English  |  正體中文  |  简体中文  |  Items with full text/Total items : 27533/39387
Visitors : 2541639      Online Users : 26
RC Version 4.0 © Powered By DSPACE, MIT. Enhanced by NTU Library IR team.
Scope Adv. Search
LoginUploadHelpAboutAdminister

Please use this identifier to cite or link to this item: http://ntour.ntou.edu.tw:8080/ir/handle/987654321/49402

Title: 基於影像質地與機器學習的模糊影像識別
Blur Detecting Based on Image Texture and Machine Learning
Authors: Chiu, Kuan-Chih
邱冠智
Contributors: NTOU:Department of Computer Science and Engineering
國立臺灣海洋大學:資訊工程學系
Keywords: 無影像參考評估;機器學習;影像質地;心理量測函數;貝索夫平滑空間
Blind quality assessment;machine learning;image texture;psychometric function;Besov spaces
Date: 2017
Issue Date: 2018-08-22T06:57:23Z
Abstract: 本論文討論如何藉由影像質地與機器學習的方法分類模糊影像畫質。許多心理學的研究顯示,人類感知的刺激物變化量會與初始刺激物強度成正比。因此人類視覺對模糊的偵測會受到影像內容質地的影響。本論文提出了一個新的客觀量測指標,它能夠測量固有的平滑紋理也可以預測模糊失真所引起的差異。另外我們還開發了一個模糊影像畫質量測評估模型,在學習了再模糊影像的Besov常模下降率後,能夠將影像特徵分類到五個等級。我們接著利用訓練影像和心理測量函數來導出模型中的最小可視模糊門檻(JNB)。然後將測試影像所分類的模糊等級與導出的JNB進行比較,以確定模糊程度是否達到人類感知門檻。在三個影像資料庫上實驗的結果驗證了模型的有效性和穩定性。
This thesis investigates how to blindly classify the visual quality of a blurred image by machine learning approach. Substantial psychological research shows that human perceive change in stimuli is proportional to the initial stimuli. Hence, human vision detects blurriness might be influenced by the texture of image contents. This paper presents a new objective metric designed as both measuring the inherent smooth texture and predicting the difference induced by blur distortion. In addition, we develop a blind blurred image quality assessment model, which learns the descending rate of Besov norm of the re-blurred images to train the image features into five grades. We also utilize the training images and psychometric function to derive the just noticeable blur (JNB) threshold in the model. Then, the graded blurred level of the test image is compared with the derived JNB label to determine whether the blurriness attain to the human-perceived level or not. Experimental results obtained on three simulated databases verify the model’s effectiveness and robustness.
URI: http://ethesys.lib.ntou.edu.tw/cgi-bin/gs32/gsweb.cgi?o=dstdcdr&s=G0010457037.id
http://ntour.ntou.edu.tw:8080/ir/handle/987654321/49402
Appears in Collections:[資訊工程學系] 博碩士論文

Files in This Item:

File Description SizeFormat
index.html0KbHTML13View/Open


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

 


著作權政策宣告: 本網站之內容為國立臺灣海洋大學所收錄之機構典藏,無償提供學術研究與公眾教育等公益性使用,請合理使用本網站之內容,以尊重著作權人之權益。
網站維護: 海大圖資處 圖書系統組
DSpace Software Copyright © 2002-2004  MIT &  Hewlett-Packard  /   Enhanced by   NTU Library IR team Copyright ©   - Feedback