Research on the image design of clothing patterns

From the Internet, periodical literature, clothing sales and fashion trend websites, we collected and screened the perceptual adjectives describing clothing patterns, and got 108 adjectives. Then, young consumers aged between 20 and 30 were given questionnaires and asked to choose from a list of 108 adjectives that were suitable for describing clothing patterns. Sixty questionnaires were sent out, and 56 were valid, with a recovery rate of 93.3%. Finally, 36 adjectives were obtained by counting the words with a 1/3 tick rate in the valid questionnaire.

The selected adjectives had to be extracted and classified to form the desired image adjectives. The experiment was conducted by using the open card classification method. The subjects were asked to put the cards of similar words, 36 words all-together, according to their personal feelings, with the number of groups controlled to 5~8. SPSS 21.0 software was used for systematic cluster analysis to determine the cluster number as 5, then K-means cluster analysis was conducted to specify the target cluster number as 5, and representative adjectives such as ordinary, simple, delicate, innovative and interesting were selected from each group. Then the antonyms of these five adjectives were matched. Finally, 5 groups of image adjectives were obtained: ordinary-personality, simple-complex, rough-delicate, innovative- traditional, interesting-boring. For the convenience of follow-up research, these 5 groups of image adjectives are numbered as A1~ A5.

Paper-cut is loved by people for its transparent visual sense and simple artistic sense. According to the effect of clothing, designers re-create paper-cut patterns of different styles in clothing design, which greatly enriches the visual beauty of clothing. In this paper, 160 representative paper-cut patterns in the Fujian region of China were collected through field investigation and a literature review. Twenty students and experts with a design background were divided into four groups, with one expert and four students in each group. The experts and students conducted cluster analysis together according to the similarity of the samples. The number of classified groups and the specific number of each group were not required, and finally 160 samples were divided into 18 categories. Finally, 5 experts were invited to recommend representative samples in the 18 categories, and one sample with the highest votes in each category was selected, with a total of 18 representative samples. In order to exclude the influence of paper-cut pattern color on subjective preference, grayscale processing was performed on the samples first, and in order to avoid the influence of clothing style, fabric, pattern color, application location and other factors on the experimental results, this research used PS software to uniformly place the paper-cut pattern on the front center of the T-shirt favored by the public, and finally obtained 18 clothing pattern samples, as shown in Figure 1.

Fig. 1.

In this paper, through the research of related literature of pattern design and the discussion of the inheritors of the paper-cut pattern, paper-cut pattern design elements were screened and selected by the method of hierarchical analysis, and the secondary design elements that have a weak impact on the pattern image and are easy to interfere with the user’s judgment were removed. The main design elements were determined that best characterize the pattern, including mode of composition, outer contour shape, technique of expression, theme (pattern content), line type and paper-cut techniques. Then these 6 design elements were deconstructed according to the morphological disassembly method in order to obtain the corresponding sub-elements of each design element, and encoded, as shown in Table 1.

Quantitative theory is a branch of multivariate statistics, which is divided into four kinds according to the different purposes of the research problem, respectively called quantitative theory I, II, III & IV. Among them, the quantitation theory type I is regarded as a classification multiple regression analysis method, which allows the inclusion of categorical and qualitative independent variables that can specify which design parameters are associated with triggering a certain psychological feeling. In quantitative theory, qualitative variables are often called ‘item’, and the various “values” of qualitative variables are called ‘category’. Suppose there are m items, and the category of the i-th item is c_i, then for n samples, δ_k (i, j) (i = 1,2,…, m; j = 1,2,…, c_i; k = 1,2,…, n) is called the reaction of item i category j in sample k, which can be obtained: (1)δk(i,j)=1,thekthsample,thequalitativedataofitemiiscategoryj0,(others)$$\matrix{ {{\delta _k}(i,j) = } \cr {\left\{ {\matrix{ {1,\left( {\matrix{ {the{\rm{ }}kth{\rm{ }}sample,{\rm{ }}the{\rm{ }}qualitative{\rm{ }}data} \cr {of{\rm{ }}item{\rm{ }}i{\rm{ }}is{\rm{ }}category{\rm{\;}}j} \cr } } \right)} \cr {0,(others{\rm{\;}})} \cr } } \right.} \cr } $$

It is assumed that there is a linear relationship between the dependent variable and the response of each item and category object, then a mathematical model can be established, (2)yk=∑i=1m∑i=1ciδk(i,j)bij+εk$${y_k} = \sum\nolimits_{i = 1}^m {\quad \sum\nolimits_{i = 1}^{ci} {{\delta _k}(i,j){b_{ij}} + {\varepsilon _k}} } $$

In the formula: b_ij - only depends on the coefficient of category j of item i; ε_k - random error in the k-th sampling.

100 young consumers aged 18-30 years old were investigated with the semantic difference method and 5-level Likert scale, and 97 valid questionnaires were collected. In order to test the validity of the survey data, 97 valid questionnaire data were tested by the Kendall cooperation coefficient of multi-paired samples, the results of which are shown in Table 2.

Table 2 shows that Kendall’s W cooperation coefficient is 0.8, greater than 0.5 and close to 1. The P value of the test probability is less than the significance level of 0.05, which means that the 97 young consumers have significant consistency in the perceptual evaluation of 18 clothing pattern samples, and the survey data are all valid. Figure 2 shows the evaluation results of the perceptual image of the clothing pattern.

