利用分群濾波器於影像雜訊濾除

Abstract

本論文是研究一個可保存影像細節的濾波器, 用於影像 。它是改良自一 種影像壓縮的前置濾波器,由Wong所提出分群濾波器。分群濾波器基本上 是一種平均濾波器,由指數函數控制加權值。加權值可被調整使得平滑區 被抹平而邊緣區被完整保留。所做的改變可分兩方面: 第一, 我們提出另 一種調整加權值的方法。第二, 我們設計一種尋找脈衝雜訊之法則。由模 擬結果得知, 此改良型的分群濾波器有良好的濾除雜訊能力同時保持影像 邊緣及細節。 In this thesis, we develop an efficient detail-preserving filter for image enhancement. It is modified from the clustering filter by Wong which is used as a preprocessing filter for image compr- ession. Basically, the clustering filter is an averaging filter with an exponentially weighting window. Weights inside the window can be adaptively adjusted, such that homogeneous regions are highly smoothed and edges are sharply preserved. Two modifi- cations have been made: First, we propose a different weights- adjusting strategy. Second, we devise a criterion to find the pixels corrupted by impulse. Simulation results show that the proposed filter can achieve a high smoothing efficiency and at the same time well preserves edges and details.