Commercial vehicle color design evaluation method and system, terminal and medium

Abstract

The invention relates to the field of industrial design, and particularly provides a commercial vehicle color design evaluation method and system, a terminal and a medium, and the method comprises the steps: carrying out the calculation through an analytic hierarchy process and fuzzy comprehensive evaluation, and obtaining an analytic hierarchy process score value containing the five dimensions of brand, environment, user, function and cost; the method comprises the following steps: analyzing fixation data collected by an eye tracker, and quantifying to obtain an objective preference score of a user for a preset interest area; clustering analysis is carried out on a user color preference image, and a color preference matching degree score of a design scheme is obtained through calculation. And inputting the multi-source quantitative score and the current main color feature of the scheme into a pre-trained color design optimization suggestion generation model, and directly outputting specific adjustment parameter suggestions for hue, brightness and purity by the model. According to the invention, standardization, objectification and engineering of commercial vehicle color design evaluation are realized.

Description

Technical Field

[0001] This invention relates to the field of industrial design, specifically to an evaluation method, system, terminal, and medium for color design of commercial vehicles. Background Technology

[0002] Currently, the evaluation and decision-making of color design schemes for commercial vehicles mainly rely on the experience-based review of design teams, market research questionnaires, or simple focus group discussions. Traditional review models are primarily qualitative, with evaluation results dependent on the individual aesthetic preferences and experience of reviewers. This leads to significant discrepancies in evaluation standards among different review subjects, making it difficult to form stable, reproducible, and objective evaluation conclusions. This fails to meet the technical requirements of consistency and reliability in commercial vehicle color design. Existing user research methods such as questionnaires and interviews can only obtain subjectively stated user preferences, failing to capture users' subconscious visual attention behaviors and true preference tendencies during visual perception without interference. This results in insufficient accuracy in user preference data collection, directly affecting the accuracy and effectiveness of evaluation results. Related technologies have failed to establish a mapping relationship between multi-dimensional evaluation indicators and specific design parameters. They cannot transform multi-source evaluation information such as brand recognition, user preferences, and process costs into quantitative adjustment parameters for hue, brightness, and saturation. Consequently, evaluation results cannot directly guide the iterative optimization of design schemes and lack engineering practicality. Summary of the Invention

[0003] To address the aforementioned issues, this invention provides a method, system, terminal, and medium for evaluating the color design of commercial vehicles, thereby achieving standardization, objectification, and engineering of the evaluation of commercial vehicle color design.

[0004] In a first aspect, the technical solution of the present invention provides a method for evaluating the color design of commercial vehicles, comprising the following steps: Based on an evaluation index system that includes brand recognition, environmental adaptability, target audience preferences, functional semantics, and process cost, the weights of each index are determined by the analytic hierarchy process (AHP), and the analytic hierarchy process score of the color design scheme is calculated using a fuzzy comprehensive evaluation model. Collect user eye movement data generated when target users observe color design samples, and obtain user preference scores by calculating the total access time or the percentage of fixation times in preset interest areas; Color clustering analysis and color system mapping are performed on the color preference images provided by users to obtain user color preference data. By calculating the similarity between the scheme to be evaluated and the user preference data in terms of main color features, the color preference matching score is obtained. The hierarchical analysis score, user preference score, color preference matching score, and the main color feature of the current color scheme are concatenated to form an input vector. This input vector is then fed into a pre-trained color design optimization suggestion generation model, which outputs suggestions for adjusting hue, brightness, or saturation.

[0005] Secondly, the technical solution of the present invention provides an evaluation system for the color design of commercial vehicles, including: The hierarchical analysis evaluation module is used to determine the weight of each indicator based on an evaluation index system that includes brand recognition, environmental adaptability, target audience preferences, functional semantics, and process cost, and to calculate the hierarchical analysis score of the color design scheme using the hierarchical analysis method and the fuzzy comprehensive evaluation model. The user preference evaluation module is used to collect user eye movement data generated when target users observe color design samples. By calculating the total access time or the percentage of fixations in preset interest areas, a user preference score is obtained. The color preference evaluation module is used to perform color clustering analysis and color system mapping on the color preference images provided by the user to obtain the user's color preference data. By calculating the similarity between the scheme to be evaluated and the user's preference data in terms of main color features, the color preference matching score is obtained. The parameter adjustment suggestion generation module is used to concatenate the hierarchical analysis score, user preference score, color preference matching score, and the main color feature of the current color scheme to form an input vector. The input vector is then fed into a pre-trained color design optimization suggestion generation model, which outputs adjustment parameter suggestions for hue, brightness, or saturation.

[0006] Thirdly, the technical solution of the present invention provides a terminal, comprising: Memory for storing evaluation programs for commercial vehicle color designs; The processor is used to implement the steps of the commercial vehicle color design evaluation method described above when executing the evaluation program for the commercial vehicle color design.

[0007] Fourthly, the present invention provides a computer-readable storage medium storing an evaluation program for commercial vehicle color design, wherein the evaluation program for commercial vehicle color design, when executed by a processor, implements the steps of the above-described evaluation method for commercial vehicle color design.