Fig. 2.

SPSS26.0 statistical analysis software was used to calculate and analyze the sub-element score, partial correlation coefficient, rank, constant term, multiple correlation coefficient and coefficient of determination by using the QTT I, details of which are shown in Table 4.

The QTT I model is based on mathematical, physical and chemical multivariate linear regression equations. With 28 sub-elements as independent variables and the evaluation value of image adjectives as dependent variables, the qualitative and multiple linear regression equations of design elements and the perceptual evaluation value were constructed. According to the constant terms and sub-elements scores of each group image adjectives in Table 3, the multiple linear regression equation of each group’s image adjectives can be obtained. For example, the multiple linear regression equation of the ordinary-personality image adjectives is as follows: y1=4.771−0.413X11−0.357X12+0.126X13−0.462X14−0.259X15+0.474X16+0.426X17+0.337X18−0.324X21−0.238X22−0.135X23−0.269X24−0.272X25+0.396X260.693X31+0.422X32−0.431X33−0.363X41+0.187X42−0.247X43−0.464X44+0.489X45+0.652X460.089X51+0.137X52+0.108X61−0.261X62+0.178X63$$\matrix{ {{{\rm{y}}_1} = 4.771 - 0.413{{\rm{X}}_{11}} - 0.357{{\rm{X}}_{12}} + 0.126{{\rm{X}}_{13}} - } \cr {0.462{{\rm{X}}_{14}} - 0.259{{\rm{X}}_{15}} + 0.474{{\rm{X}}_{16}} + 0.42} \cr {6{{\rm{X}}_{17}} + 0.337{{\rm{X}}_{18}} - 0.324{{\rm{X}}_{21}} - 0.238{{\rm{X}}_{22}} - } \cr {0.135{{\rm{X}}_{23}} - 0.269{{\rm{X}}_{24}} - 0.272{{\rm{X}}_{25}} + 0.396{{\rm{X}}_{26}} - } \cr {0.693{{\rm{X}}_{31}} + {{0.422}_{{\rm{X}}32}} - 0.431{{\rm{X}}_{33}} - } \cr {0.363{{\rm{X}}_{41}} + 0.187{{\rm{X}}_{42}} - 0.247{{\rm{X}}_{43}} - } \cr {0.464{{\rm{X}}_{44}} + 0.489{{\rm{X}}_{45}} + 0.652{{\rm{X}}_{46}} - } \cr {0.089{{\rm{X}}_{51}} + 0.137{{\rm{X}}_{52}} + 0.108{{\rm{X}}_{61}} - } \cr {0.261{{\rm{X}}_{62}} + 0.178{{\rm{X}}_{63}}} \cr } $$

The multiple linear regression equations of the remaining 4 groups of image adjectives can be analogized according to Table 4. It can be seen from Table 4 that the coefficient of determination R² of the five groups of image adjectives are all greater than 80%, which indicates that the prediction equations of each group of image adjectives have a good degree of fit. In order to verify the reliability of the QTT I linear model for 5 groups of adjectives, a T test was performed on the perceptual evaluation value and QTT I predicted value of 18 test samples by SPSS software, the results of which show that the P values are all greater than 0.05, indicating that there is no significant difference; hence the QTT I model in this research is reliable.

Through the T test of the prediction models of 5 groups of adjectives, it can be seen that the QTT I model in this research is reliable, hence the correlation between the perceptual image of clothing patterns and design elements can be analyzed. The item partial correlation coefficient represents the correlation between various design elements and image adjectives. The higher the value, the stronger the degree of correlation with the image adjectives, that is, the greater the influence on the consumer image. For example, in the image adjective of ordinary-personality, the most relevant design element is X4, followed by X3, X1, X2, X5, X6. Among them, the sub-design elements of design elements X4 have negative scores, which belong to ordinary images. The larger the absolute value of negative values, the more ordinary images are indicated. Therefore, the order of influence degree of sub-design elements of design elements X4 on ordinary images is X44, X41, X43. The positive score of sub-design elements belongs to the image of personality, and the order of influence on the image of personality is X46, X45, X42. The correlation analysis of other image adjectives and design elements can be analogized. In order to facilitate designers’ reference and highlight the influence degree of sub-elements, the top two sub-design elements in the absolute value score are selected to analyze their correlation with perceptual vocabulary, and Table 4 is finally obtained.

According to the above analysis as a reference for clothing pattern design, take the ordinary - personality image adjective as an example, select design sub-elements X44, X31, X14, X21, X51, X62 in relation to the ordinary image from Table 5, and design sub-elements X46, X32, X16, X26, X52, X63 in relation to the image of personality to form a new scheme. The effect of the design scheme is shown in Figure 3.

Fig. 3.

In order to verify the validity of the design scheme, the semantic difference method was used to invite 20 target users to evaluate the perceptual image of the new design scheme, respectively. The evaluation mean value is shown in Table 5.

As can be seen from the results of Table 6, the image scores of scheme 1 and Scheme 2 for ordinary images are 1.374 and 1.147, respectively, which indicates that the design pattern has ordinary characteristics. For the image of personality, the image scores of design scheme 1 and 2 are above 4 points, indicating that the design schemes are consistent with the image of consumers. It can be seen that the design scheme generated by the method in this research is in line with the target user’s image expectation.