適用手持式掃瞄裝置之影像接合演算法

Abstract

本論文提出一個適用於新式手持式掃描裝置之影像縫合演算法。這種新式手持式掃描裝置結合了攝影鏡頭與光學滑鼠感測器，利用滑鼠感測器估計攝影鏡頭之位移量，可以將連續拍攝的多張影像接合為一張大圖。然而滑鼠感測器測量得到之位移量有一定程度的誤差，本論文提出之演算法可以有效校正此誤差，進而得到更好之縫合結果。

An image stitching algorithm for a novel type of hand-held scanning devices is proposed in this thesis. These hand-held scanning devices are combined from video cameras and optical mouse sensors. With optical mouse sensors, we can estimate the movement of video camera when users scan the document. The scanner can stitch image frames captured by the video camera to a whole image with the movement estimation from the mouse sensor. However the estimations obtained from the mouse sensors have a certain amount of errors. In this thesis, we propose an algorithm to correct these measurement errors. Better stitched results can be achieved with the correction by our algorithm.