[0008] As can be seen from the above technical solutions, this application has the following advantages: By constructing a multi-dimensional evaluation index system that includes brand recognition, environmental adaptability, target audience preferences, functional semantics, and process costs, and combining the analytic hierarchy process (AHP) to quantify the weights of the indicators, it integrates objective visual attention features captured by user eye-tracking data and user preference matching degrees obtained from color clustering analysis to form a quantitative evaluation score based on multi-source data fusion. Furthermore, through a pre-trained optimization suggestion generation model, the evaluation results are combined with heat map distribution features and style vectors to output directly executable hue, brightness, and saturation adjustment parameter suggestions. This invention effectively solves the technical defects of traditional evaluation methods, such as strong subjectivity, insufficient capture of real preferences, and lack of design guidance, and realizes the standardization, objectification, and engineering of color design evaluation for commercial vehicles. Attached Figure Description

[0009] To more clearly illustrate the technical solution of this application, the accompanying drawings used in the description will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1 This is a schematic diagram of a method for evaluating the color design of commercial vehicles, provided as an embodiment of the present invention.

[0011] Figure 2 This is a schematic block diagram of an evaluation system for color design of commercial vehicles provided in an embodiment of the present invention.

[0012] Figure 3 This is a schematic diagram of the structure of a terminal provided in an embodiment of the present invention. Detailed Implementation

[0013] To make the purpose, features, and advantages of this application more apparent and understandable, specific embodiments and accompanying drawings will be used to clearly and completely describe the technical solution protected by this application. Obviously, the embodiments described below are only some embodiments of this application, and not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0014] Unless otherwise defined, all technical and scientific terms used in this application have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used in this application and in the specification of this invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.

[0015] Figure 1This is a schematic flowchart illustrating a method for evaluating the color design of commercial vehicles, provided as an embodiment of the present invention. Figure 1 The executing entity can be a commercial vehicle color design evaluation system. The commercial vehicle color design evaluation method provided in this embodiment is executed by a computer device, and correspondingly, the commercial vehicle color design evaluation system runs on the computer device. Depending on different needs, the order of the steps in this flowchart can be changed, and some steps can be omitted.

[0016] like Figure 1 As shown, the method includes the following steps.

[0017] S1, based on an evaluation index system that includes brand recognition, environmental adaptability, target audience preferences, functional semantics and process cost, determines the weight of each index through the analytic hierarchy process and uses a fuzzy comprehensive evaluation model to calculate the analytic hierarchy score of the color design scheme.

[0018] S2 collects user eye movement data generated when the target user observes the color design sample, and obtains the user preference score by calculating the total access time or the percentage of fixations in the preset interest area.

[0019] S3. Perform color clustering analysis and color system mapping on the color preference image provided by the user to obtain the user's color preference data. Calculate the similarity between the scheme to be evaluated and the user's preference data in terms of primary color features to obtain the color preference matching score.

[0020] S4 concatenates the hierarchical analysis score, user preference score, color preference matching score, and the main color feature of the current color scheme to form an input vector. The input vector is then fed into a pre-trained color design optimization suggestion generation model, which outputs suggestions for adjusting hue, brightness, or saturation.

[0021] As a refinement and extension of the specific implementation of the above embodiments, in order to fully explain the specific implementation process of this embodiment, the following will provide possible embodiments to describe the specific implementation of the above steps in a non-limiting manner.

[0022] In some alternative implementations, step S1, obtaining the hierarchical analysis score, specifically includes the following steps S1.1 to S1.5.

[0023] S1.1 Construct a hierarchical evaluation index system for commercial vehicle color design. The index system includes a target layer, a criterion layer, and an indicator layer. The criterion layer consists of primary indicators, including brand recognition, environmental adaptability, target audience preference, functional semantics, and process cost. The indicator layer consists of secondary indicators, including at least three specific evaluation items belonging to each primary indicator.

[0024] Target layer (A): The final output, namely "the comprehensive evaluation value of the color design scheme for commercial vehicles"; Criterion Layer (B): This refers to the primary indicators. This embodiment identifies five key dimensions that influence the overall value of color design in commercial vehicles: Brand recognition (B1): Measures the fit between the color scheme and the corporate visual identity system and the ability to convey brand personality; Environmental adaptability (B2): Measures the visual harmony and durability of a color scheme in different climates and usage scenarios; Target audience preferences (B3): Measures the degree to which a color scheme matches the aesthetic preferences and psychological feelings of the intended users; Functional semantics (B4): Measures the effectiveness of color in indicating vehicle functions, enhancing safety, and conforming to industry practices; Process cost (B5): Measures the technical feasibility, cost controllability, and ease of maintenance of the color scheme during mass production painting.

[0025] Indicator Layer (C): These are the secondary indicators, which are the concretization and operationalization of the criteria layer. Each primary indicator has at least three secondary indicators that can be directly observed or evaluated. For example, "Brand Recognition (B1)" can be specifically broken down into "Company Standard Color Compliance (C11)", "Brand Image Communication (C12)", and "Visual Recognition (C13)", etc.

[0026] S1.2, construct judgment matrix A based on the results of pairwise importance comparisons of indicators at the same level using the 1-9 scale method employed by experts.

[0027] Several experts from the fields of commercial vehicle design, engineering, and marketing were invited to conduct pairwise importance comparisons of indicators at the same level (such as the five primary indicators of the criteria level) based on the “1-9 scale method” and its meaning shown in Table 1.

[0028] Table 1: Specific meanings of the 1-9 scale method

[0029] Based on the comparative judgments of all experts, a judgment matrix A is constructed, where the elements are... This represents the importance scale of indicator i relative to indicator j.

[0030] S1.3, the eigenvalue method is used to calculate the largest eigenvalue of the judgment matrix A and its corresponding eigenvector, and the eigenvector is normalized to obtain the weight vector W of each index; the consistency ratio CR of the judgment matrix A is calculated, and the judgment weight is valid when CR < 0.10.

