Downloads & Free Reading Options - Results

The Mathematical Pattern Recognition and Image Analysis (MPRIA) Project is concerned with basic research problems related to the study of the Earth from remotely sensed measurement of its surface characteristics. The program goal is to better understand how to analyze the digital image that represents the spatial, spectral, and temporal arrangement of these measurements for purposing of making selected inference about the Earth.

The automated contour recognition and classification in aerial phtography requires a feature of contours that is invariant to shift, rotation and scaling. General polygons are characterized by their angles and their normalized sides. The normalization of the sides may be achieved by dividing either with the longest side or with the circumference of the polygon. For a curved contour, the simplest invariant under shift and rotation is the curvature. To make the curvature invariant to scaling, one may divide it by the largest curvature or multiply it with the length of the contour; the multiplication with the length of the contour is better because of the finite resolution of photographs and equipment. The principle of contour recognition and classification based on angles and curvature is worked out for practical displays that produce square patterns, such as liquid crystal or plasma tube displays. The mechanical and electronic design for experimental equipment is carried out based on a plasma tube display with 512 x 512 resolved points. (Author)

The application of mathematical and statistical analyses techniques to imagery obtained by remote sensors is described by Principal Investigators. Scene-to-map registration, geometric rectification, and image matching are among the pattern recognition aspects discussed.

The application of mathematical and statistical analyses techniques to imagery obtained by remote sensors is described by Principal Investigators. Scene-to-map registration, geometric rectification, and image matching are among the pattern recognition aspects discussed.

The application of mathematical and statistical analyses techniques to imagery obtained by remote sensors is described by Principal Investigators. Scene-to-map registration, geometric rectification, and image matching are among the pattern recognition aspects discussed.

This work investigates the use or Bayesian multiscale techniques for image analysis and automatic target recognition. We have developed two new techniques. First, we have develop a wavelet-based approach to image restoration and deconvolution problems using Bayesian image models and an alternating-maximation method. Second, we have developed a wavelet-based framework for target modeling and recognition that we call TEMPLAR (TEMPlate Learning from Atomic Representations) . TEMPLAR is can he used to automatically extract low-dimensional wavelet representations (or templates) or target objects from observation data, providing robust and computationally efficient target classifiers. On a more theoretical level, we have developed a framework for multiresolution analysis or likelihood functions, which extends wavelet-like analysis to a wide class or non-Gaussian processes. In another line of investigation, we are exploring a new imaging application known as network tomography. The goal of this work is to characterize the internal performance of communication networks based only on external measurements at the edge (sources and receivers) of the network. In the coming year, we plan to focus on four key research areas. First, we will develop theoretical hounds on the performance of multiscale/wavelet estimators in non-Gaussian environments including Poisson imaging applications. Second, we will study the use of complex wavelets in image restoration and target recognition problems. Third, we will develop automatic methods for segmenting imagery (SAR, FLIR, LADAR) based on complexity-regularization methods. Fourth, we will continue to develop a unified framework for communication network tomography and investigate new tools for network performance visualization.

This work investigates the use or Bayesian multiscale techniques for image analysis and automatic target recognition. We have developed two new techniques. First, we have develop a wavelet-based approach to image restoration and deconvolution problems using Bayesian image models and an alternating-maximation method. Second, we have developed a wavelet-based framework for target modeling and recognition that we call TEMPLAR (TEMPlate Learning from Atomic Representations) . TEMPLAR is can he used to automatically extract low-dimensional wavelet representations (or templates) or target objects from observation data, providing robust and computationally efficient target classifiers. On a more theoretical level, we have developed a framework for multiresolution analysis or likelihood functions, which extends wavelet-like analysis to a wide class or non-Gaussian processes. In another line of investigation, we are exploring a new imaging application known as network tomography. The goal of this work is to characterize the internal performance of communication networks based only on external measurements at the edge (sources and receivers) of the network. In the coming year, we plan to focus on four key research areas. First, we will develop theoretical hounds on the performance of multiscale/wavelet estimators in non-Gaussian environments including Poisson imaging applications. Second, we will study the use of complex wavelets in image restoration and target recognition problems. Third, we will develop automatic methods for segmenting imagery (SAR, FLIR, LADAR) based on complexity-regularization methods. Fourth, we will continue to develop a unified framework for communication network tomography and investigate new tools for network performance visualization.

