摘要 在數位影像擷取元件日益發達的今日社會，透過各式各樣的影像擷取裝置，如CCD及CMOS去擷取一個影像已經是十分平常的問題。 在整個成像元件所組成的系統中，如何鑑別一個影像是否為最佳聚焦影像是一個相當重要的研究方向。在主動式的自動聚焦方法中，經常透過雷射光及超音波等的主動訊號發射器，再配合訊號接收器。如此一來便可知道光學系統與物體之間的關係。但是這樣一來對系統的整個體積、重量以及製造成本方面，都是比較大的負擔。 在本篇論文中介紹了靜態的影像鑑別方法，透過系統離焦對影像對比的改變來對系統的狀態做定量的工作。由空間頻率開始，計算空間頻率隨著離焦程度的衰減。再將衰減的量反應至邊緣的損耗最後透過統計的方法求得一定值。除了自行設計的物空間圖案外更對真實空間的人物及風景做一評鑑，在距焦附近的效果是顯著的。在文章的最後除了直接對灰階值做運算處理之外，更對系統的穩定度及運算時間做一比較及說明。本文所做的一切都希望將來能應用為離線操作。 Abstract In modern society the digital imaging devices are popular productions. Generally, CCD and CMOS image sensor are used to be an important element in common digital device. The most importance of an imaging system is to evaluate the image quality and the entire system situation and then modulate itself to the perfect situation. In active auto focusing method the laser and ultrasonic emitter source and signal detector are necessary. Active methods are accurate than passive methods, but the cost of manufacture and volume and weight of the entire system are higher than passive methods. In this thesis, we introduce a static image evaluation. The decrease of contrast will be measured to evaluate the situation of the entire system. We calculate the decay of spatial frequency spectrum versus defocus level first. The information related with edges will be calculated to evaluate the situation of the system by statistics methods. Besides the designed pattern object, we use the real image to be a test of our method. The effect is obvious near the focal point. Finally, the stability of the system is measured. The methods mentioned in this thesis will be used in off line situation, in the future.
