透過您的圖書館登入
IP:3.139.238.76
  • 學位論文

利用三維頭部資料建立男性國人頭型及臉型分類與應用

Classification and Application of Taiwanese Males’ Head Shapes and Face Shapes by Using 3-D Head Scanning Data

指導教授 : 王茂駿

摘要


本研究利用三維頭型掃描方法重新建立符合現代台灣人頭部資訊資料庫,共蒐集300位18~30歲之男性國人頭部資訊,進行三維頭型掃描儀的精確度(precision)與準確度(accuracy)驗證,以確保數據的真實性;並將所蒐集到的頭部資訊分別進行頭型與臉型兩部份的分析與探討。 根據三維頭型掃描儀掃描頭部資訊可進而計算出63項頭部相關尺寸。頭型部份利用主成分分析找出「頭長」、「頭高」與「頭寬」三項主要影響頭型之重要尺寸參數,並利用資料挖礦中之群集分析法,將國人頭型分為「短淺寬型」、「短淺窄型」、「長深寬型」和「長深窄型」四種代表類型。頭型分類尺碼部分,考量實務運用與相關研究之做法,分別建立四種頭型分類尺碼,涵蓋率皆達93%以上。在實務應用方面,套用一套合邏輯的縮放機制,針對頭型資料進行重複驗證,頭型縮放之平均誤差值為10.20 mm,並利用電腦輔助設計軟體,將縮放機制實際應用於本研究所提出之分群後代表頭型,以期達到具有快速且具高解釋程度之效,以應用在頭部相關產品設計之參考。 臉型部份利用主成分分析找出「臉長(型態面)」、「臉寬(顴骨間距)」與「鼻樑點至右耳珠水平深度」三項主要影響臉型之重要尺寸參數,並利用資料挖礦中之群集分析法,將國人臉型分為「長淺窄型」、「短淺窄型」、「長深寬型」和「短深寬型」四種代表類型。臉型分類尺碼部分,考量實務運用與相關研究之做法,分別建立兩種頭型分類尺碼,涵蓋率皆達92%以上。在實務應用方面,套用一套合邏輯的縮放機制,針對臉型資料進行重複驗證,臉型縮放之平均誤差值為7.49 mm,並利用電腦輔助設計軟體,將縮放機制實際應用於本研究所提出之分群後代表臉型,以期達到具有快速且具高解釋程度之效,以應用在臉部相關產品設計之參考。

並列摘要


This study used 3D scanning method to collect head and face anthropometric data of 300 Taiwanese males aged 18~30 years. In addition to the analysis of head shapes and face shapes, the precision and accuracy of the 3D head scanner were also evaluated in this study. Principal component analysis of head shapes was applied to obtain the important dimensions, including head length, total head height and head breadth. Clustering analysis was performed to identify four head shapes including: short-shallow-wide, short-shallow-narrow, long-deep-wide and long-deep-narrow. In addition, we established four head shape sizing systems with coverage over 93%. Moreover, this study used a reasonable deformation logic to rapidly change head shapes with at mean error 10.20 mm. This result can provide important reference for designing and manufacturing head products. Next, principal component analysis of face shapes was applied to obtain the important dimensions, including face length (morphologic face height), face breadth (bizygomatic breadth) and sellion to right tragion depth. Clustering analysis was performed to identify four face shapes including: long-shallow-narrow, short-shallow-narrow, long-deep-wide and short-deep-wide. In addition, we established two face shape sizing systems with coverage over 92%. Moreover, this study used a reasonable deformation logic to rapidly change face shapes with mean error around 7.49 mm. This results can provide useful information for designing and manufacturing face products.

參考文獻


6. 許馨文,2012,以三維足型資料建置客製化鞋楦,國立清華大學工業工程與工程管理研究所碩士論文。
4. 李建璁,林彥輝,楊宜學,陳志勇,2006,應用國人頭型資料庫探討安全帽尺寸分類,勞工安全衛生研究季刊,民國95年,第14卷第1期,第48-57頁。
9. 游志雲、葉文裕、楊宜學、張碧慧,1996,勞工頭型模式之研究:台灣勞工頭型資料庫與標準頭型之建立,勞工安全衛生研究季刊,民國85年6月,第4卷第2期,第31-46頁。
44. She, Q. Y. (2002). Practice book of knowledge and standard of individual protect equipment. Hubei Press of Science and Technology, Vol. 14-6.
10. Anderberg, M. R. (1973). Cluster Analysis for Applications. New York: Academic Press.

延伸閱讀