This work investigates the use or Bayesian multiscale techniques for image analysis and automatic target recognition. We have developed two new techniques. First, we have develop a wavelet-based approach to image restoration and deconvolution problems using Bayesian image models and an alternating-maximation method. Second, we have developed a wavelet-based framework for target modeling and recognition that we call TEMPLAR (TEMPlate Learning from Atomic Representations) . TEMPLAR is can he used to automatically extract low-dimensional wavelet representations (or templates) or target objects from observation data, providing robust and computationally efficient target classifiers. On a more theoretical level, we have developed a framework for multiresolution analysis or likelihood functions, which extends wavelet-like analysis to a wide class or non-Gaussian processes. In another line of investigation, we are exploring a new imaging application known as network tomography. The goal of this work is to characterize the internal performance of communication networks based only on external measurements at the edge (sources and receivers) of the network. In the coming year, we plan to focus on four key research areas. First, we will develop theoretical hounds on the performance of multiscale/wavelet estimators in non-Gaussian environments including Poisson imaging applications. Second, we will study the use of complex wavelets in image restoration and target recognition problems. Third, we will develop automatic methods for segmenting imagery (SAR, FLIR, LADAR) based on complexity-regularization methods. Fourth, we will continue to develop a unified framework for communication network tomography and investigate new tools for network performance visualization.

This work investigates the use or Bayesian multiscale techniques for image analysis and automatic target recognition. We have developed two new techniques. First, we have develop a wavelet-based approach to image restoration and deconvolution problems using Bayesian image models and an alternating-maximation method. Second, we have developed a wavelet-based framework for target modeling and recognition that we call TEMPLAR (TEMPlate Learning from Atomic Representations) . TEMPLAR is can he used to automatically extract low-dimensional wavelet representations (or templates) or target objects from observation data, providing robust and computationally efficient target classifiers. On a more theoretical level, we have developed a framework for multiresolution analysis or likelihood functions, which extends wavelet-like analysis to a wide class or non-Gaussian processes. In another line of investigation, we are exploring a new imaging application known as network tomography. The goal of this work is to characterize the internal performance of communication networks based only on external measurements at the edge (sources and receivers) of the network. In the coming year, we plan to focus on four key research areas. First, we will develop theoretical hounds on the performance of multiscale/wavelet estimators in non-Gaussian environments including Poisson imaging applications. Second, we will study the use of complex wavelets in image restoration and target recognition problems. Third, we will develop automatic methods for segmenting imagery (SAR, FLIR, LADAR) based on complexity-regularization methods. Fourth, we will continue to develop a unified framework for communication network tomography and investigate new tools for network performance visualization.

This work investigates the use or Bayesian multiscale techniques for image analysis and automatic target recognition. We have developed two new techniques. First, we have develop a wavelet-based approach to image restoration and deconvolution problems using Bayesian image models and an alternating-maximation method. Second, we have developed a wavelet-based framework for target modeling and recognition that we call TEMPLAR (TEMPlate Learning from Atomic Representations) . TEMPLAR is can he used to automatically extract low-dimensional wavelet representations (or templates) or target objects from observation data, providing robust and computationally efficient target classifiers. On a more theoretical level, we have developed a framework for multiresolution analysis or likelihood functions, which extends wavelet-like analysis to a wide class or non-Gaussian processes. In another line of investigation, we are exploring a new imaging application known as network tomography. The goal of this work is to characterize the internal performance of communication networks based only on external measurements at the edge (sources and receivers) of the network. In the coming year, we plan to focus on four key research areas. First, we will develop theoretical hounds on the performance of multiscale/wavelet estimators in non-Gaussian environments including Poisson imaging applications. Second, we will study the use of complex wavelets in image restoration and target recognition problems. Third, we will develop automatic methods for segmenting imagery (SAR, FLIR, LADAR) based on complexity-regularization methods. Fourth, we will continue to develop a unified framework for communication network tomography and investigate new tools for network performance visualization.