[0031] The eigenvalue method is used to calculate the largest eigenvalue of the judgment matrix A. And its corresponding eigenvector. The eigenvector is then normalized to obtain the vector. This is the weight vector for each indicator. Weights The magnitude of the value directly reflects the relative importance of the corresponding indicator in the evaluation system.

[0032] To eliminate potential logical contradictions in expert judgments, such as A being more important than B, B being more important than C, but C being more important than A, a consistency check must be performed. A consistency index is calculated. and consistency ratio Where RI is the average random consistency index, which can be obtained by looking up a table. The consistency of the judgment matrix is ​​considered acceptable if and only if CR < 0.10, and the calculated weight vector... Only then will it be valid. If it fails the test, feedback needs to be provided to experts to readjust the judgment until a logically consistent judgment matrix is ​​obtained.

[0033] S1.4, Establish the evaluation set The frequency of each secondary indicator belonging to each evaluation level is counted by experts, the membership degree is calculated, and a fuzzy evaluation matrix R is constructed. Using a weighted average fuzzy synthesis operator, the weight vector W and the fuzzy evaluation matrix R are synthesized to obtain the fuzzy evaluation result vector B.

[0034] Considering the inherent fuzziness in design evaluation, a fuzzy comprehensive evaluation method is introduced to transform the qualitative evaluation of experts into quantitative data.

[0035] First, define a general set of evaluation levels, for example... =Excellent, Good, Average, Poor, Poor. An expert panel evaluates each of the lowest-level secondary indicators (e.g., C11) and tallies the categories they are considered to belong to. The number of experts. This is determined by calculating membership degrees. This forms the membership vector of the indicator, where Assuming the index Cij belongs to the level The number of experts is D, where D is the total number of experts. The membership vectors of all the lower-level indicators are combined to form the fuzzy evaluation matrix R of the corresponding upper-level indicator.

[0036] Using the weight vector W obtained in step S1.3 and the fuzzy evaluation matrix R, fuzzy synthesis operations are performed. This invention preferably uses a weighted average operator for calculation, the formula of which is: ,in This operator can fully utilize all indicator information to obtain a fuzzy evaluation result vector B at the criterion level (first-level indicators) and even the final target level (overall scheme). Vector B represents the degree to which the scheme belongs to each level in the evaluation set V.

[0037] S1.5, defuzzify the fuzzy evaluation result vector B to obtain the hierarchical analysis score.

[0038] For each level in the evaluation set V Assign a specific score For example, F=(100,80,60,40,20).

[0039] The weighted average method was used to calculate the final analytic hierarchy process score. The calculation formula is:

[0040] This score This refers to the comprehensive quantitative evaluation result of the design scheme in five dimensions: brand, environment, user, function, and cost, which integrates the wisdom of a group of experts, is based on a scientific weighting system, and handles the ambiguity of evaluation.

[0041] This embodiment constructs a complete multi-level evaluation index system. It uses brand recognition, environmental adaptability, target audience preferences, functional semantics, and process cost as the core criteria layer, coupled with at least three specific secondary evaluation items under each criterion, forming an evaluation dimension network covering design, market, and engineering dimensions. This solves the technical defects of traditional evaluation indicators being singular and one-sided. It employs the Analytic Hierarchy Process (AHP) combined with the 1-9 scaling method to construct a judgment matrix. The index weights are calculated using the eigenvalue method and verified through consistency testing (CR < 0.10) to ensure weight validity. Expert experience is transformed into a quantified weight vector, replacing the traditional subjective weighting model and making the determination of index importance reproducible. A fuzzy comprehensive evaluation model is introduced. By constructing an evaluation set and statistically analyzing membership degrees to form a fuzzy evaluation matrix, a weighted average fuzzy synthesis operator is used to achieve quantitative synthesis and defuzzification of multiple indicators. This transforms qualitative evaluation into precise analytic hierarchy process scores, effectively reducing the interference of individual aesthetic preferences of reviewers on the evaluation results. This reduces the dispersion of conclusions from different evaluation subjects by more than 30%, significantly improving the stability and credibility of the evaluation results.

[0042] In some optional implementations, step S2, which obtains the user preference score, specifically includes the following steps S2.1 to S2.4.

[0043] S2.1, Use an eye tracker to collect raw eye movement data when at least one target user observes a color design scheme sample to be evaluated, the raw eye movement data including fixation point coordinates and fixation duration recorded in time series.

[0044] An eye tracker was used as the hardware foundation for data acquisition. At least one target user was required to naturally observe a sample of the commercial vehicle color design scheme to be evaluated in a controlled experimental environment. This sample was typically presented as a high-fidelity digital image or a screen-rendered model. During this process, the eye tracker continuously recorded the user's raw eye movement data. This data is a time-stamped sequence, including: Gaze coordinates: The position of the user's visual focus on the two-dimensional plane of the sample image, represented by pixel coordinates (x, y); Gaze duration: The length of time a user remains focused on a single point of gaze.

[0045] This step transforms the user's subjective, implicit visual attention process into objective time-space sequence data that can be processed by a computer.

[0046] S2.2, Based on the image of the color design scheme sample, define at least one region of interest (ROI) with pixel coordinate range according to its different color regions and functional component regions; map and match the gaze coordinates in the original eye-tracking data with the pixel coordinate range of the ROI to identify the effective gaze points falling within each ROI.

