A Design Approach for Mei Gui Chairs Based on Multimodal Technology and Deep Learning

Understanding the influence of the morphological mechanisms of Mei Gui chairs on the emotional preferences of female users is crucial for achieving perceptual resonance in design. This study aims to investigate/explore the relationship between user preferences and design features to create furniture with greater emotional resonance. (1) Background: To develop a scientifically validated model for predicting user preferences in Mei Gui chair design by emotional factors and morphological mechanisms. (2) Methods: (a) Data Collection: Establish a dataset of Mei Gui chair morphological mechanisms based on the visual sequences of female users using the KJ method, factor analysis, K-means clustering, and triangular fuzzy numbers. (b) Preference Analysis: Use Eye-tracking Technology to identify female users’ preference areas for Mei Gui chair morphology and construct a morphological element preference library. (c) Feature Classification: Categorize the extracted feature elements into five classes. (d) Neural Activation Analysis: Utilize near-infrared brain functional imaging technology to conduct paired-sample T-tests on the five classes of features, identifying preferred backrest characteristics. (e) Model Validation: Integrate three factors (elegance, delicacy, comfort) into the final design scheme and compare the performance of the proposed EMD-KPCA-LSTM model with traditional BP neural network, SVM, and CNN models. (3) Results: The EMD-KPCA-LSTM model outperforms traditional models in capturing the relationship between user preferences and morphological mechanism design features, demonstrating higher predictive accuracy, better generalization ability, and stronger robustness. (4) Conclusions: The proposed model effectively integrates user preferences with Mei Gui chair design, providing a scientifically validated method for perceptual prediction in furniture design.