This work investigates the use or Bayesian multiscale techniques for image analysis and automatic target recognition. We have developed two new techniques. First, we have develop a wavelet-based approach to image restoration and deconvolution problems using Bayesian image models and an alternating-maximation method. Second, we have developed a wavelet-based framework for target modeling and recognition that we call TEMPLAR (TEMPlate Learning from Atomic Representations) . TEMPLAR is can he used to automatically extract low-dimensional wavelet representations (or templates) or target objects from observation data, providing robust and computationally efficient target classifiers. On a more theoretical level, we have developed a framework for multiresolution analysis or likelihood functions, which extends wavelet-like analysis to a wide class or non-Gaussian processes. In another line of investigation, we are exploring a new imaging application known as network tomography. The goal of this work is to characterize the internal performance of communication networks based only on external measurements at the edge (sources and receivers) of the network. In the coming year, we plan to focus on four key research areas. First, we will develop theoretical hounds on the performance of multiscale/wavelet estimators in non-Gaussian environments including Poisson imaging applications. Second, we will study the use of complex wavelets in image restoration and target recognition problems. Third, we will develop automatic methods for segmenting imagery (SAR, FLIR, LADAR) based on complexity-regularization methods. Fourth, we will continue to develop a unified framework for communication network tomography and investigate new tools for network performance visualization.

Fundamental problems encountered while attempting to develop automated techniques for applications of remote sensing are discussed under the following categories: (1) geometric and radiometric preprocessing; (2) spatial, spectral, temporal, syntactic, and ancillary digital image representation; (3) image partitioning, proportion estimation, and error models in object scene interference; (4) parallel processing and image data structures; and (5) continuing studies in polarization; computer architectures and parallel processing; and the applicability of "expert systems" to interactive analysis.

Fundamental problems encountered while attempting to develop automated techniques for applications of remote sensing are discussed under the following categories: (1) geometric and radiometric preprocessing; (2) spatial, spectral, temporal, syntactic, and ancillary digital image representation; (3) image partitioning, proportion estimation, and error models in object scene interference; (4) parallel processing and image data structures; and (5) continuing studies in polarization; computer architectures and parallel processing; and the applicability of "expert systems" to interactive analysis.

Fundamental problems encountered while attempting to develop automated techniques for applications of remote sensing are discussed under the following categories: (1) geometric and radiometric preprocessing; (2) spatial, spectral, temporal, syntactic, and ancillary digital image representation; (3) image partitioning, proportion estimation, and error models in object scene interference; (4) parallel processing and image data structures; and (5) continuing studies in polarization; computer architectures and parallel processing; and the applicability of "expert systems" to interactive analysis.

Computer programs for analysis of connectivities and shapes of regions in digitized images

Computer programs for analysis of connectivities and shapes of regions in digitized images

Computer programs for analysis of connectivities and shapes of regions in digitized images

Computer programs for analysis of connectivities and shapes of regions in digitized images

Computer programs for analysis of connectivities and shapes of regions in digitized images

Topics addressed include: multivariate spline method; normal mixture analysis applied to remote sensing; image data analysis; classifications in spatially correlated environments; probability density functions; graphical nonparametric methods; subpixel registration analysis; hypothesis integration in image understanding systems; rectification of satellite scanner imagery; spatial variation in remotely sensed images; smooth multidimensional interpolation; and optimal frequency domain textural edge detection filters.

The research conducted under the grant was concerned with three theoretical aspects of image analysis: (1) statistical image modeling and texture analysis; (2) digital geometry (geometrical properties of subsets of digital images); and (3) parallel image processing (including formal models for parallel image recognition). During the period 1 May 77-31 Dec 82, 116 technical reports were issued on the grant. Nearly all of these have been published or in process of acceptance for publication. A bibliography of these reports is given. Seven students wrote Ph.D dissertations with the support of the grant. (Author)

Pattern Recognition Software and Techniques for Biological Image Analysis

Pattern Recognition Software and Techniques for Biological Image Analysis

Pattern Recognition Software and Techniques for Biological Image Analysis

Substantial work was done on the method of deformable templates for object detection, recognition, and interpretation. New measures of deformation were introduced, and these yielded new and more efficient algorithms for template matching. Experiments were carried out for target tracking in radar imagery and for anatomical labeling in a variety of medical imaging applications. A new detection and tracking algorithm was developed. The algorithm uses ideas from classical filtering, but in the context of a nonlinear model and modern computational tools. Experiments were performed on challenging imagery obtained by filming several fish swimming in a large fish tank containing an elaborate array of objects.

Image Pattern Recognition attempts to infer properties of the world from image data. Such capabilities are crucial for making measurements from satellite or telescope images related to Earth and space science problems. Such measurements can be the required product itself, or the measurements can be used as input to a computer graphics system for visualization purposes. At present, the field of image pattern recognition lacks a unified scientific structure for developing and evaluating image pattern recognition applications. The overall goal of this project is to begin developing such a structure. This report summarizes results of a 3-year research effort in image pattern recognition addressing the following three principal aims: (1) to create a software foundation for the research and identify image pattern recognition problems in Earth and space science; (2) to develop image measurement operations based on Artificial Visual Systems; and (3) to develop multiscale image descriptions for use in interactive image analysis.

