- 學位論文
以特徵圖資訊為基礎的網球影片虛擬廣告植入方法
Virtual Advertising for Tennis Highlight Video using Salience Information
摘要
隨著多媒體內容在網路上的普及。近幾年這趨勢引發了虛擬廣告技術的誕生，尤其是針對能帶來巨大商機的運動影片。虛擬廣告的想法是希望將品牌商標和其他廣告訊息置入影片，使觀眾可以專注在觀看比賽的同時又會不自覺的對廣告產生印象。然而，利用人工方式來做這項工作是很費時且沒有效率的。基於上述理由，在這篇論文中，我們提出了一個全方面性的虛擬廣告置入方法用在網球影片上，希望提升廣告的有效性同時又不會增加觀眾的視覺干擾。在廣告置入時間方面，我們根據網球視頻結構分析和視覺特徵來檢測適當的置入時間點。置入影片位置的選擇，我們則是基於人類的視覺系統分析，考慮空間和時間關注度來避免廣告遮蔽住影片重要部分。同時，為了增加廣告的有效性，廣告位置的重要程度也納入決定考量。接著，廣告選擇機制和色彩呈現方式用來提升用戶體驗同時也減少視覺干擾。由實驗結果證明，我們提出的自動偵測置入方式是有效且適當的。除此之外依照規定的準則來選擇廣告置入時間和位置的結果顯示和使用者測試研究也是一致的。
並列摘要
The community contributed media content on the Internet have become one of the primary sources. This trend triggers the virtual advertising comes up recent years, especially for sport videos which bring tremendous commercial potentials. The concept of virtual advertising is to bring brand names or other advertising messages into scenes, thus the audience can focus on the game while impressed with advertisement unconsciously. However, the task is labor intensive and inefficient by manually. In this thesis, a novel approach to build up a more comprehensive scheme which will equilibrate on promoting cognitive effectiveness but curtailing visual intrusiveness for tennis highlight videos is proposed. This approach determines appropriate insertion timing according to the structure of video sequence and the visual features gathered in each frame. The inappropriate locations are masked using the proposed attention analyzing procedure through spatial and temporal examination. In order to promote advertising effectiveness, two considerations of significant localization also take into account in the implementation. Then, an advertising selection mechanism with color relevance is implemented to reduce visual interference from the perspective of user experiences. The objective assessments demonstrate that this advertising insertion approach is both effective and adequate. In addition, the subjective evaluation results show that the insertion method with predefined criteria is accordant to user study experiments.
並列關鍵字
無資料參考文獻
延伸閱讀
- Hsieh, K. Y. (2008). 以視覺模型為輔助且適用於網球影片之虛擬探照游移廣告系統 [master's thesis, National Taiwan University]. Airiti Library. https://doi.org/10.6342/NTU.2008.00818
- 林志煒（2005）。平面廣告圖像符號之隱喻性多空間模式探討與應用－以時報廣告金像獎（1997?2003）得獎作品與時報廣告金犢獎（2005）創作作品為例〔碩士論文，崑山科技大學〕。華藝線上圖書館。https://doi.org/10.6828/KSU.2005.00082
- 陳禮猷（2005）。網路虛擬經驗類型組合之廣告效果〔碩士論文，國立臺北科技大學〕。華藝線上圖書館。https://doi.org/10.6841/NTUT.2005.00272
- 張鴻閔（2007）。以視覺特徵擷取為基礎之影像視訊內容認證技術〔碩士論文，國立中央大學〕。華藝線上圖書館。https://www.airitilibrary.com/Article/Detail?DocID=U0031-0207200917350594
- Yu, C. H., Wu, C. C., Wang, J. S., Chen, H. Y., & Lin, Y. T. (2020). Learning Tennis through Video-based Reflective Learning by Using Motion-Tracking Sensors. Educational Technology & Society, 23(1), 64-77. https://doi.org/10.30191/ETS.202001_23(1).0005