數位取像系統中像素內插、影像縮放及對比加強等技術之研究

Abstract

在本論文中將討論數位取像系統中的四樣工作：像素內插、影像縮放、邊角加強、亮度對比加強。其中，在像素內插的研究，我們採用了一套根據圖樣(pattern)類型再配以多項式近似的方法，所得內插結果不論是假色(false color)以及格子效應(lattice effect)出現的機率較原始方法都減少很多。影像縮放中，Cubic-spline內插及B-spline內插都是缺陷很少，計算點也少的有效內插方法。邊角加強技術上，我們採用了調整各點放大倍率的方式，得到了邊界銳利、環狀缺陷(ringing artifact)不明顯、且雜訊不被加強的加強結果。在亮度對比加強中，我們採用調整亮度值柱狀圖(histogram)的方式：首先，我們選擇了邊界上的亮度值為我們統計柱狀圖的資料來源。再來，我們根據統計出的柱狀圖特性做相對應的調整。而得到一個亮度分配較有效率、對小面積物體間的對比值有較適當調整、且較符合影像原始感覺的演算法。

In this thesis, we propose several approaches to do pixel interpolation, image scaling, edge enhancement, and brightness enhancement in digital image acquisition systems. About pixel interpolation, we propose a pattern-based method and an interpolation method based on polynomial fitting. These two approaches reduce the appearance of false color and lattice effect. About image scaling, we discuss both cubic spline interpolation and B-spline interpolation. About edge enhancement, we adjust the enhancement factor at each pixel according to the image content near that pixel. This approach can provide sharp edges, fewer ringing artifacts, and lower noise enhancement. About brightness enhancement, we select the pixels around edges and build a hitogram for hem. Then, by modifying the histogram in a proper way, this approach provides enhanced brightness without affecting the perception of the original image.