Abstract

VEGETATION data are used mainly for multiple date applications. This implies high-accuracy geometric requirements, especially for multitemporal registration. To comply with these requirements, VEGETATION1 image location is improved by a systematic use of a database of ground control points (GCPs), based on VEGETATION1 chips extracted worldwide. Due to this systematic processing, VEGETATION1 has shown a very accurate and stable geometric performance, ever since its launch onboard SPOT-4 in March of 1998. VEGETATION2, launched onboard SPOT-5 in May of 2002, now complements VEGETATION1 for operational production. It was, therefore, essential to ensure geometric continuity between both sensors. Onboard SPOT-5, a stellar sensor provides high-accuracy satellite attitude, thus enabling high accuracy in the absolute image location. Consequently, it is not necessary to improve VEGETATION2 image geolocation using GCPs. However, it is necessary to perform a fine geometric calibration of VEGETATION2 cameras. This calibration is performed during the commissioning phase using GCPs from the VEGETATION1 database. The monitoring of VEGETATION2 has shown that due to this calibration its geometric performance became at least as accurate and stable as that of VEGETATION1, and that geometric continuity between both sensors was guaranteed. This chapter describes the VEGETATION1 and VEGETATION2 sensors, the method used to build the VEGETATION database of GCPs, as well as the operational method used to register VEGETATION data.

Introduction

The SPOT-4 (Satellite Pour l'Observation de la Terre 4) satellite, launched in March 1998, provides data continuity for high-resolution SPOT data users through its High Resolution in the Visible and Infra-Red (HRVIR) instruments.

Abstract

We consider various algorithmic solutions to image registration based on the alignment of a set of feature points. We present a number of enhancements to a branch-and-bound algorithm introduced by Mount, Netanyahu, and Le Moigne (Pattern Recognition, Vol. 32, 1999, pp. 17–38), which presented a registration algorithm based on the partial Hausdorff distance. Our enhancements include a new distance measure, the discrete Gaussian mismatch, and a number of improvements and extensions to the above search algorithm. Both distance measures are robust to the presence of outliers, that is, data points from either set that do not match any point of the other set. We present experimental studies, which show that the new distance measure considered can provide significant improvements over the partial Hausdorff distance in instances where the number of outliers is not known in advance. These experiments also show that our other algorithmic improvements can offer tangible improvements. We demonstrate the algorithm's efficacy by considering images involving different sensors and different spectral bands, both in a traditional framework and in a multiresolution framework.

Introduction

Image registration involves the alignment of two images, called the reference image and the input image, taken of the same scene. The objective is to determine the transformation from some given geometric group that most nearly aligns the input image with the reference image.

Abstract

This chapter provides an overview of the use of mutual information (MI) as a similarity measure for the registration of multisensor remote sensing images. MI has been known for some time to be effective for the registration of monomodal, as well as multimodal images in medical applications. However, its use in remote sensing applications has only been explored more recently. Like correlation, MI-based registration is an area-based method. It does not require any preprocessing, which allows the registration to be fully automated. The MI approach is based on principles of information theory. Specifically, it provides a measure of the amount of information that one variable contains about the other. In registration, we are concerned with maximizing the dependency of a pair of images. In this context, we discuss the computation of mutual information and various key issues concerning its evaluation and implementation. These issues include the estimation of the probability density function and computation of the joint histogram, normalization of MI, and use of different types of interpolation, search and optimization techniques for finding the parameters of the registration transformation (including multiresolution approaches).

Introduction

In this chapter, we discuss the application of mutual information (MI) as a similarity metric for cross-registering images produced by different imaging sensors. These images may be from different sources taken at different times and may be produced at different spectral frequencies and/or at different spatial resolutions.

Abstract

Performance bounds can be used as a performance benchmark for any image registration approach. These bounds provide insights into the accuracy limits that a registration algorithm can achieve from a statistical point of view, that is, they indicate the best achievable performance of image registration algorithms. In this chapter, we present the Cramér-Rao lower bounds (CRLBs) for a wide variety of transformation models, including translation, rotation, rigid-body, and affine transformations. Illustrative examples are presented to examine the performance of the registration algorithms with respect to the corresponding bounds.

Introduction

Image registration is a crucial step in all image analysis tasks in which the final information is obtained from the combination of various data sources, as in image fusion, change detection, multichannel image restoration, and object recognition. See, for example, Brown (1992) and Zitová and Flusser (2003). The accuracy of image registration affects the performance of image fusion or change detection in applications involving multiple imaging sensors. For example, the effect of registration errors on the accuracy of change detection has been investigated by Townshend et al. (1992), Dai and Khorram (1998), and Sundaresan et al. (2007). An accurate and robust image registration algorithm is, therefore, highly desirable.

