Engineering topics

Swain, P. H.

Publications and source records attributed to Swain, P. H..

At least 19 records

NASA NTRS1993

Conjugate-Gradient Neural Networks in Classification of Multisource and Very-High-Dimensional Remote Sensing Data

Application of neural networks to classification of remote sensing data is discussed. Conventional two-layer backpropagation is found to give good results in classification of remote sensing data but is not efficient in training. A more efficient variant, based on conjugate-gradient optimization, is used for classification of multisource remote sensing and geographic data and very-high-dimensional data. The conjugate-gradient neural networks give excellent performance in classification of multisource data, but do not compare as well with statistical methods in classification of very-high-dimentional data.

Benediktsson, J. A.↗

NASA NTRS1993

Parallel Consensual Neural Networks

A new neural network architecture is proposed and applied in classification of remote sensing/geographic data from multiple sources. The new architecture is called the parallel consensual neural network and its relation to hierarchical and ensemble neural networks is discussed. The parallel consensual neural network architecture is based on statistical consensus theory. The input data are transformed several times and the different transformed data are applied as if they were independent inputs and are classified using stage neural networks. Finally, the outputs from the stage networks are then weighted and combined to make a decision. Experimental results based on remote sensing data and geographic data are given. The performance of the consensual neural network architecture is compared to that of a two-layer (one hidden layer) conjugate-gradient backpropagation neural network. The results with the proposed neural network architecture compare favorably in terms of classification accuracy to the backpropagation method.

Benediktsson, J. A.↗

NASA NTRS1991

A method for classification of multisource data using interval-valued probabilities and its application to HIRIS data

A method of classifying multisource data in remote sensing is presented. The proposed method considers each data source as an information source providing a body of evidence, represents statistical evidence by interval-valued probabilities, and uses Dempster's rule to integrate information based on multiple data source. The method is applied to the problems of ground-cover classification of multispectral data combined with digital terrain data such as elevation, slope, and aspect. Then this method is applied to simulated 201-band High Resolution Imaging Spectrometer (HIRIS) data by dividing the dimensionally huge data source into smaller and more manageable pieces based on the global statistical correlation information. It produces higher classification accuracy than the Maximum Likelihood (ML) classification method when the Hughes phenomenon is apparent.

Kim, H.↗

NASA NTRS1991

A consensual neural network

A neural network architecture called a consensual neural network (CNN) is proposed for the classification of data from multiple sources. Its relation to hierarchical and ensemble neural networks is discussed. CNN is based on the statistical consensus theory and uses nonlinearly transformed input data. The input data are transformed several times, and the different transformed data are applied as if they were independent inputs. The independent inputs are classified using stage neural networks and outputs from the stage networks are then weighted and combined to make a decision. Experimental results based on remote-sensing data and geographic data are given.

Benediktsson, J. A.↗

NASA NTRS1990

A method for classification of multisource data using interval-valued probabilities and its application to HIRIS data

A method of classifying multisource data in remote sensing is presented. The proposed method considers each data source as an information source providing a body of evidence, represents statistical evidence by interval-valued probabilities, and uses Dempster's rule to integrate information based on multiple data sources. The method is applied to the problems of ground-cover classification of multispectral data combined with digital terrain data such as elevation, slope, and aspect. Then this method is applied to simulated 201-band High Resolution Imaging Spectrometer (HIRIS) data by dividing the dimensionally huge data source into smaller and more manageable pieces based on the global statistical correlation information. It produces higher classification accuracy than the Maximum Likelihood (ML) classification method when the Hughes phenomenon is apparent.

Kim, H.↗

NASA NTRS1989

Combining multispectral and ancillary data in remote sensing using interval-valued probabilities

This paper presents a method for classifying multisource data in remote sensing and geographic information systems using interval-valued probabilities. In this method, each data source is considered as an information source which provides a body of statistical evidence. In order to integrate information obtained from multiple data sources, the method adopts Dempster's rule for combining multiple bodies of evidence. Preliminary experiments have been undertaken to illustrate the use of the method in a supervised ground-cover classification on multispectral data combined with digital elevation data. They demonstrate the ability of the method in capturing information provided by inexact and incomplete evidence when there are not enough training samples to estimate statistical parameters.

