應用感性語彙與意象認知於電腦輔助銀行業幾何型商標設計之建構

The Application of Kansei Vocabulary and Image Recognition Analysis to the Computer-Aided Bank Geometric Logo Design

企業識別主要包含三個要素，即：（1）理念識別、（2）活動識別與（3）視覺識別，其中以視覺識別的商標在視覺化設計上最具影響力。不過，由於商標設計在構想發展階段，仍有賴於設計師或企業決策主管個人主觀見解、經驗與構思，使得商標設計的過程，不易反應消費者的認知需求，值得進一步探討與改善。本研究運用解構商標元素，並重新建構商標設計的法則，以探討應用設計法則於商標設計之可行性。整體發展過程中，以國內外現有銀行業的幾何型商標為案例，除用感性語彙探討其與國內外銀行業幾何商標間之關聯性外，並使用語意差異法，配合感性工學語彙之評量，對受測者進行測試。最後，將所得資料運用指示型多元迴歸與相關性分析，探討感性語彙與商標建構特徵間的關聯性，以轉譯為法則，並透過電腦程式之發展產生相關之商標，來協助設計師及企業決策者設計商標時的方向與參考依據。本研究所導出之幾何型商標設計法則包含：（1）移動且旋轉、（2）分割、（3）重複且移動、（4）重疊且移動、與（5）旋轉且縮放等五種型態。本研究之結果並經受測者的認知驗證，具80%以上之接近度，預期將可協助設計師有效地發展商標設計，並可快速產生較符合顧客需求的可行商標構想案。

The characteristics of a general corporate identity system includes: （1）mind identity, （2）behavior identity and （3）visual identity. The visual identity plays the most influential role in the procedure of visual design, especially in logo design. Unfortunately, most logo designs are still heavily relied on designers or business chiefs' subjective opinions, experience, or ideas that make the logo design can not meet customer's needs. The objective of this research effort is to disassemble logo elements and restructure logo design rules for the development of a computer-aided logo design system. The banks with geometric logos are chosen as a case to explore the feasibility of the research. In the development process, concepts of Kansei vocabulary, semantic differential method and statistical indicator regression analysis are applied to measure the relationship between customers' logo recognition and logo features. The relationships are then transformed into design rules. The rules connect customer requirements and geometric forms that will be used in the development of computer program for the generation of related logos. In this research, the rules for the geometric logo design includes: （1） movement and rotation, （2） segmentation, （3） repetition and movement, （4） overlapping and movement, and （5） rotation and enlargement/reduction. A validation of the developed system is performed by the same group of subjects. It appears that logos generated by the system have about 80% of closeness to the customers' image recognition. Therefore, the results should provide a useful method with designers in logo design that can meet customers' image requirements.