Search

CN-121980740-A - Impression-driven scientific and technological wood design method and system

CN121980740ACN 121980740 ACN121980740 ACN 121980740ACN-121980740-A

Abstract

The invention discloses a technological wood color design method and system based on impression driving, which are characterized in that first impressions, overall impressions and subsequent impressions of a user are acquired by setting observation time periods of 1s, 5s and 5min, the characteristics of color quantity, texture complexity, visual saliency and the like of the colors are extracted, a correlation model of the impressions and the color characteristics is established, and based on ' innovation, fun, nature, classics, uniqueness and high-level ' six-dimensional impression dimensionality of visual complexity and saliency division ', the relation between subjective impressions and objective colors is quantized, the automatic generation of impression-oriented color parameters is realized, the matching degree of design and user perception is improved, and the method is suitable for color design in the fields of technological wood, furniture decoration and the like.

Inventors

  • LI CHAO
  • LI KEXIN
  • QIN JIAPEI
  • HAN DONGQI

Assignees

  • 浙江理工大学

Dates

Publication Date
20260505
Application Date
20251208

Claims (10)

  1. 1. The design method of the technical wood design based on impression driving is characterized by comprising the following steps of: Step (1) constructing a standardized technological wood color image set, namely selecting technological wood color images which can cover wood grain types, various color numbers and various color combination modes; Setting three observation time periods corresponding to three layers of impressions, namely acquiring a first impression I 1 in a short time, acquiring an overall impression I 2 in a medium time period and acquiring a subsequent impression I 3 in a delayed return visit, designing subjective scales of a plurality of questions for each layer of impressions, acquiring scores, and taking an average value of the questions in the same layer; Extracting objective characteristics of each image, namely extracting the objective characteristics of each image and establishing a sample characteristic table, wherein the sample characteristic table comprises the number N of colors, texture complexity calculated based on the image information entropy H and color combination contrast; Verifying impression grading rationality by discriminant analysis or multi-class logistic regression, namely, determining statistical distinction of grading difference by combining single-factor variance analysis by taking observation time length as a class label and grading of questions as independent variables; establishing a correlation model of impression and color characteristics, namely establishing a linear or generalized linear model by taking an impression mean value as a dependent variable and the number of colors and the complexity of textures as independent variables; Calculating an information entropy H and a visual saliency mean M s of each image to form a sample data table as a basis of subsequent threshold estimation and dimension division, wherein a threshold is determined based on robust statistics, namely, a sample median H c of the information entropy H is used as a complexity threshold, and a 25% quantile M L and a 75% quantile M H of the saliency mean are used as saliency thresholds; Dividing six-dimensional impression dimensionality, namely dividing the impression into six dimensionalities based on two-dimensional coordinate axes of complexity and saliency, wherein H is less than or equal to H c ,M<M L , H is less than or equal to H c ,M L ≤M≤M H , H is less than or equal to H c ,M>M H , H is less than or equal to H c and M, H is more complex and more significant, H is more than H c ,M L ≤M≤M H , H is more than or equal to H c ,M<M L ; And (8) classifying the descriptive vocabulary into six-dimensional dimensions through designer experiments, and counting high-frequency words as dimension names.
  2. 2. The impression-driven scientific wood color design method according to claim 1, wherein the calculation model of the number of colors in the step (5) is as follows: Wherein N H is the number of colors predicted by the information entropy channel, N s is the number of colors predicted by the saliency channel, H is the information entropy, and M s is the visual saliency mean; the number of colors after final visual channel synthesis is: Wherein round is rounded, and N v is the number of colors after vision channel synthesis.
  3. 3. The impression-driven scientific and technological wood color design method according to claim 2, wherein the representative names of six dimensions in the step (7) are high-complexity high-significance correspondence "innovation", low-complexity high-significance correspondence "fun", low-complexity middle-significance correspondence "nature", low-complexity low-significance correspondence "classical", high-complexity low-significance correspondence "unique", and high-complexity middle-significance correspondence "advanced"; the impression dimension is used to predict the overall complexity and visual saliency range L [ H, M S ], L being the impression dimension. The range of different label values is as follows: The range of values is defined as follows: The complexity level p=1, H >5.77 belongs to high complexity, the complexity level p=0, H >5.77 belongs to low complexity, the visual saliency level q=0, MS <0.346 belongs to low saliency, the visual saliency level q= 1,0.275 is less than or equal to MS <0.346 belongs to medium saliency, the visual saliency level q=2, and MS >0.346 belongs to high saliency.
  4. 4. A method for designing a scientific wood color based on impression driving according to claim 3, wherein the correction model of the number of colors by the first impression is: N=round(clip((1+0.17×I 1 )×N v ,1,5)) Wherein I 1 is the first impression intensity, takes the value of [0,1], clip is limiting the result between 1 and 5, N is the number of colors, N is a positive integer of [1,5 ].
  5. 5. The impression-driven scientific wood color design method according to claim 1 or 4, wherein the texture complexity in the step (3) is calculated by the image information entropy H: wherein p i is the probability of occurrence of the ith gray in the image.
  6. 6. The impression-driven scientific and technological wood color design method according to claim 1, wherein the main scale in the step (2) is a five-point scoring system, and the questions do not display hierarchical labels; the five-point scoring is used for quantifying the subjective feeling, and specifically, the five-point scoring is very non-conforming to-1, not very conforming to-2, generally conforming to-3, comparatively conforming to-4 and very conforming to-5; The questions do not display hierarchy labels so as to avoid the subjects from knowing the evaluation purpose in advance to adjust the scores deliberately and ensure the objectivity of the data.
  7. 7. The impression-driven scientific wood color design method according to claim 1, wherein the statistical discrimination in the step (4) is determined by a probability value p <0.05 of a one-factor analysis of variance. The single factor analysis of variance is specifically that whether the topic score mean values of a short-time 1s observation group, an equal-time-length 5s observation group and a delayed return visit 5min observation group have significant differences or not is compared; When p <0.05 is calculated, the probability that the difference is randomly and accidentally caused is less than 5%, and the problem grading difference of different observation periods is considered to be truly existing, namely, grading of each impression level is statistically differentiated.
  8. 8. The scientific and technological wood design system based on impression driving is characterized by comprising an image acquisition module, an impression acquisition module, a feature extraction module, a model calculation module and an output module which are in communication connection; The image acquisition module is used for constructing a standardized technological wood color image set; the impression acquisition module is used for setting the observation time length and acquiring the scores of the three layers of impressions; the feature extraction module is used for calculating information entropy, visual saliency mean value and color features of the image; a model calculation module for performing the model calculations and dimension partitioning of claims 1-7; And the output module is used for generating the color parameters conforming to the target impression.
  9. 9. An impression-driven scientific and technological wood color design system according to claim 8, wherein the feature extraction module calculates visual saliency using CIELAB color space.
  10. 10. An impression-driven scientific wood color design system according to claim 8, wherein the output module supports automatic generation of color combinations from impression dimensions, high saliency corresponding to warm hues and low saliency corresponding to cool hues.

