含脈衝、高斯及混合式之影像雜訊清除法

Abstract

在很多實際的系統雜訊是不可避免的，雜訊的存在將使得系統的輸出大大的受到影響，因此雜訊的移除是必須且重要的。本文著重在影像雜訊的去除，這裡所討論的雜訊主要是脈衝與高斯分佈或兼含兩者之雜訊。以往的濾波器是往往是用來處理某種單一類型雜訊，但是通常雜訊是以混合的方式存在，也就是同時含有上述兩種的雜訊，因此當混合式雜訊存在時，這時先前設計應付一種類型雜訊的濾波器將無法達到所需的要求。不同以往的濾波器設計，我們提出一套能同時處理上述兩種雜訊的方法，首先我們先偵測影像像素是否為脈衝雜訊所污染，如果是才以新型之中值濾波處理，這樣能避免誤處理原始像素，接著處理過後的影像再由基於模摸規則(fuzzy rule-based)的濾波器去除剩下的主要含高斯雜訊部分，此濾波器是一種權重平均的輸出，設計是基於三個影響系統輸出的參數:像素間灰階值的差距，像素間的距離、方向，以及處理區域之像素間的變異。經由最小的平均誤差平方(LMS)演算法，我們可以得到此濾波的歸屬(權重)函數。最後由實驗結果我們可以證明我們所提出的發法在影像雜訊移除的效果及其強健性。

The noise is commonly observed in many practical system, and the system’s output is greatly affected by the existence of the noise. Therefore, it is necessary and important to remove the noise. The thesis introduces several schemes for the image noise removal: the noise discussed here contains the impulse and Gaussian noise distribution, and both. First filters were designed to process some single noise type. Observe that the noise usually appears in mixed mode in a practical system. When the noise appears in mixed mode, the filters for single noise type mentioned above can not facilitate an effective filter action. Different from the design of filters above, we propose a method that can remove the noise mixed with impulse and Gaussian noise from an image. First, we detect the image pixel to check whether the image pixel is impulse noise contaminated or not. If the pixel is corrupted by the impulse noise, a new median filtering is applied to replace the noisy pixel, which will prevent to distract the original image pixel. Then the processed image is filtered by the use of the fuzzy rule-based filter to remove the remaining noise, whose noise contains Gaussian type mostly. The fuzzy rule-based filter’s output is a weighted-average sum. It is based on the three parameters: the gray level difference between pixels, the spatial distance and direction between pixels, and the variance in the local window. Using the LMS algorithm we can determine the membership function for the filter. From the simulation results, our proposal scheme has demonstrated the effectiveness and robustness, in comparison with other filters in image noise removal.