Robert F. Cromp

Support vector machines for hyperspectral remote sensing classification

The Support Vector Machine provides a new way to design classification algorithms which learn from examples (supervised learning) and generalize when applied to new data. We demonstrate its success on a difficult classification problem from hyperspectral remote sensing, where we obtain performances of 96%, and 87% correct for a 4 class problem, and a 16 class problem respectively. These results are somewhat better than other recent result on the same data. A key feature of this classifier is its ability to use high-dimensional data without the usual recourse to a feature selection step to reduce the dimensionality of the data. For this application, this is important, as hyperspectral data consists of several hundred contiguous spectral channels for each exemplar. We provide an introduction to this new approach, and demonstrate its application to classification of an agriculture scene.

An automated parallel image registration technique based on the correlation of wavelet features

With the increasing importance of multiple multiplatform remote sensing missions, fast and automatic integration of digital data from disparate sources has become critical to the success of these endeavors. Our work utilizes maxima of wavelet coefficients to form the basic features of a correlation-based automatic registration algorithm. Our wavelet-based registration algorithm is tested successfully with data from the National Oceanic and Atmospheric Administration (NOAA) Advanced Very High Resolution Radiometer (AVHRR) and the Landsat Thematic Mapper (TM), which differ by translation and/or rotation. By the choice of high-frequency wavelet features, this method is similar to an edge-based correlation method, but by exploiting the multiresolution nature of a wavelet decomposition, our method achieves higher computational speeds for comparable accuracies. This algorithm has been implemented on a single-instruction multiple-data (SIMD) massively parallel computer, the MasPar MP-2, as well as on the CrayT3D, the Cray T3E, and a Beowulf cluster of Pentium workstations.

Analyzing hyperspectral data with independent component analysis

Hyperspectral image sensors provide images with a large number of contiguous spectral channels per pixel and enable information about different materials within a pixel to be obtained. The problem of spectrally unmixing materials may be viewed as a specific case of the blind source separation problem where data consists of mixed signals and the goal is to determine the contribution of each mineral to the mix without prior knowledge of the minerals in the mix. The technique of independent component analysis (ICA) assumes that the spectral components are close to statistically independent and provides an unsupervised method for blind source separation. We introduce contextual ICA in the context of hyperspectral data analysis and apply the method to mineral data from synthetically mixed minerals and real image signatures.

Probabilistic neural network architecture for high-speed classification of remotely sensed imagery

Abstract In this article we discuss a neural network architecture (the probabilistic neural net or the PNN) that, to the best of our knowledge, has not previously been applied to remotely sensed data. The PNN is a supervised nonparametric classification algorithm as opposed to the Gaussian maximum likelihood classifier (GMLC). The PNN works by fitting a Gaussian kernel to each training point. The width of the Gaussian is controlled by a tuning parameter called the window width. If very small widths are used, the method is equivalent to the nearest neighbor method. For large windows, the PNN behaves like the GMLC. The basic implementation of the PNN requires no training time at all. In this respect it is far better than the commonly used backpropagation neural network (BPNN), which can be shown to the O(N6) time for training where N is the dimensionality of the input vector. In addition, the PNN can be implemented in a feed-forward mode in hardware. The disadvantage of the PNN is that it requires all the training data to be stored. Some solutions to this problem are discussed in the article. Finally, we discuss the accuracy of the PNN with respect to the GMLC and the BPNN. The PNN is shown to be better than GMLC and not as good as the BPNN with regard to classification accuracy.

Data mining of multidimensional remotely sensed images

As scientific spatial databases grow at unprecedented rates, new approaches are necessary to enable scientists to efficiently locate data sets pertinent to their needs, One method that shows promise is the augmentation of a metadatabase with information on image content so that users can be heuristically guided to appropriate data sets. This paper motivates a need for building more intelligent databases and examines methods for automatically extracting content from image data, The paper concludes with a discussion of automated discovery aa it pertains to multidimensional remotely sensed images.

Automatic labeling and characterization of objects using artificial neural networks

Abstract Existing NASA supported scientific databases are usually developed and managed by a team of database administrators whose main concern is the efficiency of the databases in terms of normalization and data search constructs. The populating of the database is usually done in a manual fashion by row and column as the data become available, and the data dictionary is usually defined by the same team (at times with little input from the end science user). This process is tedious, error prone and self-limiting in terms of what can be described in a relational Data Base Management System (DBMS). The next generation Earth remote sensing platforms [i.e., Earth Observing System (EOS)] will be capable of generating data at a rate of over 300 Megabits per second from a suite of instruments designed for different applications. What is needed is an innovative approach that creates object-oriented data-bases that segment, characterize, and catalog, and are manageable in a domain-specific context, and whose contents are available interactively and in near-real-time to the user community. This paper describes work in progress that utilizes an artificial neural net approach to characterize satellite imagery of undefined objects into high-level data objects. The characterized data is then dynamically allocated to an object-oriented database where it can be reviewed and accessed by a user. The definition, development, and evolution of the overall data system model are steps in the creation of an application-driven knowledge-based scientific information system.

An intelligent information fusion system for handling the archiving and querying of terabyte‐sized spatial databases

NASA’s Intelligent Data Management Project is conducting research into the development of data management systems that can handle the archiving and querying of data produced by Earth and space missions. Several unique challenges drive the design of these systems, including the volume of the data, the use and interpretation of the data’s temporal, spatial, and spectral components, the size of the userbase, and the desire for fast response times.The Intelligent Data Management group has developed an Intelligent Information Fusion System (IIFS) for testing approaches to handling the archiving and querying of terabyte‐sized spatial databases. Major components of this system are the mass storage and its interactions with the rest of the system; the real‐time planning and scheduling for processing the data; the extraction of metadata and subsequent construction of fast indices for organizing the data along various search dimensions; and the overall user interface.The IIFS design is novel in a number of areas. S...

Design of neural networks for classification of remotely sensed imagery

Abstract As currently planned, future Earth remote sensing platforms (i.e., Earth Observing System [EOS]) will be capable of generating data at a rate of over 50 Megabits per second. To address this issue the Intelligent Data Management (IDM) project at NASA/GSFC has prototyped an Intelligent Information Fusion System (IIFS) that uses backpropagation neural networks for the classification of remotely sensed imagery. This is part of the IDM strategy of providing archived data to a researcher through a variety of discipline-specific indices. In this paper we discuss classification accuracies of a backpropagation neural network and compare it with a maximum likelihood classifier (MLC) with multivariate normal class models. We have found that, because of its nonparametric nature, the neural network outperforms the MLC in this area. In addition, we discuss techniques for constructing optimal neural nets on parallel hardware like the MasPar MP-1 currently at NASA/GSFC. Other important discussions are centered around training and classification times of the two methods, and sensitivity to the training data. Finally we discuss future work in the area of classification and neural nets.

Mission to Planet Earth: AI views the world

In the late 1990s, NASA will launch a series of satellites to study the Earth as a dynamic system. The enormous size and complexity of the resulting data holdings pose several challenges and promise to test the limits of practical AI techniques.

Techniques for managing very large scientific databases

Discusses issues relating to the state of the art in scientific data management. Management of scientific data sets or databases is reviewed. The generic science requirements, as well as a case example that drives the underlying data management system architecture are explored, showing current technology limitations. A concept of intelligent information fusion with sufficient detail on how to integrate advanced technologies to enhance scientific production, is presented. Emphasis is on user interfaces, spatial data structure, uses of neural networks for extracting information from scientific imagery, uses of object-oriented database management systems, animation, and visualization techniques.<<ETX>>