Kim, H.↗

NASA NTRS1982

A means for utilizing ancillary information in multispectral classification

Attention is given to a method which makes it possible to incorporate information from ancillary data sources into the results of an existing classification of remotely sensed data. The method is based upon probabilistic label relaxation procedures which are used for imbedding spatial context data in image-labeling problems. The source of ancillary information is utilized in the form of a set of probabilities. These are introduced into a modified relaxation method called supervised relaxation labeling which, on application, develops a labeling for remotely sensed data that strikes a balance in consistency between spectral, spatial, and ancillary data sources of information. A description is given of results of a forestry classification in which accuracy is improved from 68% to 81% by incorporating topographic elevation in the outlined manner.

Richards, J. A.↗

NASA NTRS1982

Estimation of context for statistical classification of multispectral image data

Recent investigations have demonstrated the effectiveness of a contextual classifier that combines spatial and spectral information employing a general statistical approach. This statistical classification algorithm exploits the tendency of certain ground cover classes to occur more frequently in some spatial contexts than in others. Indeed, a key input to this algorithm is a statistical characterization of the context: the context function. An unbiased estimator of the context function is discussed which, besides having the advantage of statistical unbiasedness, has the additional advantage over other estimation techniques of being amenable to an adaptive implementation in which the context-function estimate varies according to local contextual information. Results from applying the unbiased estimator to the contextual classification of three real Landsat data sets are presented and contrasted with results from noncontextual classifications and from contextual classifications utilizing other context-function estimation techniques.

Tilton, J. C.↗

NASA NTRS1982

A quantitative applications-oriented evaluation of Thematic Mapper design specifications

Simulated Thematic Mapper data sets were created from aircraft Multispectral Scanner data in order to predict the sensitivity of applications-oriented classification results to variations in selected sensor parameters. It was found that band-to-band misregistration as little as 0.3 pixel can have a pronounced effect on the classification of both field-center pixels and pixels associated with edges and small objects in the scene. It was also determined that a significant improvement in classification accuracy could be achieved if the resolution of the thermal band of the Thematic Mapper were comparable to the resolution of its other spectral bands.

Swain, P. H.↗

NASA NTRS1981

Incorporating spatial context into statistical classification of multidimensional image data

Compound decision theory is employed to develop a general statistical model for classifying image data using spatial context. The classification algorithm developed from this model exploits the tendency of certain ground-cover classes to occur more frequently in some spatial contexts than in others. A key input to this contextural classifier is a quantitative characterization of this tendency: the context function. Several methods for estimating the context function are explored, and two complementary methods are recommended. The contextural classifier is shown to produce substantial improvements in classification accuracy compared to the accuracy produced by a non-contextural uniform-priors maximum likelihood classifier when these methods of estimating the context function are used. An approximate algorithm, which cuts computational requirements by over one-half, is presented. The search for an optimal implementation is furthered by an exploration of the relative merits of using spectral classes or information classes for classification and/or context function estimation.

Bauer, M. E.↗

NASA NTRS1981

Key Issues in the Analysis of Remote Sensing Data: A report on the workshop

The procedures of a workshop assessing the state of the art of machine analysis of remotely sensed data are summarized. Areas discussed were: data bases, image registration, image preprocessing operations, map oriented considerations, advanced digital systems, artificial intelligence methods, image classification, and improved classifier training. Recommendations of areas for further research are presented.

Swain, P. H.↗

NASA NTRS1981

On the accuracy of pixel relaxation labeling

An analysis of pixel labeling by probabilistic relaxation techniques is presented to demonstrate that these labeling procedures degenerate to weighted averages in the vicinity of fixed points. A consequence of this is that undesired label conversions can occur, leading to a deterioration of labeling accuracy at a stage after an improvement has already been achieved. Means for overcoming the accuracy deterioration are suggested and used as the basis for a possible design strategy for using probabilistic relaxation procedures. The results obtained are illustrated using simple data sets in which labeling on individual pixels can be examined and also using Landsat imagery to show application to data typical of that encountered in remote sensing applications.