The purpose of image registration is to find the transformation parameters, so that the two given images that represent the same scene are aligned. There are many factors that might affect the performance of registration algorithms.

In recent years, image registration has become extremely important in remote sensing applications. Image registration refers to the fundamental task in image processing to match two or more pictures which have been taken of the same object or scene, for example, at different times, from different sensors, or from different viewpoints.

The main reason for the increased significance of image registration in remote sensing is that remote sensing is currently moving towards operational use in many important applications, both at social and scientific levels. These applications include, for example, the management of natural disasters, assessment of climate changes, management of natural resources, and the preservation of the environment; all of which involve the monitoring of the Earth's surface over time. Furthermore, there is an increasing availability of images with different characteristics, thanks to shorter revisiting times of satellites, increased flexibility of use (different acquisition modalities) and the evolution of sensor technologies. Therefore, a growing need emerges to simultaneously process different data, that is, remote sensing images, for information extraction and data fusion. This includes the comparison (integration or fusion) of newly acquired images with previous images taken with different sensors or with different acquisition modalities or geometric configurations – or with cartographic data. The remote images can, therefore, be multitemporal (taken at different dates), multisource (derived from multiple sensors), multimode (obtained with different acquisition modalities), or stereo-images (taken from different viewpoints).

The publication of this Atlas, in accordance with the desires of Professor Barnard, was assured by a grant made by the Carnegie Institution of Washington in 1907. The long delay in its appearance calls for an explanation. Mr. Barnard was in the throes of preparing for publication a volume of his pioneer celestial photographs made at the Lick Observatory in the years 1889-1895. He had difficulty in satisfying himself that any mode of reproduction could adequately depict the qualities of the original photographs.

That handsome work, which forms Volume XI of the Publications of the Lick Observatory, was not printed until 1913. It was natural and proper that the preparation of the present volume should have been delayed while the task of completing the earlier volume was in hand. The mode of reproduction to be adopted for the splendid photographs of this Atlas had not been selected at the time the original grant was made, and consequently considerable investigation and experiment were necessary in reaching a decision on this important matter. The attempts made with the photogravure and other processes did not give the assurance of uniformity that was desired, and finally the author was persuaded that actual photographic prints would be more satisfactory and hardly more expensive than any other available method of reproduction. After this decision had been reached and had been approved by the Carnegie Institution of Washington, Professor Barnard began the task of making the reproducing negatives, and then took upon himself the heavy duty of personally inspecting every print of the 35,700 needed in the issue of an edition of 700 copies.

My principal aim in presenting these photographs has been to give pictures of some of the most interesting portions of the Milky Way in such form that they may be studied for a better understanding of its general structure. They are not intended as star charts. Such photographic charts have already been made by Wolf and Palisa and by Franklin-Adams. They are probably more useful for the identification of individual stars. But these do not give us a true picture of the parts of the sky shown, for there are structures and forms that cannot well be depicted in ordinary charts, and it has seemed to me that some of these are of the utmost importance in the study of the universe at large. These photographs may, therefore, be considered as supplementary to the regular charts in that they show the details of the clouds, nebulosities, etc. In this form, however, it is always difficult to identify the individual small stars. To overcome this difficulty charts have been prepared corresponding to each photograph and giving on the same scale a set of co-ordinates, and all the principal stars and objects of especial interest. The most useful reference stars are numbered, as are the dark objects. These charts and the tables, which give fuller data about the reference stars, will be found in Part II. It is recommended that in studying any photograph the reader should open Part II to the corresponding chart, and then he will have before him the photograph or plate, the author's text descriptive of it, the chart, with its co-ordinates, including most of the stars of the Bonner Durchmusterung, and the table supplementary to the chart.

The positions of Professor Barnard's dark objects are given here for equinox J2000.0. These positions will allow the reader to locate them on contemporary star maps, on photographs of the Milky Way, or visually in the sky. The angular size of these objects, if known, is given in arc minutes, as well as the Plate(s) on which the objects appear in this Atlas. There are 31 dark objects in Barnard's catalogue that do not appear on the 50 plates contained in this edition, although they are still provided in this listing.

As stated in the Addendum, during the editing of the 1927 edition of Barnard's Atlas, Frost and Calvert noted three objects from Barnard's list, Nos. 52, 131a, and 172, had been duplicated. This omission may have been in error and the corrections are presented. All of these omitted regions were found using Barnard's paper in the Astrophysical Journal January, 1919 (49, 1-23), where the positions provided are given in B1875.0 equinox. The Barnard catalogue numbers, positions, angular measurements, and descriptions are taken from this paper. The explanations following Professor Barnard's descriptions are those of this author.

