Browsing by Author "Phillips, Rhonda D."

Now showing 1 - 9 of 9
  • Thumbnail Image
    An Adaptive Noise Filtering Algorithm for AVIRIS Data with Implications for Classification Accuracy
    Phillips, Rhonda D.; Blinn, Christine E.; Watson, Layne T.; Wynne, Randolph H. (Department of Computer Science, Virginia Polytechnic Institute & State University, 2008)
    This paper describes a new algorithm used to adaptively filter a remote sensing dataset based on signal-to-noise ratios (SNRs) once the maximum noise fraction (MNF) has been applied. This algorithm uses Hermite splines to calculate the approximate area underneath the SNR curve as a function of band number, and that area is used to place bands into “bins” with other bands having similar SNRs. A median filter with a variable sized kernel is then applied to each band, with the same size kernel used for each band in a particular bin. The proposed adaptive filters are applied to a hyperspectral image generated by the AVIRIS sensor, and results are given for the identification of three different pine species located within the study area. The adaptive filtering scheme improves image quality as shown by estimated SNRs, and classification accuracies improved by more than 10% on the sample study area, indicating that the proposed methods improve the image quality, thereby aiding in species discrimination.
  • Thumbnail Image
    Continuous Iterative Guided Spectral Class Rejection Classification Algorithm: Part 2
    Phillips, Rhonda D.; Watson, Layne T.; Wynne, Randolph H.; Ramakrishnan, Naren (Department of Computer Science, Virginia Polytechnic Institute & State University, 2009)
    This paper describes in detail the continuous iterative guided spectral class rejection (CIGSCR) classification method based on the iterative guided spectral class rejection (IGSCR) classification method for remotely sensed data. Both CIGSCR and IGSCR use semisupervised clustering to locate clusters that are associated with classes in a classification scheme. In CIGSCR and IGSCR, training data are used to evaluate the strength of the association between a particular cluster and a class, and a statistical hypothesis test is used to determine which clusters should be associated with a class and used for classification and which clusters should be rejected and possibly refined. Experimental results indicate that the soft classification output by CIGSCR is reasonably accurate (when compared to IGSCR), and the fundamental algorithmic changes in CIGSCR (from IGSCR) result in CIGSCR being less sensitive to input parameters that influence iterations. Furthermore, evidence is presented that the semisupervised clustering in CIGSCR produces more accurate classifications than classification based on clustering without supervision.
  • Thumbnail Image
    Continuous Iterative Guided Spectral Class Rejection Classification Algorithm: Part 1
    Phillips, Rhonda D.; Watson, Layne T.; Wynne, Randolph H.; Ramakrishnan, Naren (Department of Computer Science, Virginia Polytechnic Institute & State University, 2009)
    This paper outlines the changes necessary to convert the iterative guided spectral class rejection (IGSCR) classification algorithm to a soft classification algorithm. IGSCR uses a hypothesis test to select clusters to use in classification and iteratively refines clusters not yet selected for classification. Both steps assume that cluster and class memberships are crisp (either zero or one). In order to make soft cluster and class assignments (between zero and one), a new hypothesis test and iterative refinement technique are introduced that are suitable for soft clusters. The new hypothesis test, called the (class) association significance test, is based on the normal distribution, and a proof is supplied to show that the assumption of normality is reasonable. Soft clusters are iteratively refined by creating new clusters using information contained in a targeted soft cluster. Soft cluster evaluation and refinement can then be combined to form a soft classification algorithm, continuous iterative guided spectral class rejection (CIGSCR).
  • Thumbnail Image
    Enrichment Procedures for Soft Clusters: A Statistical Test and its Applications
    Phillips, Rhonda D.; Watson, Layne T.; Wynne, Randolph H.; Ramakrishnan, Naren (Department of Computer Science, Virginia Polytechnic Institute & State University, 2010-02-01)
    Clusters, typically mined by modeling locality of attribute spaces, are often evaluated for their ability to demonstrate ‘enrichment’ of categorical features. A cluster enrichment procedure evaluates the membership of a cluster for significant representation in pre-defined categories of interest. While classical enrichment procedures assume a hard clustering definition, in this paper we introduce a new statistical test that computes enrichments for soft clusters. We demonstrate an application of this test in refining and evaluating soft clusters for classification of remotely sensed images.