Richards, J. A.↗

NASA NTRS1981

Pixel labeling by supervised probabilistic relaxation

A simple modification to existing probabilistic relaxation procedures is suggested which allows the information contained in initial labels to exert an influence on the direction of relaxation throughout the process. In this manner, the initial labels assume more importance than with conventional algorithms and are used in combination with the outcome of relaxation at each iteration to produce a cooperative estimate of the correct label for a particular object. Pixel labeling examples are presented which show the performance that can be obtained with the modified algorithm. The procedure is readily generalized to allow other data to influence the process.

Richards, J. A.↗

NASA NTRS1981

Contextual classification on a CDC Flexible Processor system

A potential hardware organization for the Flexible Processor Array is presented. An algorithm that implements a contextual classifier for remote sensing data analysis is given, along with uniprocessor classification algorithms. The Flexible Processor algorithm is provided, as are simulated timings for contextual classifiers run on the Flexible Processor Array and another system. The timings are analyzed for context neighborhoods of sizes three and nine.

Smith, B. W.↗

NASA NTRS1981

Contextual classification of multispectral image data - An unbiased estimator for the context distribution

Recent investigations have demonstrated the effectiveness of a contextual classifier that combines spatial and spectral information employing a general statistical approach. This statistical classification algorithm exploits the tendency of certain ground-cover classes to occur more frequently in some spatial contexts than in others. Indeed, a key input to this algorithm is a statistical characterization of the context: the context distribution. Here a discussion is given of an unbiased estimator of the context distribution which, besides having the advantage of statistical unbiasedness, has the additional advantage over other estimation techniques of being amenable to an adaptive implementation in which the context distribution estimate varies according to local contextual information. Results from applying the unbiased estimator to the contextual classification of three real Landsat data sets are presented and contrasted with results from noncontextual classifications and from contextual classifications utilizing other context distribution estimation techniques.

Tilton, J. C.↗

NASA NTRS1981

Contextual classification of multispectral image data: An unbiased estimator for the context distribution

A key input to a statistical classification algorithm, which exploits the tendency of certain ground cover classes to occur more frequently in some spatial context than in others, is a statistical characterization of the context: the context distribution. An unbiased estimator of the context distribution is discussed which, besides having the advantage of statistical unbiasedness, has the additional advantage over other estimation techniques of being amenable to an adaptive implementation in which the context distribution estimate varies according to local contextual information. Results from applying the unbiased estimator to the contextual classification of three real LANDSAT data sets are presented and contrasted with results from non-contextual classifications and from contextual classifications utilizing other context distribution estimation techniques.

Tilton, J. C.↗

NASA NTRS1981

The quantitative methodology of remote sensing: An educator's perspective

Strategies for teaching the fundamentals of quantitative remote sensing to students from diverse disciplines and with a broad range of mathematical backgrounds are described. Because the concepts are fairly abstract and complex, the successful educator must take care to motivate and assist the students in mastering the concepts. He/she should strive to communicate intuitive notions of the concepts couched in settings as familiar as possible and use these settings to illustrate each mathematical or statistical formulation which is introduced. The utilization of laboratory-type experiences and lectures from experts is also suggested.

Swain, P. H.↗

NASA NTRS1981

Contextual classification of multispectral image data

A general method is presented for exploiting both spatial and spectral information when classifying multispectral image data. This statistical classification algorithm utilizes the tendency of certain ground cover classes to be more likely to occur in some contexts than others. The theoretical model assumes the two-dimensional array of random observations and a 0-1 loss function, a distribution of the p-context array that is spatially invariant, and class-conditional independence for the observations. The problems that prevent the immediate use of this context classifier are the need for a generally applicable method for making adequate estimates of the context distribution and a reduction in the computational intensivity of the classifier. The former problem is being approached by a method that raises the relative frequency value for each class configuration to a power and uses the result as the context distribution estimate. The second is being approached by searching for a less computationally intensive algorithm.

Tilton, J. C.↗