Please use this identifier to cite or link to this item:
標題: 數位相機模糊自動對焦
Fuzzy Auto- Focus Approaches for Digital Cameras
作者: 陳英進
Chen, Ying-Chin
關鍵字: 數位單眼相機;自動對焦;模糊邏輯
出版社: 電機工程學系
本論文主要發展數位相機之模糊自動對焦演算法,用以增強數位相機之對焦功能,該項研究的目的是針對自動對焦系統受光線、亮度、被攝物等條件的影響以致難以對焦,以及一些特殊情況下會出現對焦偏差等問題,提出適用的模糊演算處理模式,使數位相機能判定當時對焦狀況,並修正對焦量,進而實現拍攝正確對焦的成像。本文首先詳述多點自動對焦系統與校正技術,應用模糊自動對焦演算方式於多重基點記憶影像感應器(Multi BASIS)對焦感測處理,利用模糊關聯量運算理論完成被拍攝物體在低明暗對比時的自動對焦,以及運用最小平方法觀念計算出運動物體拍攝前所須補正的離焦量。許多實驗結果被用以說明所提議方法之可行性與有效性。本項開發技術將可使用於數位相機、攝影機、監視器以及需要正確對焦的光機電設備。

This thesis presents fuzzy autofocus algorithms for single-len digital cameras. These algorithms are used to strengthen or improve present autofocus functions of the digital cameras; in particular, they are aimed to slove for some difficult focusing problems occurring in presence of high light-dark constrast, low luminanace, and fast-moving objects, etc. To overcome the aforementioned problems, the developed approaches will enable the digital camera to judge its focusing information and then execute designed fuzzy rules to calculate the appropriate focusing increment in order to obtain correct image formation while shooting. In the thesis, the sytem configuration and calibration of a multi-point autofocus camera with muti-base point strored image sensors are described in detail. A practical fuzzy autofocusing algorithm is proposed via the multi-base point stored image sensors, and then used to deal with the autofocus problem of the subjects with high light-dark contrast. To shoot the fast-moving objects in a more clear way, a fuzzy autofocus method together with the well-known least-squares motion prediction method is presented to calculate adequate focusing increment before shooting. Several experimental results are used to show the feasibility and validity of the proposed methods. These technologies are expected to be useful for commercial digital cameras, cinecameras, monitors and any optical-electromechanical equipment with autofocusing.
Appears in Collections:電機工程學系所

Show full item record
TAIR Related Article

Google ScholarTM


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.