- 學位論文
運用重疊線來實現二維對三維影像轉換系統與晶片設計
Using Overlapping Line to Achieve Two Dimensional to Three Dimensional Image Conversion System and Chip Implementation
摘要
近年來隨著3D電影工業的蓬勃發展，也帶動了整個3D產業發展，除了有越來越多3D電影出現，也陸續有廠商相中此波3D熱潮紛紛推出相關產品，例如立體螢幕、搭載立體螢幕之筆記型電腦、立體相機、立體電視、立體手機等電子產品，由此可發現廠商大多推出之產品皆是顯示端，反倒是內容端來源不如顯示端發展迅速，使3D產業發展受到限制，為了改善此困境便思考如何增加內容物，最直覺的方式便是利用專門硬體設備拍攝，但此種方式雖能解決此窘境但卻面臨幾種隨之而來的問題，除了會增加硬體成本之外，拍攝出之影片通常還需要經由專業後製處理而大大提高成本，隨之而來的解決方案便是二維影像至三維影像之轉換技術，而在二維影像至三維影像之轉換技術之中，最重要的要點莫過於深度資訊圖之產生，深度資訊圖的優劣決定了後面產生之三維影像品質，而所謂深度資訊圖便是圖片當中物件的深遠關係圖，將圖片中所有物件之遠近關係找出來後，建立出一張深度資訊圖，再透過DIBR演算法在立體顯示螢幕上重現立體影像，本論文預計提出一個自動產生深度資訊圖的演算法，主要是利用重疊線來判斷物體之間之遠近關係，藉此重建出完整的深度資訊圖。
並列摘要
In recent years, the 3D film industry is growing fast. It has also led to the development of the entire 3D industry. In addition to increase more and more 3D movies, the manufacturers have introduced related products in this 3D craze. The growing of content source is less than displayed equipment. It will restrict the development of 3D industry. To improve this problem, we proposed a two dimensional to three dimensional image conversion system. The generation of depth map is most important in the system. Depth map can define the relation of object position. The quality of depth map will affect the 3D display result. A depth map image plus an original image can be synthesized to multi-view image by DIBR algorithm. We will propose an automatic depth map generation algorithm, it uses the overlapping line to achieve two dimensional to three dimensional image conversion system and chip implementation in this thesis.