Description

Impression-driven scientific and technological wood design method and system Technical Field The invention relates to a scientific and technological wood color design method and system based on impression driving, in particular to a scientific and technological wood color feature extraction, analysis and design method and system based on subjective impression driving of a user, and belongs to the technical field of wood color design. Background In fields of furniture manufacturing, interior decoration and the like, scientific and technological wood gradually becomes a substitute material of natural wood due to the advantages of controllable texture, stable cost and the like, and the rationality of design and color of the scientific and technological wood directly influences the market acceptance of products. However, the existing technical wood design pattern has the remarkable limitation that 1) the design process is highly dependent on subjective experience of a designer, and quantitative association of real perception of a user is lacking. In the traditional design, a designer determines parameters such as the color number, the texture complexity and the like of the flower color according to the popular trend of the industry or the aesthetic sense of individuals, but the relation between the impression perception of the flower color by a user (such as 'visual impact of the first time' and 'comfort of long-term watching') and the parameters is not analyzed by a system, so that the problem of disjointing the designed flower color, namely 'the designer thinks the flower color is attractive but the acceptance of the user is low', particularly in the market with the increasingly-growing customization demands, the empirically-driven mode is difficult to match with diversified user preferences, and 2) the relation between the flower color characteristics and the visual perception lacks a scientific quantification model. The prior researches focus on the influence of colors and textures on visual experience, but do not distinguish the level differences of 'short-time quick perception (first impression)', 'medium-time observation (whole impression)', and 'delayed memory return visit (subsequent impression)', and do not establish dynamic mapping relation between characteristic parameters (such as color quantity) and different impression levels. Meanwhile, threshold definition of key perception indexes such as visual complexity, saliency and the like is fuzzy, an accurate parameter interval cannot be provided for design, so that different user impressions can be transmitted from the same parameter colors, a great amount of trial and error cost exists in the design process, and 3) the automatic generation of impression guidance is difficult to realize by the existing design tools. Most design software only provides basic color and texture editing functions, corresponding color parameters cannot be automatically calculated according to target impressions, a designer needs to manually adjust a plurality of variables, efficiency is low, and rationality of parameter combination is difficult to ensure. In summary, the existing technical wood design depends on the experience of designers, the quantitative association of subjective impressions of users is lacking, the matching degree between the design result and the perception of users is low, meanwhile, the relation between the color characteristics (such as the number of colors and textures) and the visual perception is lacking in a system model, and the accurate impression guiding design is difficult to realize. Therefore, research on a technical wood color design system and a method thereof with a user impression as a core, quantitative characteristics and a perception relation, and support for parameterization generation are needed to assist in realizing accurate conversion construction from subjective impression to objective color parameters, and improve design scientificity and market matching degree. Disclosure of Invention In order to solve the above prior art problems, the present invention provides a method and a system for designing a scientific and technological wood design based on impression driving. And establishing a correlation model of the impression and the color characteristics by quantifying multiple layers of impressions of the user on the wood color, and dividing six-dimensional impression dimensions based on visual complexity and significance to realize conversion from subjective impressions to objective color parameters. In order to achieve the above purpose, the present invention is realized by the following technical scheme: The invention discloses a scientific and technological wood design method based on impression driving, which comprises the following steps of: Step (1) constructing a standardized technological wood color image set, namely selecting technological wood color images which can cover wood grain categories, 1-5 color numbers and