基于MATLAB的圖像融合算法

論文編號:TX244  論文字數(shù):19816,頁數(shù):34

摘要
 
 圖像融合能夠?qū)⒉煌�類型傳感器獲取的同一對象的圖像數(shù)據(jù)進行空間配準。并且采用一定的算法將各圖像數(shù)據(jù)所含的信息優(yōu)勢或互補性有機的結(jié)合起來產(chǎn)生新的圖像數(shù)據(jù)。這種新數(shù)據(jù)具有描述所研究對象的較優(yōu)化的信息表征，同單一信息源相比，能減少或抑制對被感知對象或環(huán)境解釋中可能存在的多義性、不完全性、不確定性和誤差，最大限度的利用各種信息源提供的信息。
    圖像融合分為像素級、特征級、決策級三個層次，其中像素級圖像融合能夠提供其它層次上的融合處理所不具有的更豐富、更精確、更可靠的細節(jié)信息，有利于圖像的進一步分析、處理和理解，它在整個圖像融合技術(shù)中是最為復(fù)雜、實施難度最大的融合處理技術(shù)。本文的研究工作是圍繞像素級圖像融合展開的，針對像素級圖像融合技術(shù)中需要解決的關(guān)鍵問題，研究了多種像素級圖像融合方法。
 本論文的主要的研究內(nèi)容有：
 首先介紹了圖像信息融合的概念、優(yōu)勢、發(fā)展歷史和應(yīng)用領(lǐng)域，并介紹了圖像融合的三個層次及常用的空域圖像融合方法,空域融合方法有像素平均法、像素最大最小法、像素加權(quán)平均法，頻域融合方法包括圖像的多尺度分解、圖像的小波變換、基于小波變換的圖像融合方法。圖像的預(yù)處理有濾波（鄰域平均濾波法、中值濾波法）和圖像配準。最后，對于圖像融合系統(tǒng)來說，融合圖像質(zhì)量的評價顯得特別重要，本文探討了圖像融合質(zhì)量的評價問題，總結(jié)了融合效果的主、客觀評價標準，作為本課題性能分析的判斷標準。
 

關(guān)鍵詞：圖像配準；圖像融合；空域融合法；小波變換；評價標準
 
 

MATLAB-based image fusion algorithm

Abstract
 
 The same object gotten from different sensors can be registered spatially by mage fusion. The information advantages or the complements of all the image data can be combined to produce new image data using some fusion algorithms. The new data can describe the optimized information of the studied object. Compared with single information source, the new data can reduce or restrain the ambiguity, the incompleteness, the uncertainty and the error, which may appears in the explanation of the studied object or the environment, and make full use of the information provided by all kinds of resources.
 Image fusion consists of such three levels as the Pixel level，the feature level and the decision level，among which the Pixel level image fusion can Provide more
abundant, accurate and reliable detailed information that doesn’t exist on the other levels and It is the most complicated in the whole image fusion techniques and also is the most difficult to implement in the fusion Processing techniques. this dissertation Progresses mainly around the Pixel level image fusion and proposes a variety of Pixel level image fusion techniques according to the key Problems in the Pixel level image fusion techniques.
 The major research and findings are as follows:
 First we introduce the concepts， advantages，developments and applications. Three levels of image fusion and image fusion techniques in common use are also reviewed. Airspace Image Fusion such as simple fusion method (pixel average, maximal or minimal pixel selection), Frequency-domain image fusion methods include the multiresolution image fusion techniques based on multi-scale pyramid decomposition, and the image fusion method based on wavelet transform Image Pre-processing like Filter processing (neighborhood average filter, median filtering method) and Image Registration. in the end, evaluation for fusion image is vital to fusion system. This dissertation probes into the image fusion quality assessment and deduces a set of indexes as the criteria to analyze the performances of this discussion.

Keywords: Image Registration；Image Fusion；Airspace integration method；Wavelet Transform；Evaluation criteria

目錄
第一章 緒論 6
1.1 圖像融合的概念 6
1.2圖像融合的主要研究內(nèi)容 7
1.2.1 圖像融合的層次 7
1.2.2 圖像融合算法的發(fā)展 9
1.2.3圖像融合的步驟 9
1.3 圖像融合技術(shù)的發(fā)展現(xiàn)狀 10
1.4 本文的研究工作 10
第二章 圖像預(yù)處理 11
2.1 圖像的校正 11
2.2 圖像濾波技術(shù) 11
2.2.1 鄰域平均法 12
2.2.2 中值濾波 12
2.3 圖像配準 13
2.3.1 圖像配準概述 13
2.3.2 手動圖像配準 14
2.3.3 基于圖像特征的匹配算法 15
第三章 圖像融合 16
3.1 加權(quán)平均融合法 16
3.2 像素灰度值選大/小融合方法 16
3.3 主分量融合法 17
3.4 IHS變換法 19
3.5 小波變換融合法 20
3.5.1 小波的定義及特點 20
3.5.2 基于小波變換的圖像融合方法原理 25
3.5.3 圖像融合規(guī)則及融合因子 26
第四章 圖像融合效果評價 27
4.1 主觀評價 27
4.2 客觀評價 27
4.2.1 基于光譜特征的評價 27
4.2.2 基于信息量的評價 28
4.2.3 基于統(tǒng)計特性的評價 29
4.2.4 基于信噪比的評價 30
總結(jié)與展望 31
謝辭 32
參考文獻 32