This paper summarizes a method for extracting 3D information and indices of refraction from a scene by means of a pair of polarimetric passive imaging sensors. Each sensor provides the Stokes vector at each sensor pixel location, from which, degree and angle of linear polarization are computed. Angle of linear polarization provides the azimuth angle of the surface normal vector. Two cases are considered. For the special case when the two sensors have a common azimuth plane the index of refraction can be find analytically in terms of the degrees of polarization and the angle between the lines of sight from the two sensors, from which the depression angle of the surface normal can be computed. For the second and more general case the surface normal is estimated from the cross-product of the azimuth vectors from the two sensors and the inner product of the line of sight vectors and surface normal. Once the de-presion angles are estimated the index of refraction can be computed.

We report on the result of study to do a comparative analysis of various image fusion algorithms. As part of this study we have defined a set of measures of effectiveness (MOE) for image fusion performance evaluation and have used them to compare the performance of a variety of image fusion algorithms on real multi-sensor imagery.

Any radar setup is limited by the physical properties of the electromagnetic radiation used to gather information about a target. Some of the limits on accuracy are due to the quantum nature of the photons, others due to the limitations of the electronic equipment used to perform the modulations. The techniques used to record the characteristics of the radiation are as important as the method by which that information is analyzed and interpreted. In each stage of the travels of the radiation, from creation and propagation to recording and analysis, the physical properties of the photons involved play a key role in the behavior of the radiation. The techniques by which these properties are modulated, monitored and recorded are also important. In this paper, these topics are discussed and weighed in detail. From this understanding of the flow of information a theoretical apparatus will be constructed that will be capable of recovering a maximum of that information for later use.

Addresses issues related to classification of images in complex spaces. The image is represented in terms of phase and amplitude components. The classifier optimizes functions of joint real and imaginary conditional probability density functions. A bound on the total probability of errors in terms of Rayleigh quotient is derived and compared to the cases where a non-complex amplitude-only signal is used. Examples of application of the proposed approach on polarimetric radar imagery indicate several orders of magnitude improvement in performance.

Automatic detection and recognition of targets by means of passive IR sensors suffer from limitations due to lack of sufficient contrast between the targets and their background, and among the facets of a target. The results of a suite of polarization-sensitive automatic target detection and recognition algorithms on sets of simulated and real polarimetric IR imagery are presented. A custom designed polarimetric IR (PIR) imaging sensor is used for collecting real polarimetric target data under a variety of conditions. Then a set of novel algorithms are designed and tested that uses the target and background Stokes parameters for detection, segmentation and classification of targets. The empirical performance results in terms of the probabilities of detection, false alarm rate, segmentation accuracy, and recognition probabilities as functions of the number of pixels on target, aspect and depression angles and several background conditions (clutter densities) of applying this ATR algorithms on the polarimetric data and its comparison with a typical IR only ATR are demonstrated, showing that a noticeable improvement can be achieved.

A comparison of four separability measures which are useful in the registration of two-dimensional images is presented. Bayes probability of error, Chernoff bound, Bhattacharyya bound, and Fisher's criteria are used in the selection of the appropriate reference images from a set of aerial pictures, each exhibiting a different view of a scene, i.e., down-looking and target-looking. The experiment is repeated for corresponding synthetic images of this scene. Both area and edge correlations are used. A comparison of the measures and the resulting probabilities of error is made. The results show that for real images the target-looking view performs better than the down-looking view for both area and edge correlations. For synthetic images, the down-looking view performs better than the target-looking view for both area and edge correlations. The variation in synthetic images between the target-looking and the down-looking views is larger than in the real images for both area and edge correlations.

Recognition of three-dimensional objects independent of size, position, and orientation is an important and difficult problem of scene analysis. The use of three-dimensional moment invariants is proposed as a solution. The generalization of the results of two-dimensional moment invariants which had linked two-dimensional moments to binary quantics is done by linking three-dimensional moments to ternary quantics. The existence and number of nth order moments in two and three dimensions is explored. Algebraic invariants of several ternary forms under different orthogonal transformations are derived by using the invariant property of coefficients of ternary forms. The result is a set of three-dimensional moment invariants which are invariant under size, orientation, and position change. This property is highly significant in compressing the data which are needed in three-dimensional object recognition. Empirical examples are also given.