[0047] To associate continuous eye-tracking data with specific areas of the design scheme, regions of interest (ROIs) are predefined for analysis. The ROI definition method in this embodiment is not simply gridding, but rather incorporates commercial vehicle design semantics, including the following steps S2.21 to S2.23.

[0048] S2.21, Gaussian filtering is applied to the RGB image of the color design scheme sample for noise reduction.

[0049] First, Gaussian filtering is applied to the RGB images of the color design scheme samples to smooth out image noise and avoid small, meaningless color fragments caused by noise in subsequent color segmentation.

[0050] S2.22, convert the denoised image from the RGB color space to the LAB color space; use a clustering algorithm based on the color difference ΔE in the LAB color space to divide the image into several color blocks.

[0051] The preprocessed image is converted from the RGB color space to the CIELAB color space. The advantage of the LAB color space lies in its perceptual uniformity; its color difference ΔE can well reflect the color differences perceived by the human eye. Subsequently, a clustering algorithm based on the color difference ΔE in the LAB color space is used to cluster the image pixels, merging visually uniform color pixel areas, thereby dividing the entire design image into several "color blocks." The colors within each block are similar, while there are perceptible color differences between blocks.

[0052] S2.23, Based on the pre-stored component template coordinate mapping table of the standard three-view of commercial vehicles, the color blocks are merged to obtain at least one region of interest; wherein, the component template coordinate mapping table defines the coordinate range of the outline polygon of each physical component in the standard view.

[0053] To imbue these color blocks with engineering semantics, this embodiment introduces a pre-stored coordinate mapping table of standard three-view component templates for commercial vehicles. This mapping table defines the coordinate range of the outline polygons of each physical component (such as the cab, cargo box, bumper, wheels, and brand logo area) in standard side views, front views, and other views.

[0054] a) Determination of component ownership.

[0055] For any color block, if more than a preset proportion of its pixels fall within the outline polygon range of a component in the mapping table, then the color block is determined to belong to that component.

[0056] Specifically, for each segmented color block, the coordinates of all its pixels are calculated. If more than a preset proportion (e.g., more than 50%) of the pixels in the block fall within the outline polygon of a component A in the component template, then the block is determined to belong to component A.

[0057] b) Spatial connectivity check and merging.

[0058] For all color blocks belonging to the same component, calculate the minimum distance between the boundary pixels of any two blocks; if the distance is less than a preset adjacent threshold, then the two blocks are determined to be adjacent.

[0059] Specifically, for all color blocks identified as belonging to the same component A, spatial adjacency is determined. The minimum distance between the boundary pixels of any two such blocks is calculated; if this distance is less than a preset adjacency threshold (e.g., 5 pixels), they are considered adjacent. Finally, using a connected component search algorithm from graph theory, all blocks belonging to the same component A and spatially connected (directly or indirectly adjacent) are merged.

[0060] Specifically, for any color block obtained after segmentation, its single-pixel-wide contour boundary is first extracted using image processing techniques, denoted as the boundary point set. For two color blocks belonging to the same component, the minimum Euclidean distance between their boundaries is calculated, which can be achieved by calculating the distances between all pairs of points in the two boundary point sets and taking the minimum value. A preset adjacency threshold is defined; if the minimum Euclidean distance is less than or equal to the adjacency threshold, the two color blocks are considered spatially adjacent. All color blocks belonging to the same component are considered as vertices of an undirected graph G. For any two vertex color blocks, if they satisfy the adjacency determination condition, an edge is established between them. Subsequently, a depth-first search or breadth-first search algorithm is run on this graph G to find all connected components of the graph. Each found connected component represents a set of color blocks that are spatially adjacent to each other (directly or indirectly connected through other blocks of the same component). For each found connected component, i.e., the set of blocks, the pixel regions of all blocks in the set are logically joined to form a continuous and complete image region. This merged region is finally identified as an independent region of interest for the corresponding component. If a color block is not adjacent to any other block of the same component, it constitutes an independent connected component and is also defined as an independent region of interest.

[0061] c) Generation of regions of interest.

[0062] For all color blocks belonging to the same component, find all sets of blocks that are connected to each other through adjacency. Each set of connected blocks is merged and marked as an independent region of interest.

[0063] Specifically, each connected color region that is ultimately merged and corresponds to a specific physical component is designated as an independent region of interest and defined by the range of pixel coordinates of the region's minimum bounding rectangle or actual polygon in the image.

[0064] S2.3, Calculate the total fixation duration of all valid fixation points within each region of interest. And calculate its total fixation duration along with that of all fixation points in this observation task. The ratio; or, the number of gazes accessed by each of the said regions of interest by effective gaze points. And calculate its number of fixations compared to the total number of fixations in this observation task. The ratio of .

[0065] Iterate through all gaze point data and obtain the coordinates of each gaze point. Match the pixel coordinates of all regions of interest. If a gaze point falls within a region of interest... If the data point falls within a certain range, it is identified as a valid fixation point for that region of interest. Further validity filtering conditions can be set, such as removing data points with excessively short or long fixation durations.

[0066] For each defined area of ​​interest Perform the following calculation: count all those falling within... The total fixation time of the effective fixation point within the range is denoted as . ; Calculate the total fixation duration for all fixations or all valid fixations in this user observation task. ; Calculate the ratio: .

[0067] Similarly, it is possible to count areas of interest. Number of visits by effective focal points and total number of visits ratio The fixation duration ratio better reflects the user's sustained attention level, while the fixation frequency ratio reflects the frequency with which the area attracts the user.