  • Thumbnail Image
    Feature Reduction using a Singular Value Decomposition for the Iterative Guided Spectral Class Rejection Hybrid Classifier
    Phillips, Rhonda D.; Watson, Layne T.; Wynne, Randolph H.; Blinn, Christine E. (Department of Computer Science, Virginia Polytechnic Institute & State University, 2007)
    Feature reduction in a remote sensing dataset is often desirable to decrease the processing time required to perform a classification and improve overall classification accuracy. This work introduces a feature reduction method based on the singular value decomposition (SVD). This feature reduction technique was applied to training data from two multitemporal datasets of Landsat TM/ETM+ imagery acquired over a forested area in Virginia, USA and Rondonia, Brazil. Subsequent parallel iterative guided spectral class rejection (pIGSCR) forest/nonforest classifications were performed to determine the quality of the feature reduction. The classifications of the Virginia data were five times faster using SVDbased feature reduction without affecting the classification accuracy. Feature reduction using the SVD was also compared to feature reduction using principal components analysis (PCA). The highest average accuracies for the Virginia dataset (88.34%) and for the Rondonia dataset (93.31%) were achieved using the SVD. The results presented here indicate that SVDbased feature reduction can produce statistically significantly better classifications than PCA.
  • Thumbnail Image
    Improving the Performance of a Hybrid Classification Method Using a Parallel Algorithm and a Novel Data Reduction Technique
    Phillips, Rhonda D. (Virginia Tech, 2007-05-09)
    This thesis presents both a shared memory parallel version of the hybrid classification algorithm IGSCR (iterative guided spectral class rejection) and a novel data reduction technique that can be used in conjuction with pIGSCR (parallel IGSCR). The parallel algorithm is motivated by a demonstrated need for more computing power driven by the increasing size of remote sensing datasets due to higher resolution sensors, larger study regions, and the like. Even with a fast algorithm such as pIGSCR, the reduction of dimension in a dataset is desirable in order to decrease the processing time further and possibly improve overall classification accuracy. pIGSCR was developed to produce fast and portable code using Fortran 95, OpenMP, and the Hierarchical Data Format version 5 (HDF5) and accompanying data access library. The applicability of the faster pIGSCR algorithm is demonstrated by classifying Landsat data covering most of Virginia, USA into forest and non-forest classes with approximately 90 percent accuracy. Parallel results are given using the SGI Altix 3300 shared memory computer and the SGI Altix 3700 with as many as 64 processors reaching speedups of almost 77. This fast algorithm allows an analyst to perform and assess multiple classifications to refine parameters. As an example, pIGSCR was used for a factorial analysis consisting of 42 classifications of a 1.2 gigabyte image to select the number of initial classes (70) and class purity (70%) used for the remaining two images. A feature selection or reduction method may be appropriate for a specific lassification method depending on the properties and training required for the classification method, or an alternative band selection method may be derived based on the classification method itself. This thesis introduces a feature reduction method based on the singular value decomposition (SVD). This feature reduction technique was applied to training data from two multitemporal datasets of Landsat TM/ETM+ imagery acquired over a forested area in Virginia, USA and Rondonia, Brazil. Subsequent parallel iterative guided spectral class rejection (pIGSCR) forest/non-forest classifications were performed to determine the quality of the feature reduction. The classifications of the Virginia data were five times faster using SVD based feature reduction without affecting the classification accuracy. Feature reduction using the SVD was also compared to feature reduction using principal components analysis (PCA). The highest average accuracies for the Virginia dataset (88.34%) and for the Amazon dataset (93.31%) were achieved using the SVD. The results presented here indicate that SVD based feature reduction can produce statistically significantly better classifications than PCA.
