.51	4.29	5.76	6.6	8.24	6.29
Figure 5	6	2.92	5.33	7.75	4.74	5.91	6.35	7.63	7.49
Figure 6	7	7.77	3.42	6.24	5.79	7.13	5.68	8.32	7.47
Figure 7	6.95	6.74	3.9	6.9	4.34	7.32	5.73	4	7.63
Figure 8	6.02	4.51	6.18	6.58	3.89	6.2	7	8.46	6
Figure 9	6.59	4.41	3.39	7.67	4.56	7.35	5.93	5.44	5.6
Figure 10	5.83	4.07	4.76	7.62	5.59	5.41	5.91	6.83	7.43
Figure 11	5.75	6.26	5.62	6.07	6.1	5.8	6.96	4.68	5.99
Figure 12	6.8	8.6	5.25	5.98	4.08	6.18	6.16	8.11	7.03

3.2.2

The relationship between the visual effect of cultural creative design products and semantic variables

The results of the correlation analysis between visual aesthetic effects and semantic variables are shown in Table 7. According to the results of the correlation, coordination (r=0.736, p=0.005), orderliness (r=0.845, p=0.002), and innovativeness (r=0.836, p=0.001) are significantly correlated with the average visual quality. The results of the correlation analysis between the semantic variables showed that peculiarity, cultural and naturalness (r=-0.836, p=0.001, r=-0.988, p=0.002) were significantly correlated, vividness and diversity (r=0.836, p=0.001) were significantly correlated, innovativeness and coherence (r=0.825, p=0.001) were very significantly correlated , orderliness, culture and peculiarity (r=0.63, p=0.035. r=0.765, p=0.005) are significantly correlated, and innovativeness and orderliness (r=0.689, p=0.018) are significantly correlated.

Table 7.

The correlation between the visual and semantic variables

	Naturality	Diversity	Coordination	Peculiarity	Ordinal	Animality	Culture	Innovation	AVQ
Naturality	1
Diversity	0.512	1
Coordination	0.446	0.287	1
Peculiarity	-0.836**	-0.215	-0.175	1
Ordinal	-0.466	-0.168	0.524	0.63	1
Animality	0.295	0.836**	0.325	-0.069	-0.096	1
Culture	-0.988**	-0.485	-0.378	0.765**	0.524	-0.269	1
Innovation	0.002	0.078	0.825**	0.112	0.689*	0.352	0.095	1
AVQ	-0.155	0.172	0.736**	0.458	0.845**	0.351	0.215	0.836**	1

4

Conclusion

The color matching method based on deep learning technology has become an important method in cultural creative design because it can improve the efficiency of color matching. This paper carries out an in-depth study on the deep learning model and visual integration path in cultural creative design.

In the experimental analysis part, this paper takes six color matching schemes as the main research objects, and finds that the regression equations of “atmospheric” and “implicit” in the color intention effect are E=0.089+0.63x1+0.25x2 and E=0.39+0.3x1+0.05x2, respectively. After providing an evaluation of the visual aesthetic effect of cultural creative design products, it is found that there is no statistically significant difference between the eight semantic variables (P=0.447>0.05), which shows that the effects of naturalness, diversity, coordination, oddity, orderliness, vividness, culture and innovativeness demonstrated by the cultural creative design products are equal, i.e., it shows that the method proposed in this paper has good utility on the cultural creative product design has good utility.

Funding:

This article represents a phased achievement of the Key Project of Jiangsu Province’s Educational Science Planning for 2023, titled “A Study on the Integration of Aesthetic Education Courses in Art Museums of Higher Vocational Colleges,” with project number B/2023/02/109.

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

Deep Learning Modeling and Visual Aesthetics Integration Path in Cultural and Creative Designs

Hailong Shen

Qingguo Sun

Publié en ligne: 21 mars 2025

Reçu: 02 nov. 2024

Accepté: 07 févr. 2025

DOI: https://doi.org/10.2478/amns-2025-0649

Mots clésAdaptive primary color extraction, Color matching, Deep learning model, Cultural creative design

© 2025 Hailong Shen et al., published by Sciendo

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

Mots clés
Adaptive primary color extraction, Color matching, Deep learning model, Cultural creative design