VI Preface
segmentation, change detection, endmember extraction for spectral unmixing,
and feature extraction. Diverse kinds of ArtificialNeuralNetworks, Mathemati-
cal Morphology and Markov Random Fields are applied to these tasks. The kind
of images are mostly multispectral-hyperspectral images, with some examples of
processing Synthetic Aperture Radar images, whose appeal lies in its insensitiv-
ity to atmospheric conditions. Two specific applications stand out. One is forest
fire detection and prevention, the other is quality inspection using hyperspectral
images.
Chapter 1 provides a review of current Imaging Spectrometer designs. They
focus on the spectral unit. Three main classes are identified in the literature:
filtering, dispersive and interferometric. The ones in the first class only transmit
a narrow spectral band to each detector pixel. In dispersive imaging spectrome-
ters the directions of light propagation change by diffraction, material dispersion
or both as a continuous function of wavelength. Interferometric imaging spec-
trometers divide a light beam into two, delay them and recombine them in the
image plane. The spectral information is then obtained by performing a Fourier
transform.
Chapter 2 reviews the state of the art in the application of Data Compression
techniques to Remote Sensing images, specially in the case of Hyperspectral im-
ages. Lossless, Near-Lossless and Lossy compression techniques are reviewed and
evaluated on well known benchmark images. The chapter includes summaries of
pertinent materials such as Wavelet Transform, KLT, Coding and Quantization
algorithms, compression quality measures, etc.
Chapter 3 formulates the watermarking of digital images as a multi-objective
optimization problem and proposes a Genetic Algorithm to solve it. The two
conflicting objectives are the robustness of the watermark against manipulations
(attacks) of the watermarked image and the low distortion of the watermarked
image. Watermarking is proposed as adding the image mark DCT coefficients to
some of the watermarked image DCT coefficients. In the case of hyperspectral
images the DCT is performed independently on each band image. The careful
definition of the robustness and distortion fitness functions to avoid flat fitness
landscapes and to obtain fast fitness evaluations is described.
Chapter 4 refers the current efforts at the European Space Agency to provide
Service Support Environments (SSE) that: (1) Simplify the access to multiple
sources of Earth Observation (EO) data. (2) Facilitate the extraction of infor-
mation from EO data. (3) Reduce the barrier for the definition and prototyping
of EO Services. The objective of the chapter is to provide an overview of the
systems which can be put in place to support various kinds of user needs and
to show how they relate each other, as well as how they relate with higher level
user requirements. The chapter reviews several apparently un-related research
topics: service oriented architecture, service publishing, service orchestration,
knowledge based information mining, information and feature extraction, and
content based information retrieval. The authors stress their relative roles and
integration into a global web-based SSE for EO data.
我的内容管理
收起
我的收益 登录查看自己的收益
我的积分
登录查看自己的积分
我的C币
登录后查看C币余额
我的收藏 
我的下载 
下载帮助 
评论2