Study on Denoising and Unmixing of Hyperspectral Images Exploiting Spectral Linearity

フォーマット:

学位論文

責任表示:

ミヤ　リズキニヤ

言語:

英語

著者名:

ミヤ　リズキニヤ  

バージョン:

ETD

概要:

This study aims to generalize color line to M-dimensional spectral line feature (M>3) and introduce methods for denoising and unmixing of hyperspectral images based on the spectral linearity.For denoising, we propose a local spectral component decomposition method based on the spectral line. We first calculate the spectral line of an M-channel image, then using the line, we decompose the image into three components: a single M-channel image and two … gray-scale images. By virtue of the decomposition, the noise is concentrated on the two images, thus the algorithm needs to denoise only two grayscale images, regardless of the number of channels. For unmixing, we propose an algorithm that exploits the low-rank local abundance by applying the unclear norm to the abundance matrix for local regions of spatial and abundance domains. In optimization problem, the local abundance regularizer is collaborated with the L2, 1 norm and the total variation. 続きを見る

URL:

http://id.nii.ac.jp/1077/00000567/