Edward emerson barnard was born on 16 December, 1857, in Nashville, Tennessee. Edward was the second child of Reuben and Elizabeth Jane (Haywood) Barnard. Tragically, his father passed away three months before Edward was born. Edward had an older brother, Charles, born in 1854, but not much is known of Charles.

Their mother moved to Nashville shortly after Reuben died and she tried to provide for her children by fashioning wax flowers. Soon after arriving in Nashville, the Civil War broke out and the small town in central Tennessee became the hub of battles between the North and the South. Edward had only two months of formal schooling and, just prior to his ninth birthday, he took a job at a portrait studio to augment the family income. The studio was owned by John H. Van Stravoren and the first duties of young Edward were to assist with portrait enlargements using a machine named “Jupiter.” The Jupiter enlargement machine required the aperture end to be kept in alignment with the Sun, so as to provide natural sunlight projected onto the portrait frame. Tedious work for a young lad, but it set the stage for Barnard's later work in astrophotography; that is, patience and diligence. Barnard continued to work at the portrait studio for over 16 years. The Van Stavoren studio was sold to Rodney Poole in 1871 and the Jupiter enlargement machine was dismantled. Barnard's duties became that of “sign painter” and later of taking photographs and developing the glass plates.

I WAS twelve years old when I first discovered Edward Emerson Barnard's famous work, A Photographic Atlas of Selected Regions of the Milky Way. In my continued explorations of the life of E. E. Barnard and his work, and through my personal discoveries of the cosmos, I noticed a striking set of parallels between his life and mine that made strong my devotion to his work. We were both born in December exactly 100 years apart: Barnard in 1857 and I in 1957. Barnard came from an impoverished family and grew up in civil unrest and war in America. My beginnings were much the same, and although the civil unrest was in Vietnam, it had permeated into everyday American life. At the age of nine, we both entered the workforce outside the home in an effort to financially assist our families. We both struggled through our education, always keeping a focus on our passion for astronomy. And, like Barnard, I've devoted my life to observational astronomy, with a fanaticism for the dark nebulosity within our Galaxy.

Barnard spent most of his astronomical career photographing the night sky. What he captured in those images provided greater detail than the eye could discern through the telescope. The most interesting regions found in these photographs were dark patches that Barnard called “black holes”; unique objects that are, of course, not the black holes that astronomers refer to today, but masses of dust and gas silhouetted against the brightness of the Milky Way.

This second part of the Atlas has been provided to aid in the convenient use and study of the photographs contained in Part I, for reasons which were stated in notes by Professor Barnard as follows:

When comparing astronomical photographs made with long exposures with star charts I have frequently had much trouble through the want of an approximate position, in identifying stars and other objects on the photographs. Also, very often, the colors of the stars so change their relative intensities that they are not easily recognized on the chart. The photographs in the present work are intended as pictures of the sky and it would have been impossible to mark co-ordinates on them without spoiling their pictorial value. It was therefore decided to make a map, with co-ordinates, corresponding to each photograph and on the same scale. Though this has required much work, the charts assist greatly in the approximate location of any object shown on the photographs. They have been of great service to me in studying the photographs and I believe will be a welcome addition to the Atlas.

The photographs are not all enlarged in the same proportion, and therefore are not uniform in scale. All of the fainter stars shown on the Durchmusterung charts were not put on the diagrams, but it is believed enough of them are given to permit a ready identification of objects in any part of a photograph. Four stars on each photograph, located near the corners, were identified and used for determining its scale and for locating the system of co-ordinate lines. The epoch 1875.0 was adopted and is used throughout this work.

In his article in the Astrophysical Journal for January, 1919 (49, 1-23) Professor Barnard gave a list of 182 dark objects in the sky. For the convenience of the user of this Atlas this catalogue is printed here. Three of the objects in that list have been omitted here, viz., Nos. 52, 131a, and 172, because by inadvertence the same object had been listed twice.

Mr. Barnard had begun a second list, most of the objects for which he had himself selected. Their positions were determined by Miss Calvert. It seemed best to begin the second list with No. 201; accordingly, there are no objects having the numbers from 176 to 200.

Where the space in the column for description was insufficient, a note has been added at the end of the catalogue, and this is indicated by a dagger (†) at the end of the last column.

Each dark object of the list falling within the field of a plate has been sketched in by Miss Calvert, with its number, on the corresponding chart in Part II. Where these objects have been referred to, in the descriptions of the photographs, and on the charts and in the tables, their numbers have been preceded by the letter “B.”