Preface-This book began as a rather conservative revision of my earlier textbook, Techniques for Image Processing and Classification in Remote Sensing.
Like many “limited” endeavors, however, it soon
grew to be a much larger project! When it became clear that simply a revision would not suffice, I
gave considerable thought on a new way to present the subject of image processing in the context of
remote sensing.
After much mental wandering about, it became clear that there was a unifying
theme through many of the image processing methods used in remote sensing, namely that they are
based, directly or indirectly, on models of physical processes. In some cases there is a direct dependence, for example, on physical models that describe orbital geometry or the reflectance of radiation. In other cases, the dependence is indirect. For example, the common assumption of data
similarity implies that neighboring pixels in the space or spectral domains are likely to have similar
values. The origin of this similarity is in the physical processes leading up to the acquisition of the
data, and in the acquisition itself. In nearly all cases, the motivation and rationale for remote sensing image processing algorithms can be traced to an assumption of one or more such models. Thus,
I settled on this viewpoint for the book.
It was obvious from the beginning that the book should be an entirely digital production. The computer tools currently available for desktop publishing easily support this, and given the subject matter, seem almost obligatory. Therefore, extensive use is made of computer-generated graphics and image processing. Nearly all figures are entirely new and produced especially for this edition. Three-dimensional graphing programs were used to visualize multidimensional data, and an assortment of desktop image processing programs was used to produce the images.
These include in particular, IPT, a development version of the MacSADIE image processing software from my laboratory, and Multi Spec, a multi-spectral classification program from David Landgrebe’s laboratory at Purdue University. To enhance the use of the book in the classroom, exercises are included for each chapter. They range from conceptual, “gedanken” experiments to mathematical derivations.
The exercises are intended to promote an understanding of the material presented in the chapter. Extensive bibliographies of many of the subjects covered are provided in the form of tables to conserve space and provide a compact source for further information. In the references, I’ve emphasized archival journal papers, because they are generally easiest for the reader to acquire.
Chapter 1 provides an overview of remote-sensing science and technology as of 1996.
The basic parameters for optical remote sensing are established here, and the main types of scanning sensors are described. In Chapter 2, the most important optical radiation processes in remote-sensing are described mathematically. These include solar radiation, atmospheric scattering, absorption and transmission, and surface reflectance. The wavelength region from 400nm wavelength to the thermal infrared is analyzed. Sensor models for radiometric and spatial response are explained in Chapter 3. Satellite imaging geometry is also included because of its importance for image rectification and geocoding and for extraction of elevation information from stereo images.
In Chapter 4, data models provide a transition between the physical models of Chapters 2 and 3 and the image processing methods of later chapters. Spectral and spatial statistical models for remote sensing data are described. A series of imaging simulations illustrate and explain the influence of the sensor’s characteristics on the data acquired by remote-sensing systems. Chapter 5 begins the discussion of image processing methods and covers spectral transforms, including various vegetation indicies, principal components and contrast enhancement. Chapter 6 includes convolution and Fourier filtering, multiresolution image pyramids and scale-space techniques such as wavelets.
The latter types of image analyses appear to have considerable potential for efficient and effective spatial information extraction. The concept of image spatial decomposition into two or more components is used here to provide a link among the different spatial transforms. In Chapter 7, several examples of the use of image processing for image radiometric and geometric calibration are given. The importance of image calibration for high spectral resolution imagery (“hyperspectral”) data is also discussed. The topic of multiimage fusion is addressed in Chapter 8, with reference to the spatial decomposition concept of Chapter 6.
The various approaches are explained and illustrated with Landsat TM multispectral and SPOT panchromatic image fusion. An image pyramid-based scheme for digital elevation model (DEM) extraction from a stereo image pair is also described in detail. Chapter 9 is devoted to thematic classification of remote-sensing images, including the traditional statistical approaches and newer neural network and fuzzy classification methods. Techniques specifically developed for hyperspectral imagery are also described.
It was obvious from the beginning that the book should be an entirely digital production. The computer tools currently available for desktop publishing easily support this, and given the subject matter, seem almost obligatory. Therefore, extensive use is made of computer-generated graphics and image processing. Nearly all figures are entirely new and produced especially for this edition. Three-dimensional graphing programs were used to visualize multidimensional data, and an assortment of desktop image processing programs was used to produce the images.
These include in particular, IPT, a development version of the MacSADIE image processing software from my laboratory, and Multi Spec, a multi-spectral classification program from David Landgrebe’s laboratory at Purdue University. To enhance the use of the book in the classroom, exercises are included for each chapter. They range from conceptual, “gedanken” experiments to mathematical derivations.
The exercises are intended to promote an understanding of the material presented in the chapter. Extensive bibliographies of many of the subjects covered are provided in the form of tables to conserve space and provide a compact source for further information. In the references, I’ve emphasized archival journal papers, because they are generally easiest for the reader to acquire.
Chapter 1 provides an overview of remote-sensing science and technology as of 1996.
The basic parameters for optical remote sensing are established here, and the main types of scanning sensors are described. In Chapter 2, the most important optical radiation processes in remote-sensing are described mathematically. These include solar radiation, atmospheric scattering, absorption and transmission, and surface reflectance. The wavelength region from 400nm wavelength to the thermal infrared is analyzed. Sensor models for radiometric and spatial response are explained in Chapter 3. Satellite imaging geometry is also included because of its importance for image rectification and geocoding and for extraction of elevation information from stereo images.
In Chapter 4, data models provide a transition between the physical models of Chapters 2 and 3 and the image processing methods of later chapters. Spectral and spatial statistical models for remote sensing data are described. A series of imaging simulations illustrate and explain the influence of the sensor’s characteristics on the data acquired by remote-sensing systems. Chapter 5 begins the discussion of image processing methods and covers spectral transforms, including various vegetation indicies, principal components and contrast enhancement. Chapter 6 includes convolution and Fourier filtering, multiresolution image pyramids and scale-space techniques such as wavelets.
The latter types of image analyses appear to have considerable potential for efficient and effective spatial information extraction. The concept of image spatial decomposition into two or more components is used here to provide a link among the different spatial transforms. In Chapter 7, several examples of the use of image processing for image radiometric and geometric calibration are given. The importance of image calibration for high spectral resolution imagery (“hyperspectral”) data is also discussed. The topic of multiimage fusion is addressed in Chapter 8, with reference to the spatial decomposition concept of Chapter 6.
The various approaches are explained and illustrated with Landsat TM multispectral and SPOT panchromatic image fusion. An image pyramid-based scheme for digital elevation model (DEM) extraction from a stereo image pair is also described in detail. Chapter 9 is devoted to thematic classification of remote-sensing images, including the traditional statistical approaches and newer neural network and fuzzy classification methods. Techniques specifically developed for hyperspectral imagery are also described.
Some topics that one usually finds in a remote sensing image processing book, such as classification map error analysis, were deliberately excluded. This was done not only for space reasons, but
also because I felt they departed too far from the main theme of the relation of image processing
methods to remote sensing physical models. Likewise, classification methods such as rule-based
systems that rely on higher level abstractions of the data, although effective and promising in many
cases, are not described. I also view Geographic Information Systems (GIS) as being outside the
scope of this work.
Download Link
Posts
No comments:
Post a Comment