用於智慧型手持裝置影像穩定之影像融合演算法

Abstract

現今許多智慧型手持裝置都配備了數位相機模組，隨著這些手持裝置越來越輕，伴隨而來的就是拍照時越容易晃動的問題。數位相機上都配有影像穩定的機制，但多數為光學式的防手震處理，若是使用數位式的影像穩定處理則可以達到成本降低的目的，尤其是手機這種智慧型手持裝置是很難去使用好的鏡頭來做光學式防手震處理的。 在本篇論文中，針對在智慧型手持裝置上的數位式影像穩定處理，我們提出了一個可以有效降低計算量的演算法，使得影像穩定在手機上可以完全以數位式的方式處理，而不會有運算過久或者無法運作的情況發生。

Many handheld consumer electric devices with camera module are developed to be lighter and smaller. The problem is easy to hand-shake when user take photos. To overcome this condition, many digital cameras avoid image blur with optical image stabilize system. If we use digital image stabilize to solve this problem, the total cost will be reduce a lot. It is hard to use good camera shutter to achieve optical image stabilization on hand-held consumer electronics. In this thesis, we propose an efficient algorithm for digital image stabilization on consumer electronic device. To assure image stabilization, we reconstruct a good quality image by fusion several short-exposed degraded dark image of the same scene. And this algorithm can truly work on consumer electronics with short execution time.