[0068] S2.4 The calculated gaze duration ratio or gaze frequency ratio is used as an objective preference score representing the user's preference for the color region represented by the corresponding interest area.

[0069] The ratio calculated in step S2.3 or Directly used to represent user interest areas The objective preference score for the color region it represents. For an individual user, or This refers to the individual's preference score for that region. After collecting eye-tracking data from multiple target users, it's possible to assign all users the same area of ​​interest. The arithmetic mean of the preference scores is used to obtain the final user preference score, which represents the overall preference of the target user group.

[0070] This embodiment uses an eye tracker to collect raw data such as the coordinates of the user's gaze point and the duration of gaze when observing the design sample. Objective physiological data replaces subjective stated preferences, capturing the user's subconscious visual attention behavior without interference, avoiding misjudgments of preferences caused by user expression biases in questionnaires and interviews. Based on the image segmentation and component matching region of interest definition method, color blocks are segmented through RGB-LAB color space conversion, Gaussian filtering noise reduction, and color difference ΔE clustering. These blocks are then merged using a component template coordinate mapping table from standard three-view drawings of commercial vehicles, accurately locating the correspondence between color regions and functional components, improving the accuracy of effective gaze point recognition. The proportion of gaze duration and the proportion of gaze frequency in the region of interest are used as objective preference scores. Simultaneously, the K-Means clustering algorithm extracts the main color set of the user's preferred image. After mapping through a standardized color system (hue-brightness-saturation three dimensions), the similarity to the main color of the design scheme is calculated, forming a dual user preference evaluation dimension combining subjective preference statements and objective visual behavior, improving the comprehensiveness and accuracy of preference capture.

[0071] In some optional implementations, step S3, obtaining the color preference matching score, specifically includes the following steps S3.1 to S3.4.

[0072] S3.1 Obtain at least one color preference image provided by the user; use the K-Means clustering algorithm to perform cluster analysis on all pixels of the color preference image in a predetermined color space, and extract the K colors with the highest proportion as the user's preferred main color set, wherein each main color is represented by a vector in the predetermined color space.

[0073] To avoid subjective bias caused by direct questioning, this embodiment indirectly infers the user's color preference by analyzing the color preference image provided by the user.

[0074] Obtain at least one digital color preference image provided by the user. First, perform normalization preprocessing on the image, such as size normalization and light Gaussian filtering, to eliminate the interference of size differences and subtle noise on subsequent clustering analysis.

[0075] Then, each pixel of the image is converted from the common RGB color space to a color space with better perceived uniformity, such as the CIELAB or HSL color space. Taking the CIELAB space as an example, each pixel will be represented as a three-dimensional vector [L*, a*, b*], where L* represents lightness, and a* and b* represent chromaticity coordinates.

[0076] The K-Means clustering algorithm is used to perform unsupervised clustering analysis on the LAB vectors of all pixels in the image. This algorithm is based on iterative optimization, dividing all pixels into K clusters that minimize the color vector distance (usually Euclidean distance) within the same cluster and maximize the distance between different clusters. After clustering, the proportion of pixels in each cluster to the total number of pixels is calculated. The color vectors of the center points of the K clusters with the highest proportions are extracted to form the user's preferred primary color set, denoted as . Here, the K value can be set based on experience (e.g., K=5), for each primary color. It is a three-dimensional LAB vector that objectively represents the core colors in the user's preferred image.

[0077] S3.2, using the same K-Means clustering algorithm and a predetermined color space, performs cluster analysis on the sample images of the color design schemes to be evaluated, and extracts the M colors with the highest proportion as the main color set of the design scheme.

[0078] To ensure fairness and consistency in the comparison, the same technical process as S3.1 was used to process the sample images of the color design schemes to be evaluated: the same preprocessing and color space transformation were performed on the design scheme images; the same K-Means clustering algorithm with the same parameters (same distance metric, convergence threshold, etc.) was used to extract the top M colors with the highest proportions in the image as the main color set of the design scheme, denoted as . The value of M can be the same as K or adjusted according to the design complexity.

[0079] S3.3, map the user's preferred main color set and the main color vectors in the design scheme's main color set to a predefined standardized color system for quantitative positioning; the standardized color system is a system based on three dimensions: hue, brightness, and saturation.

[0080] To perform semantic color comparisons across images, the primary color vectors need to be mapped to a unified reference system with semantic coordinates. This embodiment uses a predefined standardized color system as this reference system. This system typically organizes the color space systematically using three dimensions: hue, value / lightness, and chroma.

[0081] Set the user's preferred primary colors and the main color scheme of the design scheme Each LAB vector in the system is accurately mapped to its corresponding position in the standardized color system using a color space conversion formula. For example, a LAB vector [L*, a*, b*] can be converted into a coordinate system within that system. .

[0082] S3.4 Calculate the similarity between the user's preferred primary color set and the design scheme's primary color set in the standardized color system space; use the calculated similarity value as the color preference matching score.

[0083] Specifically, the color preference matching score is calculated based on the feature center and the distribution distance.

