最小熵解回旋滤波器应用於模糊影像还原之研究OntheStudyofthe.PDFVIP

Journal of China Institute of Technology Vol.29-2003.12 最小熵解迴旋濾波器應用 於模糊影像還原之研究 On the Study of the Blurred Image Restoration Using Minimum Entropy Deconvolution 劉建宏 宋志元 Chien-Hung Liu Chih-Yuan Sung 中華技術 學院電子系講師 中華技術學 院電子系助理教授 Department of Electronic Engineering 摘 要 一張模糊的影像(Blurred image) ，若沒有足夠之事前資訊(a priori knowledge) 而欲重建原影像是相當困難的。我們提出一個二維(2-D)之反覆迭代式最小熵解迴 旋(Minimum Entropy Deconvolution)濾波器，來針對模糊之影像直接作影像還原 (Image Restoration) 以重建影像。此去模糊的過程實際為一解迴旋運算的問題，又 因是在原影像特性未知之情況下作反 運算，故亦為一盲目解迴旋 (Blind deconvolution)的過程 。綜合言之，本法能對退化或模糊影像作去模糊之處理且演 算法實施於時域中較一般頻域方法，計算快速、方便易行、效果精確。最後本文 將提出二值或灰階影像的電腦模擬結果，來示範驗證本文所提之 2-D 最小熵解迴 旋影像還原濾波器之高性能。 關鍵詞 : 模糊影像、最小熵解迴旋、影像還原、盲目解迴旋 Abstract It is difficult to restore the blurred image when there is not enough a priori knowledge. We present a recursive type minimum entropy deconvolution algorithm to reconstruct the blurred image with image restoration process directly. Not only the process to deblur image is an image deconvolution problem, but also the characteristic of the image is unknown, so our image restoration work is a blind deconvolution process. In our plan, we use the minimum entropy deconvolution algorithm to design the 2D image restoration filter. The MED algorithm that is executed on time domain - 29 - 最小熵解迴旋濾波器應用於模糊影像還原之研究 outperform the other method carried out on frequency domain and the process of the image restoration can compute quickly, conveniently, efficiently. At last we demonstrated the good resul