多頻譜衛星影像融合與紅外線影像合成; Multi-spectral Image Fusion and Infrared Image Synthesis

NCUIR > College of Electrical Engineering & Computer Science > Graduate Institute of Computer Science and Information Engineering > Electronic Thesis & Dissertation > Item 987654321/8463

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/8463

Title:	多頻譜衛星影像融合與紅外線影像合成;Multi-spectral Image Fusion and Infrared Image Synthesis
Authors:	陳奕霖;Yi-Ling Chen
Contributors:	資訊工程研究所
Keywords:	紅外線影像合成;影像融合;多頻譜衛星影像;Infrared Image Synthesize;Image Fusion;Muti-spectral Image
Date:	2000-07-06
Issue Date:	2009-09-22 11:27:38 (UTC+8)
Publisher:	國立中央大學圖書館
Abstract:	在本論文的研究中，我們提出一個以多頻譜遙測影像來合成紅外線影像景觀的方法。這個方法共分成四個步驟：影像融合、地形效應校正、影像複合像點分類、及紅外線頻譜合成。在第一步驟中，我們以淩波轉換 (wavelet transform) 為基礎的主成分分析法 (principal component analysis)來融合高解析度單一頻譜影像及低解析度多頻譜影像成為高解析度的多頻譜影像。在第二步驟中，我們藉由迴歸交點法 (regression intersection method) 與方向餘弦超球體轉換法 (hyperspherical direction cosine transformation) 來消除融合後衛星影像中的地形效應與感應器的偏差。在第三步驟中，我們以線性複合模式 (linear mixing model) 為基礎，利用線性頻譜分析技術 (linear spectral unmixing) 來分解影像中的複合像點成份，並產生各材質的分量影像 (fraction images)。在第四步驟中，我們以 ASTER 頻譜資料庫所提供的材質反應資料結合之前所獲得的材質分量影像，利用線性複合模式合成出特定波長的紅外線影像。在本研究中，我們發現以淩波轉換為基礎的主成份分析法比傳統融合方法在融合時對多頻譜影像能保存更多的頻譜資訊；另外融合後的多頻譜影像在做地形效應修正時，能獲得較好的效果，因而能合成出更正確紅外線影像。最後，我們把此特定波長的紅外線影像貼圖到對應的地形模型 (terrain model) 上以產生 3-D 紅外線影像景觀。 In this paper, we propose a framework for synthesizing specific infrared images from multi-spectral remote-sensing images. The framework is divided into four stages: image fusion, topographic normalization, mixed-pixel analysis, and spectrum synthesis. Firstly, an image fusion method, wavelet-based principal component analysis, is proposed to fuse low-resolution multi-spectral images (Landsat TM) and high-resolution single band image (SPOT PAN) to generate high-resolution multi-spectral images. Secondly, a regression intersection method (RIM) and a hyperspheric direction cosine (HSDC) transformation are adopted to eliminate atmospheric effects, sensor bias, and topographic effects in generated high-resolution multi-spectral images. Thirdly, a linear spectral unmixing (LSU) method is employed for the mixed-pixel classification on the refined images to produce the material fraction images. Lastly, we incorporate the fraction images with reflectance tables of ASTER spectral library to synthesize specific infrared images. The proposed image fusion method really produces the high-resolution multi-spectral images preserving most original spectral information. The fusion method also improves the effect of topographic normalization and hence enhances the quality of the final synthesis images. The synthesis images will be mapped onto the corresponding terrain models to generate realistic infrared images for flight and tactical simulation.
Appears in Collections:	[Graduate Institute of Computer Science and Information Engineering] Electronic Thesis & Dissertation

Files in This Item:

File	Size	Format

社群 sharing

Loading...