[0084] Calculate separately and Statistical characteristics in the three-dimensional space of a standardized color system. An effective method is to treat each set as a three-dimensional point cloud, calculate its centroid (mean vector) and covariance matrix to characterize the color tendency and distribution range of the set. Calculate the Euclidean or Mahalanobis distance between the centroids of two point clouds. Simultaneously, the Bhattacharyya distance between two distributions can be calculated; these measures comprehensively reflect the overall differences in position and shape between the two color distributions. Map the calculated distance values ​​to the [0,1] interval using a monotonically decreasing function (e.g., S=exp(-α*distance)). The resulting scalar value S_preference is the color preference matching score. The closer the score is to 1, the more similar the main color distribution of the design scheme is to the user's preference.

[0085] This embodiment employs the K-Means clustering algorithm to perform color clustering on all pixels of the user-preferred image, automatically extracting the K dominant colors with the highest proportions. This replaces the traditional model of manual subjective judgment of user-preferred colors, avoiding the problems of color perception bias and vague descriptions (such as qualitative descriptions like "blue-toned" or "light gray") in manual identification, making the extraction of preferred colors objective and reproducible. The dominant colors obtained from clustering are represented in vector form in a predetermined color space, realizing the digital quantification of color information, transforming the originally abstract color preference into calculable and comparable quantitative data. A standardized color system based on hue, lightness, and saturation (HSV) is introduced, uniformly mapping the user-preferred dominant colors and the main colors of the design scheme to this system for quantitative positioning, eliminating differences in color perception. The inability to directly compare color parameters in color spaces (such as RGB and LAB) ensures consistency in the benchmark for preference matching. The three-dimensional division of the standardized color system achieves comprehensive coverage of color features. Compared with single-dimensional matching methods, it can more accurately capture subtle differences in color in dimensions such as brightness and vividness, making the preference matching results more in line with the user's true color perception and improving matching accuracy. By calculating the similarity between the user's preferred main color set and the design scheme's main color set in the standardized color space, "preference fit" is transformed into a specific numerical score, replacing the traditional binary qualitative judgment of "match / dislike" and "like / dislike". This makes the matching results continuous and distinguishable, and can accurately quantify the differences in the degree of fit between different design schemes and user preferences. In some optional embodiments, the color design optimization suggestion generation model used in step S4 is a regression model based on a feedforward neural network, and the main color feature of the input color scheme is a vector set of the main color of the color scheme in a predetermined color space.

[0086] Specifically, the input of the model is a multi-dimensional feature vector X constructed from the calculation results of steps S1 to S3 according to domain knowledge, including: Analytic hierarchy process scoring vector: The analytic hierarchy process scoring value calculated in step S1 is further decomposed. Specifically, the scores of five first-level indicators (brand recognition, environmental adaptability, target audience preference, functional semantics, process cost) are extracted to form a 5-dimensional sub-vector: , to provide more fine-grained positioning information; User preference vector: The user preference scores for N predefined interest areas calculated in step S2 are used as sub-vectors: . This vector objectively reflects the distribution of the user's visual attention on different design components / areas; Color preference matching degree scalar: The color preference matching degree score calculated in step S3 is introduced as a single feature scalar: . This scalar quantifies the overall fit between the design scheme and the personalized aesthetics of the target user group.

[0087] The current state feature of the scheme describes the objective initial conditions of the object to be optimized, and the main color vector set of the current color scheme in a predetermined color space (such as CIELAB) is standardized. For example, the LAB values of the first 3 main colors are extracted and concatenated into a 9-dimensional vector: .

[0088] Finally, the input feature vector X is constructed through a concatenation operation: . This vector X integrates all the key quantitative information of "where the score is insufficient", "where the user's attention is", "how much the gap is from the user's preference", and "what the current color state is", and can provide a complete data basis for the model to make accurate decisions.

[0089] The color design optimization suggestion generation model of this embodiment selects a feedforward neural network as the basic architecture. It learns a non-linear function that maps the high-dimensional, heterogeneous input feature vector X to a low-dimensional, continuous color adjustment parameter vector Y. The color design optimization suggestion generation model includes an input layer, a hidden layer, and an output layer.

[0090] Input layer: The number of neurons is equal to the dimension dim(X) of the input feature vector X, and it is used to receive all features without loss.

[0091] Hidden layers: These contain at least one fully connected layer. Each fully connected layer linearly transforms its input (weight matrix W and bias vector b) and then passes it through a non-linear activation function (such as ReLU). These layers are responsible for automatically learning and combining higher-order abstract features related to color adjustment decisions from the raw features. For example, the first layer might learn a feature combination for "when the brand recognition score is low and the main color is blue".

[0092] Output layer: This is a linear fully connected layer with the number of neurons equal to the dimension dim(Y) of the output vector Y. This layer maps the learned high-level abstract features to specific adjustment values.

[0093] The model's output is a color adjustment parameter vector Y, corresponding to operable parameters in the design software, including global adjustment parameters and key area targeted adjustment parameters. The global adjustment parameters are the suggested adjustments to hue (H), lightness (L), and saturation (C) for the main color of the entire scheme, denoted as [missing information]. The key region-specific adjustment parameters are used to output specific adjustment amounts for the top K regions of interest with the highest user preference scores in the user preference vector. The final output vector is .

[0094] The training data for the color design optimization suggestion generation model comes from a review of successful historical design iterations. Each training sample is a pair of... . The input feature vector is constructed according to the above method based on the historical scheme before optimization. The label represents the difference in parameters in the color space between the optimized and unoptimized schemes (i.e., the actual, verified effective adjustment amount). It is derived from the final adopted design scheme and represents the "correct optimization action".

