基於內容為主之雙眼立體圖片顯像調適

Abstract

本篇論文提出一基於內容為主的雙眼立體圖片顯像調適之方法。本方法將一立體圖片縮放至顯示設備之長寬比、調適其深度至顯示設備的舒適範圍、並且同時維持圖片的重要特徵。我們基於一以形變為基礎之二維影像縮放演算法，藉由對特徵點的垂直校正及水平位移的一致性加入限制， 形變較為不重要的區域，以保持重要區域的形狀及深度。此外，我們提供使用者操作介面，使其能夠互動式地改變所選取物體之大小與感知深度。實驗顯示，本方法能夠呈現出逼真的深度感受，同時最小化複視 (diplopia) 的發生與重要物體的形變。隨著立體圖片變的越來越普及，本方法能夠滿足使用者於編輯雙眼立體圖片時的以下需求：在不同的顯示設備都能適當地顯示立體圖片，以及加強深度感知的效果。

This thesis proposes a content-aware stereoscopic image display adaptation method which simultaneously resizes a binocular image to the target resolution and adapts its depth to the comfort zone of the display while preserving prominent features. Our method is based on a warping-based 2D image retargeting method. By adding constraints to enforce vertical alignment and horizontal disparity consistency of features, our method keeps the shape and depth of salient area as much as possible while distorting less prominent areas. In addition, our method allows users to interactively change the size or depth perception of the selected object. Experiments show that our method is successful in providing a vivid sense of depth while minimizing the occurrence of diplopia and distortion of prominent objects. As stereoscopic images get more and more popular, our method fills users' needs of editing their binocular images to be displayed properly on various displays or to enhance the depth perception.