市面上,產品的形態林林總總,其造形風格各有不同,如何加以分類,並以量化方式描述,值得設計界的探討。本文以轎車為例,建立汽車的圖形檔案,應用產品計測等原理,量測其相關的尺寸與角度。這些圖片由受試者依其形態分類,並從形容詞庫中,選出適合描述各類車的形容詞。其結果經群集分析,而將汽車加以初步分群。各群並經計算,找出和同群其他車輛的平均相似度最高的車輛,作為該群的典型車。同時,汽車側面輪廓線的曲線部位經界定後,從各對應曲線的坐標,求得其回歸曲線,將此曲線配合以產品形態計測對應部位的平均尺寸與角度,而架構出各群汽車的平均車型的輪廓。把各群典型車的圖像依比例轉貼到對應的平均車形輪廓上,即可產生各群的原型車。理念上,原型車將可作為該群汽車外表形態及意象的代表車型。另一方面,各類汽車意象調查的結果發現,描述汽車形態的形容詞有三個主要意象因素:豪華感,典雅感與流動感。本研究所所用汽車圖形樣本經多種群集分析,綜合歸納後,可得到5個獨立的汽車類群:靈巧的,豪華的,簡樸的,典雅的及流動的。
Products on the market have various forms with different styles. The theme on how to classify and describe them quantitatively is indeed worth exploring in the field of design. This paper compiles 40 side view pictures of recent year sedans and converts them into digital files. The key dimensions and angles of the cars in the pictures are then measured according to productometry, a technique for measuring products. A total of 40 subjects are then asked to divide the 40 car pictures into groups in terms of shape similarity, and select adjectives, from a given set, to best describe each car group. A cluster analysis reveals that, there are 4 major clusters of the car shapes. For each cluster, the car which has a least average dissimilarity to all other cars in the cluster is then identified as the typical car of the cluster. At the same time, the corresponding key curves of the car contours in the same cluster are gathered, their coordinates scatter plotted, and a regression curve calculated by regression analysis. By combining the regression curves and average dimensions and angles obtained from productometry in each cluster, an average contour shape of each car cluster can be constructed, which is then pasted with the picture image of the typical car of the same cluster. The resulted car image is then treated as a prototype car of this cluster. Conceptually, a prototype car can be considered as the representative model of its respective cluster both in shape and in image. Luxury, elegance, and dynamism are the three main factors found by factor analysis of the 21 image words used to describe the car shapes. The 40 cars are cluster analyzed again both in the image space of the original 21 image words and in the image space of the condensed three factors, resulting 3 and 4 clusters respectively. By synthesizing all the three clustering results in this research into a final classification of the cars, five independent car styles can be identified: cute, luxurious, compact, elegant, and aerodynamic.