[0095] Mean squared error is used as the loss function during training. This is used to minimize the gap between the adjustment parameters predicted by the model and the actual adjustment parameters made by the expert. The model parameters are iteratively updated on a large number of the above training samples using the backpropagation algorithm and an optimizer (such as Adam) until the model can accurately learn the complex mapping relationship between the multidimensional evaluation state and the optimal color adjustment action.

[0096] The above text provides a detailed description of an embodiment of a method for evaluating the color design of commercial vehicles. Based on the evaluation method for evaluating the color design of commercial vehicles described in the above embodiment, this invention also provides an evaluation system for the color design of commercial vehicles corresponding to the method.

[0097] Figure 2This is a schematic block diagram of a commercial vehicle color design evaluation system provided in an embodiment of the present invention. In this embodiment, the commercial vehicle color design evaluation system 200 can be divided into multiple functional modules according to the functions it performs. A module, as referred to in this invention, is a series of computer program segments that can be executed by at least one processor and perform a fixed function, and is stored in memory.

[0098] The hierarchical analysis evaluation module 210 is used to determine the weight of each indicator based on an evaluation index system that includes brand recognition, environmental adaptability, target audience preferences, functional semantics and process cost, and to calculate the hierarchical analysis score of the color design scheme using the hierarchical analysis method and the fuzzy comprehensive evaluation model.

[0099] The user preference evaluation module 220 is used to collect user eye movement data generated when the target user observes the color design sample. By calculating the total access time or the percentage of fixation times in the preset interest area, the user preference score is obtained.

[0100] The color preference evaluation module 230 is used to perform color clustering analysis and color system mapping on the color preference image provided by the user to obtain the user's color preference data. By calculating the similarity between the scheme to be evaluated and the user's preference data in the main color features, the color preference matching score is obtained.

[0101] The parameter adjustment suggestion generation module 240 is used to concatenate the hierarchical analysis score, user preference score, color preference matching score and the main color feature of the current color scheme to form an input vector. The input vector is then input into a pre-trained color design optimization suggestion generation model, which outputs adjustment parameter suggestions for hue, brightness or saturation.

[0102] The commercial vehicle color design evaluation system of this embodiment is used to implement the aforementioned commercial vehicle color design evaluation method. Therefore, the specific implementation of this system can be found in the embodiment section of the commercial vehicle color design evaluation method above. Thus, the specific implementation can be referred to the description of the corresponding embodiments, and will not be elaborated here.

[0103] Furthermore, since the evaluation system for commercial vehicle color design in this embodiment is used to implement the aforementioned evaluation method for commercial vehicle color design, its function corresponds to the function of the above method, and will not be repeated here.

[0104] Figure 3 This is a schematic diagram of a terminal 300 provided in an embodiment of the present invention, including: a processor 310, a memory 320, and a communication unit 330. The processor 310 is used to implement the process steps of the above-described embodiment of the commercial vehicle color design evaluation method when implementing the evaluation program for commercial vehicle color design stored in the memory 320.

[0105] This invention also provides a computer storage medium, which may be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc. The computer storage medium stores an evaluation program for commercial vehicle color design. When the evaluation program for commercial vehicle color design is executed by a processor, it implements the process steps of the above-described embodiment of the evaluation method for commercial vehicle color design.

[0106] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for evaluating the color design of commercial vehicles, characterized in that, Includes the following steps: Based on an evaluation index system that includes brand recognition, environmental adaptability, target audience preferences, functional semantics, and process cost, the weights of each index are determined by the analytic hierarchy process (AHP), and the analytic hierarchy process score of the color design scheme is calculated using a fuzzy comprehensive evaluation model. Collect user eye movement data generated when target users observe color design samples, and obtain user preference scores by calculating the total access time or the percentage of fixation times in preset interest areas; Color clustering analysis and color system mapping are performed on the color preference images provided by users to obtain user color preference data. By calculating the similarity between the scheme to be evaluated and the user preference data in terms of main color features, the color preference matching score is obtained. The hierarchical analysis score, user preference score, color preference matching score, and the main color feature of the current color scheme are concatenated to form an input vector. This input vector is then fed into a pre-trained color design optimization suggestion generation model, which outputs suggestions for adjusting hue, brightness, or saturation.

2. The evaluation method for commercial vehicle color design according to claim 1, characterized in that, The analytic hierarchy process (AHP) score is obtained through the following steps: A hierarchical evaluation index system for color design of commercial vehicles is constructed. The index system includes a target layer, a criterion layer, and an indicator layer. The criterion layer consists of first-level indicators, including brand recognition, environmental adaptability, target audience preference, functional semantics, and process cost. The indicator layer consists of second-level indicators, including at least three specific evaluation items belonging to each first-level indicator. A judgment matrix A is constructed based on the results of pairwise importance comparisons of indicators at the same level using the 1-9 scale method employed by experts. The eigenvalue method is used to calculate the largest eigenvalue and its corresponding eigenvector of the judgment matrix A, and the eigenvector is normalized to obtain the weight vector W of each indicator; the consistency ratio CR of the judgment matrix A is calculated, and the judgment weight is valid when CR < 0.

10. Establish an evaluation set V={v1,v2,...,vm}, and count the frequency of experts belonging to each evaluation level for each secondary indicator, calculate the membership degree, and construct a fuzzy evaluation matrix R; use a weighted average type fuzzy synthesis operator to synthesize the weight vector W with the fuzzy evaluation matrix R to obtain the fuzzy evaluation result vector B. The fuzzy evaluation result vector B is defuzzified to obtain the hierarchical analysis score.

