A Study on the Gestalt Organizational Differences in the Exterior Design of Electric Vehicles — CNN-based Visual Feature Analysis

초록

This study aims to explore the differences in the visual composition of exterior designs for electric vehicles (EVs) in China and Japan by combining Gestalt theory with a Convolutional Neural Network (CNN)-based visual feature analysis method. Through this, the study seeks to uncover the visual perception hidden behind these different design languages and to provide a theoretical foundation and practical reference materials for cross-cultural automotive design research and brand differentiation strategies. analysis methods regarding the visual composition characteristics of EV exterior designs in both countries is still lacking. To objectively compare differences in visual features and provide a new analytical perspective and methodological reference material for crosscultural automotive design research, this study conducted the following research: First, the styling elements of EV exterior design were analyzed, and research utilizing Gestalt theory and Convolutional Neural Networks (CNN) was explored. Second, the BYD Qin PLUS EV and Toyota bZ3 were selected as representative cases, and the front, side, and rear images of the two vehicles were processed identically. Finally, a comprehensive comparative analysis of the design characteristics of the two models was conducted by combining the visual feature extraction results from CNNs with Gestalt theory. The research results and content are as follows: First, Chinese electric vehicles exhibited design trends centered on visual convergence, dynamic rhythm, and hierarchical contrast, forming an expression system dominated by perceptual stimulation and visual guidance. Second, Japanese electric vehicles primarily adopted a geometric design approach, characterized by simplicity, spatial balance, and structural consistency, creating a more restrained and rational visual impression. Based on these research results, it is expected that expanding the sample to include various brands and models in the future will enhance the generalizability and explanatory power of the findings regarding design visual mechanisms.

키워드