English  |  正體中文  |  简体中文  |  Items with full text/Total items : 90453/105672 (86%)
Visitors : 11995917      Online Users : 370
RC Version 6.0 © Powered By DSPACE, MIT. Enhanced by NTU Library IR team.
Scope Tips:
  • please add "double quotation mark" for query phrases to get precise results
  • please goto advance search for comprehansive author search
  • Adv. Search
    HomeLoginUploadHelpAboutAdminister Goto mobile version
    ASIA unversity > 資訊學院 > 資訊傳播學系 > 博碩士論文 >  Item 310904400/112479

    Please use this identifier to cite or link to this item: http://asiair.asia.edu.tw/ir/handle/310904400/112479

    Title: 使用卷積類神經網路移除數位影像中的椒鹽雜訊
    Removal of Salt-and-Pepper Noise Using Convolutional Neural Network
    Authors: 陳睿瀚
    Contributors: 資訊傳播學系
    Keywords: 椒鹽雜訊;影像濾波;影像修復;卷積類神經網路;人工智慧
    Salt and pepper;Image filtering;Image restoration;Convolutional neural network;Artificial wisdom
    Date: 2019
    Issue Date: 2019-11-11 11:52:51 (UTC+8)
    Publisher: 亞洲大學
    Abstract: 在科技普及的現代,人人隨身都有具有拍攝功能的產品,但拍攝下來的影像可能在拍攝或傳送途中受到脈衝雜訊干擾,進而破壞拍攝影片的清析度。而現代對影像品質的需求也越來越高,所以影像的修復、增強能力也受到各界的重視,例如:醫學治療時,需要觀看人體X光的影像,但是常常會受到雜訊的干擾而導致觀看不易或是產生錯誤判讀,使用高效率的影像修復則可以解決以上的問題。為了使電腦能更準確的分析影像內容、機器能更精準的偵測影像中的物件,如何有效移除影像中的雜訊對現代科技極為重要。本文使用卷積類神經網路(Convolutional Neural Network, CNN)進行雜訊的移除,首先利用滑動視窗(sliding window)分析每一個雜訊像素,並且與原始圖片進行比對,找出最接近之鄰域像素(pixel),並將視窗加大擷取成微圖片,全部擷取後的微圖片便成為卷積類神經網路特徵的數據庫。最後使用卷積類神經網路進行微圖片的特徵學習,利用訓練後的卷積類神經網路,找出受雜訊干擾像素的臨域像素中最接近之像素做為替代,完成雜訊像素之修復。實驗結果顯示本文提出的卷積類神經網路確實可以有效的移除受雜訊干擾影像中的椒鹽雜訊,修復影像能看清輪廓、甚至影像的細節也能有效修復。
    A captured image may be interfered with by impulse noise during acquisition or transmission, which will deteriorate the quality of the image. The demand of image quality is also increasing, so the enhancement of corrupted images is important, such as the X-ray and other medical images. This thesis presents a new method based on the Convolutional Neural Network (CNN) for noise removal. First, each noise pixel is analyzed by using a sliding window, and compared with the original image. The nearest neighbor pixel is selected to be the restored pixel. A larger window is employed to generate training features for the CNN, enabling a greater quantity of noise-free pixels to be adopted. In turn, the CNN is used to learn the mapping relationship between the noisy and noise-free images. The restored pixel is obtained by the selection of the trained CNN from the neighboring noise-free pixels. The experimental results show that the proposed CNN can effectively remove the salt and pepper noise in noisy images with various noise densities.
    Appears in Collections:[資訊傳播學系] 博碩士論文

    Files in This Item:

    File Description SizeFormat

    All items in ASIAIR are protected by copyright, with all rights reserved.

    DSpace Software Copyright © 2002-2004  MIT &  Hewlett-Packard  /   Enhanced by   NTU Library IR team Copyright ©   - Feedback