Summary: | 碩士 === 國立雲林科技大學 === 機械工程系 === 102 === Nowadays, the equipment of photographing and sensing trend to miniaturization increasingly. And when this equipment combines with communication apparatus, it becomes more and more popular in our daily life. However, handing this equipment to record image sequence will make many chaotic components and affect the vision of video easily. That because the unusual shaking from human or external influence. In order to overcome the problem, this thesis combined existential Image Stabilization System to resolve instability of image sequence.
In this thesis there are two matching modules, including digital image stabilization system and acceleration sensor system. And there are three components operating by these two systems, including Motion Estimation, Motion Smoothing and Image Compensation. Nowadays, the resolution of video becomes much higher. In order to deal with the large calculating among when using motion estimation of higher shooting resolution, this thesis utilized a new motion estimation searching algorithmmentioned by predecessor. This algorithm provide with a large capacity and a high accuracy of searching. It used LabVIEW to combine image sensor and accelerometer to sense hand-shake. And the system can capture Pixel Shift and acceleration values in the same time. And then, it can capture the best result of coefficient of transference function of acceleration to pixel displacement by the training of LMS by LabVIEW. In this way, we can obtain transference function, including the smoothness of experimenter’s hand-shake, the unit transference of acceleration to displacement, the environment interference factor and more. Through these processes, it can present a robust and efficient image stability system.
|