An approach to the detection of specific polarimetric SAR signatures is presented. The polarimetric response in a resolution cell can be viewed as a sample of the electromagnetic scattering matrix for that resolution cell. Through the use of multiple coherent apertures, multiple samples of the scattering matrix can be obtained. With the use of a suitable decomposition and a weighted log-likelihood formulation, it is possible to estimate the relative likelihoods that the observed scattering matrix responses match known electromagnetic signatures.

Details work in our group on the use of low-level features for the identification of man-made regions in unmanned aerial vehicle (UAV) imagery. The feature sets that we have examined include classical statistical features such as the coefficient of variation in a window about a pixel, locally computed fractal dimension, and fractal dimension computed in the presence of wavelet boundaries. We discuss these techniques of feature extraction along with our approach to the classification of the features. Our classification work has focused on the use of a semiparametric probability density estimation technique. In addition, we present classification results for region of interest identification based on a set of test images from an UAV test flight.

In many applications, spatial observations must be segmented into homogeneous regions and the number, positions, and shapes of the regions are unknown a priori. Information about the underlying probability distributions are often unknown. Furthermore, the anticipated regions of interest may be small with few observations from the individual regions. This paper presents a technique designed to address these difficulties. A simple segmentation procedure can be obtained as a clustering of the disjoint subregions obtained through an initial low-level partitioning procedure. Clustering of these subregions based upon a similarity matrix derived from estimates of their marginal probability density functions yields the resultant segmentation. It is shown that this segmentation is improved through the use of a "borrowed strength" density estimation procedure wherein potential similarities between the density functions for the subregions are exploited. The borrowed strength technique is described and the performance of segmentation based on these estimates is investigated through an example from statistical image analysis.

This paper examines techniques for the efficient use of high dimensional feature sets in the detection of micro-calcifications in mammograms. The paper focuses on techniques for dimensionality reduction and discriminant analysis. The paper examines the use of principal components and Fisher's linear discriminant for dimensionality reduction along with parametric and nonparametric statistical techniques for discriminant analysis.>

A neural network architecture whose goal is the computation of a posteriori conditional class probabilities for input vectors that belong to one of two input classes is described. The network architecture has been designed to adaptively produce Voronoi tessellation partitions of the input vectors in R/sup n/ based on the Euclidean distance metric, without regard to the actual a priori class probabilities of the input vectors. These prior probabilities are then used by the network to adaptively compute the a posteriori conditional class probability for the two classes for each tessellation partition. The network presented is thus a connectionist model for vector quantization clustering and includes the process of automatic node creation necessary for many unsupervised learning applications.>