商標是一種藉由圖像或是混和圖像與文字存在的標誌，經由特定的符號或是顏色所組成，利用圖像的組成、配置、形狀、顏色、混合、字體、字體顏色等視覺元素將會帶給人們不同的視覺感受。 過去的研究中，對於視覺元素在商標中扮演的角色偏向質化的探討，通常多利用使用者測試的方法找到設計元素與商標的相關度，較少應用電腦視覺對於圖像量化分析的方法，本研究的目的在於利用各種視覺特徵的計算方式，分析各種商標中視覺元素的組成，包含商標中圖像的複雜度、和諧度、組成重複度等相關量化指標。 藉由大量透過網路上收集的商標進行分析後，可發現商標普遍具有低複雜度、高和諧度、低重複度的設計特性，本論文最後試著以商標獨特性指標代替商標辨識度以驗證本論文設計的分析方法可行性。由於在電腦視覺領域中相關於商標美學的研究並不常見，本論文亦希望能對電腦視覺分析應用在設計或美學的領域進行先導性的研究。 / A logo is a mark composed of graph or a combination of text and graph. Typical visual elements in a logo design such as layout, shape, color (foreground and background), composition, and typeface. The graphical mark can exhibit interesting properties by mixing the elements in creative ways. Most previous researches regarding the role of visual elements in logo design are of qualitative nature. In this thesis, we propose to incorporate visual feature extraction and analysis algorithms commonly utilized in computer vision to compute proper index and investigate key visual elements in logo design, including complexity, harmony and repetition. After analyzing large amount of logos collected from the internet, we find out that most logos are of low complexity, high balance and exhibit some degree of repetition. We propose a new measure of “distinctiveness” and investigate its relationship with to the aforementioned properties. We hope that the results obtained in thesis serve as a catalyst to motivate further research in applying computer vision methods to the area of aesthetics and design.