低複雜度之濾波掃瞄轉換

Abstract

本論文主要是提出低複雜度的影像升降頻的方法、簡單的濾波器設計以及快速濾波演算法。這些方法的表現及複雜度也會一併在此討論。現有的影像升降頻方法很難在複雜度和表現上兼顧，有好的表現者通常需要較高的運算量，另一方面，簡化的設計通常也容易達不到滿意的效果。因此我們將提出有效的升降頻方法來改善這些缺點。此外，為了進一步的簡化整個影像升降頻的處理過程，我們也會改良現有的一些簡單的濾波器設計方法並提出新的快速濾波演算法。

Low-complexity image resizing algorithms, filter design algorithms and fast convolution algorithms were presented in this thesis. The performance and comparison of these algorithms were also evaluated. The existing algorithms are either complicated while having good performance or low-performance while reducing the complexity. Therefore, we proposed efficient decimation and interpolation algorithms to improve these drawbacks. Moreover, simple filter design methods modified from some existing algorithms and fast convolution algorithms were developed to reduce the overall complexity in the image resizing processing.