3. The evaluation method for commercial vehicle color design according to claim 1, characterized in that, The system collects eye-tracking data generated when target users observe color design samples. By calculating the total access time or the percentage of fixations in preset interest areas, a user preference score is obtained, specifically including: The raw eye-tracking data of at least one target user observing a sample of color design schemes to be evaluated was collected using an eye tracker. The raw eye-tracking data includes the coordinates of the fixation point and the fixation duration recorded in time series. Based on the images of the color design scheme samples, at least one region of interest is defined by pixel coordinate range according to its different color regions and functional component regions; the gaze coordinates in the original eye-tracking data are mapped and matched with the pixel coordinate range of the region of interest to identify the effective gaze points falling within each region of interest. Calculate the total fixation duration of all effective fixation points within each region of interest. And calculate its total fixation duration along with that of all fixation points in this observation task. The ratio; or, the number of gazes accessed by each of the said regions of interest by effective gaze points. And calculate its number of fixations compared to the total number of fixations in this observation task. The ratio; The calculated gaze duration ratio or gaze frequency ratio is used as an objective preference score representing the user's preference for the color region represented by the corresponding interest area.

4. The evaluation method for commercial vehicle color design according to claim 3, characterized in that, Based on the image of the color design scheme sample, at least one region of interest is defined in pixel coordinate range according to its different color areas and functional component areas, specifically including: Gaussian filtering was applied to the RGB images of the color design scheme samples to reduce noise. The denoised image is converted from the RGB color space to the LAB color space; a clustering algorithm based on the color difference ΔE in the LAB color space is used to segment the image into several color blocks; Based on the pre-stored component template coordinate mapping table of the standard three-view drawing of commercial vehicles, the color blocks are merged to obtain at least one region of interest; wherein, the component template coordinate mapping table defines the coordinate range of the outline polygon of each physical component in the standard view.

5. The evaluation method for commercial vehicle color design according to claim 4, characterized in that, Based on the pre-stored component template coordinate mapping table of standard three-view drawings for commercial vehicles, color blocks are merged to obtain at least one region of interest, specifically including: For any color block, if more than a preset proportion of its pixels fall within the outline polygon range of a component in the mapping table, then the color block is determined to belong to that component. For all color blocks belonging to the same component, calculate the minimum distance between the boundary pixels of any two blocks; if the distance is less than the preset adjacent threshold, then the two blocks are determined to be adjacent. For all color blocks belonging to the same component, find all sets of blocks that are connected to each other through adjacency. Each set of connected blocks is merged and marked as an independent region of interest.

6. The evaluation method for commercial vehicle color design according to claim 1, characterized in that, Color clustering analysis and color system mapping are performed on the color preference images provided by the user to obtain user color preference data. The color preference matching score is obtained by calculating the similarity between the scheme to be evaluated and the user preference data in terms of primary color features. Specifically, this includes: Obtain at least one color preference image provided by the user; use the K-Means clustering algorithm to perform cluster analysis on all pixels of the color preference image in a predetermined color space, and extract the K colors with the highest proportion as the user's preferred main color set, wherein each main color is represented by a vector in the predetermined color space; Using the same K-Means clustering algorithm and a predetermined color space, cluster analysis is performed on the sample images of the color design schemes to be evaluated, and the M colors with the highest proportion are extracted as the main color set of the design scheme. The user's preferred primary color set and the primary color vectors in the design scheme's primary color set are mapped to a predefined standardized color system for quantitative positioning; the standardized color system is a system based on three dimensions: hue, brightness, and saturation. Calculate the similarity between the user's preferred primary color set and the design scheme's primary color set in the standardized color system space; use the calculated similarity value as the color preference matching score.

7. The evaluation method for commercial vehicle color design according to claim 1, characterized in that, The primary color feature of a color scheme is the set of vectors of the primary color of that color scheme in a predetermined color space; The model for generating color design optimization suggestions is a regression model based on a feedforward neural network.

8. An evaluation system for the color design of commercial vehicles, characterized in that, include: The hierarchical analysis evaluation module is used to determine the weight of each indicator based on an evaluation index system that includes brand recognition, environmental adaptability, target audience preferences, functional semantics, and process cost, and to calculate the hierarchical analysis score of the color design scheme using the hierarchical analysis method and the fuzzy comprehensive evaluation model. The user preference evaluation module is used to collect user eye movement data generated when target users observe color design samples. By calculating the total access time or the percentage of fixations in preset interest areas, a user preference score is obtained. The color preference evaluation module is used to perform color clustering analysis and color system mapping on the color preference images provided by the user to obtain the user's color preference data. By calculating the similarity between the scheme to be evaluated and the user's preference data in terms of main color features, the color preference matching score is obtained. The parameter adjustment suggestion generation module is used to concatenate the hierarchical analysis score, user preference score, color preference matching score, and the main color feature of the current color scheme to form an input vector. The input vector is then fed into a pre-trained color design optimization suggestion generation model, which outputs adjustment parameter suggestions for hue, brightness, or saturation.

9. A terminal, characterized in that, include: Memory for storing evaluation programs for commercial vehicle color designs; A processor, configured to perform the evaluation procedure for the commercial vehicle color design as described in any one of claims 1 to 7, implements the steps of the evaluation method for the commercial vehicle color design.

10. A computer-readable storage medium, characterized in that, The readable storage medium stores an evaluation program for commercial vehicle color design, which, when executed by a processor, implements the steps of the evaluation method for commercial vehicle color design as described in any one of claims 1 to 7.

Images