  • Thumbnail Image
    Note on the Effectiveness OF Stochastic Optimization Algorithms for Robust Design
    Iyer, Manjula A.; Phillips, Rhonda D.; Trosset, Michael W.; Watson, Layne T. (Department of Computer Science, Virginia Polytechnic Institute & State University, 2008-05-01)
    Robust design optimization (RDO) uses statistical decision theory and optimization techniques to optimize a design over a range of uncertainty (introduced by the manufacturing process and unintended uses). Since engineering ob jective functions tend to be costly to evaluate and prohibitively expensive to integrate (required within RDO), surrogates are introduced to allow the use of traditional optimization methods to find solutions. This paper explores the suitability of radically different (deterministic and stochastic) optimization methods to solve prototypical robust design problems. The algorithms include a genetic algorithm using a penalty function formulation, the simultaneous perturbation stochastic approximation (SPSA) method, and two gradient-based constrained nonlinear optimizers (method of feasible directions and sequential quadratic programming). The results show that the fully deterministic standard optimization algorithms are consistently more accurate, consistently more likely to terminate at feasible points, and consistently considerably less expensive than the fully nondeterministic algorithms.
  • Thumbnail Image
    A Probabilistic Classification Algorithm With Soft Classification Output
    Phillips, Rhonda D. (Virginia Tech, 2009-03-30)
    This thesis presents a shared memory parallel version of the hybrid classification algorithm IGSCR (iterative guided spectral class rejection), a novel data reduction technique that can be used in conjunction with PIGSCR (parallel IGSCR), a noise removal method based on the maximum noise fraction (MNF), and a continuous version of IGSCR (CIGSCR) that outputs soft classifications. All of the above are either classification algorithms or preprocessing algorithms necessary prior to the classification of high dimensional, noisy images. PIGSCR was developed to produce fast and portable code using Fortran 95, OpenMP, and the Hierarchical Data Format version 5 (HDF5) and accompanying data access library. The feature reduction method introduced in this thesis is based on the singular value decomposition (SVD). This feature reduction technique demonstrated that SVD-based feature reduction can lead to more accurate IGSCR classifications than PCA-based feature reduction. This thesis describes a new algorithm used to adaptively filter a remote sensing dataset based on signal-to-noise ratios (SNRs) once the maximum noise fraction (MNF) has been applied. The adaptive filtering scheme improves image quality as shown by estimated SNRs and classification accuracy improvements greater than 10%. The continuous iterative guided spectral class rejection (CIGSCR) classification method is based on the iterative guided spectral class rejection (IGSCR) classification method for remotely sensed data. Both CIGSCR and IGSCR use semisupervised clustering to locate clusters that are associated with classes in a classification scheme. This type of semisupervised classification method is particularly useful in remote sensing where datasets are large, training data are difficult to acquire, and clustering makes the identification of subclasses adequate for training purposes less difficult. Experimental results indicate that the soft classification output by CIGSCR is reasonably accurate (when compared to IGSCR), and the fundamental algorithmic changes in CIGSCR (from IGSCR) result in CIGSCR being less sensitive to input parameters that influence iterations.
  • Thumbnail Image
    An SMP Soft Classification Algorithm for Remote Sensing
    Phillips, Rhonda D.; Watson, Layne T.; Easterling, David R.; Wynne, Randolph H. (Department of Computer Science, Virginia Polytechnic Institute & State University, 2012)
    This work introduces a symmetric multiprocessing (SMP) version of the continuous iterative guided spectral class rejection (CIGSCR) algorithm, a semiautomated classification algorithm for remote sensing (multispectral) images. The algorithm uses soft data clusters to produce a soft classification containing inherently more information than a comparable hard classification at an increased computational cost. Previous work suggests that similar algorithms achieve good parallel scalability, motivating the parallel algorithm development work here. Experimental results of applying parallel CIGSCR to an image with approximately 10^8 pixels and six bands demonstrate superlinear speedup. A soft two class classification is generated in just over four minutes using 32 processors.