The Mathematical Pattern Recognition and Image Analysis (MPRIA) Project is concerned with basic research problems related to the study of he Earth from remotely sensed measurements of its surface characteristics. The program goal is to better understand how to analyze the digital image that represents the spatial, spectral, and temporal arrangement of these measurements for purposing of making selected inferences about the Earth. This report summarizes the progress that has been made toward this program goal by each of the principal investigators in the MPRIA Program.

The Mathematical Pattern Recognition and Image Analysis (MPRIA) Project is concerned with basic research problems related to the study of he Earth from remotely sensed measurements of its surface characteristics. The program goal is to better understand how to analyze the digital image that represents the spatial, spectral, and temporal arrangement of these measurements for purposing of making selected inferences about the Earth. This report summarizes the progress that has been made toward this program goal by each of the principal investigators in the MPRIA Program.

This research project produced results of both fundamental and practical benefit, including a quantitative description of the finite sample accuracy of the k nearest neighbor classifier for a large family of smooth pattern recognition problems, a new theoretical justification for use of a weighted euclidean metric as a similarity function, the development of a procedure for estimating the Bayes risk of practical problems, and the development of a fast approximation of a kappa nearest neighbor classifier, called the labeled cell classifier. The research resulted in a stand-alone X Windows software application, called pstool, that allows users to interactively construct training sets from multispectral digital images, six refereed conference publications, a 40 page technical report, and a journal publication in the Annals of Statistics. Seven graduate students at the University of Vermont participated in this project: four receiving Masters degrees, and one, a Ph.D in Electrical Engineering.

Face recognition is still an open problem. Many 2D face recognition approaches came into light to achieve high recognition rate. But these approaches are still challenged by the changes in illuminations, expressions, pose, noise, etc. A 3D face recognition technique is proposed to overcome such challenges and to enhance robustness to expression variations. Here, we compare the person at different age groups with higher recognition rate in comparison to 2D face recognition techniques. We propose a two stage procedure of 3D face recognition based on FLD (Fisher Linear Discriminant), SURF operator and depth-image. First, FLD is used on depth-image to perform recognition and then the SURF features of 2D gray images to carry out the refined recognition. Finally, our proposed work will increase the robustness in expression variations.

Face recognition is still an open problem. Many 2D face recognition approaches came into light to achieve high recognition rate. But these approaches are still challenged by the changes in illuminations, expressions, pose, noise, etc. A 3D face recognition technique is proposed to overcome such challenges and to enhance robustness to expression variations. Here, we compare the person at different age groups with higher recognition rate in comparison to 2D face recognition techniques. We propose a two stage procedure of 3D face recognition based on FLD (Fisher Linear Discriminant), SURF operator and depth-image. First, FLD is used on depth-image to perform recognition and then the SURF features of 2D gray images to carry out the refined recognition. Finally, our proposed work will increase the robustness in expression variations.

During the period of the grant, 2/1/93 - 1/15/95, we developed: (1) a Bayesian framework for object detection and tracking; the algorithm was successfully tested on real-data in the detection and tracking of vehicles on a highway: (2) a recognition algorithm based on stochastic hierarchical, context-free-grammars type, object representation; the study has required the development of feasible pruning techniques for dynamic programming; (3) a new acoustic model for speech recognition based on a wavelet representation of the acoustic-signal, and nonparametric prediction techniques.

During the period of the grant, 2/1/93 - 1/15/95, we developed: (1) a Bayesian framework for object detection and tracking; the algorithm was successfully tested on real-data in the detection and tracking of vehicles on a highway: (2) a recognition algorithm based on stochastic hierarchical, context-free-grammars type, object representation; the study has required the development of feasible pruning techniques for dynamic programming; (3) a new acoustic model for speech recognition based on a wavelet representation of the acoustic-signal, and nonparametric prediction techniques.

Several papers addressing image analysis and pattern recognition techniques for satellite imagery are presented. Texture classification, image rectification and registration, spatial parameter estimation, and surface fitting are discussed.

Several papers addressing image analysis and pattern recognition techniques for satellite imagery are presented. Texture classification, image rectification and registration, spatial parameter estimation, and surface fitting are discussed.

Several papers addressing image analysis and pattern recognition techniques for satellite imagery are presented. Texture classification, image rectification and registration, spatial parameter estimation, and surface fitting are discussed.