使用陀螺儀及影像合成技術達成數位影像穩定

Abstract

近年來，數位相機及手機相機的主要發展趨勢為小體積、高畫素、及高倍率變焦鏡頭。在這種條件下，拍出來的相片很有可能因為我們手部輕微的抖動，而產生模糊的、品質不佳的相片。這種現象特別容易在使用長曝光及高倍率變焦的情況下發生。為了解決這個問題，有各式各樣不同的防手震技術被提出來以提升整個拍照影像的清晰度與品質。 在這篇論文中，我們提出從三張影像並經過陀螺儀之偏移計算補償合成出一張沒有手震影像之演算法。其主要之方法就是利用提升快門速度並且連續拍出三張相片，而在拍照過程中藉由陀螺儀記錄其座標之偏移量，再藉由此資訊將3張照片合成，重建出一張亮度正常且清晰之影像。我們並且於現行之智慧型手機上實作我們的演算法、證明我們的方法是可行的、而且是高效率的。

In these years, the major trends for digital cameras and mobile phone cameras are small size, high resolution, and high magnification zoom lens. Under such conditions, the captured images may be blurred by slight hand shaking to produce the bad quality images. This phenomenon is particularly easy to occur in the case of using long exposure shutter and high-magnification zoom. To solve this problem, there are various different anti-shake technologies proposed to enhance the clarity and quality of the captured photos. In this thesis, we propose an algorithm that could produce a blur-free image by using the gyroscope to calculate the compensation during capturing and compositing the continuous 3 captured images. The major approach is by increasing the shutter speed then capturing 3 continuous images. During the capturing process, we use the gyroscope to record the coordinates offset then combine the 3 images with the compensation information to reconstruct an image with clarity and normal brightness. We also implement our algorithm under the smart phone device, the results show our approach is feasible with high efficiency.