Curvilinear targets are common in many imaging modalities. Detection of such targets can be challenging because of
their multiscale structure, their frequent obscuration in natural imagery, their turns, intersections, and merges, and the
prevalence of false positive detections based on local information. Using a spatial spectroscopy approach, we introduce
image analysis methods that use the concept of gauge frames to simplify the identification of curvilinear targets. Fast
computational approximation methods are described for gauge fields, and an experiment is described illustrating the
power of higher-order derivatives for understanding even relatively simple geometric structures. Methods for extracting
coherent curvilinear objects that exploit the larger-scale commonalities of points in the object are described.
Military operations in urban areas often require detailed knowledge of the location and identity of commonly occurring
objects and spatial features. The ability to rapidly acquire and reason over urban scenes is critically important to such
tasks as mission and route planning, visibility prediction, communications simulation, target recognition, and inference
of higher-level form and function. Under DARPA's Urban Reasoning and Geospatial ExploitatioN Technology
(URGENT) Program, the BAE Systems team has developed a system that combines a suite of complementary feature
extraction and matching algorithms with higher-level inference and contextual reasoning to detect, segment, and classify
urban entities of interest in a fully automated fashion. Our system operates solely on colored 3D point clouds, and
considers object categories with a wide range of specificity (fire hydrants, windows, parking lots), scale (street lights,
roads, buildings, forests), and shape (compact shapes, extended regions, terrain). As no single method can recognize the
diverse set of categories under consideration, we have integrated multiple state-of-the-art technologies that couple
hierarchical associative reasoning with robust computer vision and machine learning techniques. Our solution leverages
contextual cues and evidence propagation from features to objects to scenes in order to exploit the combined descriptive
power of 3D shape, appearance, and learned inter-object spatial relationships. The result is a set of tools designed to
significantly enhance the productivity of analysts in exploiting emerging 3D data sources.
This study is an initial investigation into the efficacy of texture operators for detection of military vehicle targets-in-the-clear in SAR imagery. The specific study is a very simple problem that aims to evaluate a particular feature set that arises in an approach to computer vision called spatial spectroscopy. Spatial spectroscopy begins by partitioning the image's spatial (Fourier) spectrum using a bank of filters. The filters compute a multiscale, truncated Taylor Series expansion at each pixel. Suitably extended on generic images, this feature space is capable of producing a unique pattern describing each pixel. The objective, of course, is not to uniquely distinguish each pixel but to form groups of pixels corresponding to targets in SAR that are distinct from background pixels. Thus, nonlinear operators are required to fold, twist, and bend the feature space in ways that cause pixels that make up targets to group together. The particular nonlinear operators for a study depend on the invariances and equivariances of the problem. In the present case, a large suite of operators is applied to the image data and principal discriminant analysis is used to select the most relevant features. Texture operators are found to be effective at discriminating targets from background.
There are two basic principles associated with multiscale geometry: 1) geometry involves analysis that is invariant to certain spatial transformations, including translation, rotation, and zoom, and 2) the dimension of scale is as critical as the dimensions giving spatial position, corresponding intuitively to level of detail in image space. Considered together, image space and scale is called scale space. Three families of methods based on these principles are achieving impressive results in image analysis —particularlyin their insensitivity to irrelevant detail (including image noise) and intensity blurring, and in their ability to produce stable object descriptions and pixel classifications into objects. The three families are multiscale medial axis or core-based analysis (CBA), variable conductance difftision (VCD), and multiscale geometric statistical pattern recognition (MGSPR). This pair of tutorials (morning and afternoon) covered the basic mathematics of multiscale geometry as well as all three of these families of methods. It included algorithms for computation and also illustrative results of the applications of the methods to both 2D and 3D medical images of various modalities. The morning tutorial covered the mathematics of diffusion and scale space and the definition, effect on scale space geometry, and application of cores. The afternoon tutorial covered the mathematics, algorithms, and applications of variable conductance diffusion, including approaches involving MGSPR, and it covered algorithms for segmenting objects both via VCD and via CBA.
Inhomogeneities in the fields of magnetic resonance (MR) systems cause the statistical characteristics of tissue classes to vary within the resulting MR images. These inhomogeneities must be taken into consideration when designing an algorithm for automated tissue classification. The traditional approach in image processing would be to apply a gain field correction technique to remove the inhomogeneities from the images. Statistical solutions would most likely focus on including spatial information in the feature space of the classifier so that it can be trained to model and adjust for the inhomogeneities. This paper will prove that neither of these general approaches offer a complete and viable solution. This paper will prove that neither of these general approaches offers a complete and viable solution. This paper will in fact show that not only do the inhomogeneities modify the local mean and variance of a tissue class as is commonly accepted, but the inhomogeneities also induce a rotation of the covariance matrices. As a result, gain field correction techniques cannot compensate for all of the artifacts associated with inhomogeneities. Additionally, it will be demonstrated that while statistical methods can capture all of the anomalies, the across patient and across time variations of the inhomogeneities necessitate frequent and time consuming retraining of any Bayesian classifier. This paper introduces a two stage process for MR tissue classification which addresses both of these issues by utilizing techniques from both image processing and statistics. First, a band-pass mean field corrector is used to alleviate the mean and variance deformations in each image. Then, using a kernel mixture model classifier couple to an interactive data augmentation tool, the user can selectively refine and explore the class representations for localized regions of the image and thereby capture the rotation of the covariance matrices. This approach is shown to outperform Gaussian classifiers and 4D mixture modeling techniques when both the final accuracy and user time requirements are considered.
Object-Oriented Programming is enabled by an advance in compiler technology and programming language design supporting encapsulation and inheritance. This technical adjustment has had a surprisingly broad impact on strategies for design and development of software. The methods for employing Object-Oriented Programming in software development are called Object-Oriented Design. This paper explains what Object-Oriented Programming is, why it has attracted so much interest, and then critically examines its potential impact.
A methodology for task-sensitive pixel classification is defined based on multiscale Gaussian derivatives and statistical pattern recognition methods. Multiscale Gaussian derivatives are approximated by Gaussian and offset-Gaussian filters to decrease computational requirements. A method is devised for computing a discriminant vector between classes based on class isolation and compactness. The optimal discriminant vector is converted back into image form and applied to the image to determine whether a 1-D feature space is adequate to separate the classes.
Definition of objects in medical images requires a multiscale approach because important structure appears across a wide range of scales. Object boundaries, when they are required, must be inferred from the multiscale structure of the image and a priori knowledge. For many objectbased tasks, explicit identification of boundaries is not necessary. Instead, it is possible to base object measures on medial axes and their radius functions obtained using statistical methods. A medial approach makes the easy decisions about the membership of pixels in the object first. The difficult decisions about the boundaries are made using a fuzzy measure of "objectness" that can account for edge uncertainty, partial volume effects, and a priori information. Objectness diffuses outward from the medial axis, and non-objectness diffuses inward from medial axes of surrounding regions. Their competition in boundary regions defines objectness even in the absence of an edge. The area of an object is the integral of objectness across space. Statistical pattern recognition methods (supervised and unsupervised classification; linear projections) are used to identiQy medial axes in a feature space defined by multiscale Gaussian filters. The pattern describing a pixel is formed from the response at that location and nearby locations to the filters. Approximations to derivatives of Gaussians are linear subspaces of this feature space.
Image pattern recognition involves decision-making based on image data. Statistical tools for automatic, rational decision-making are numerous and can be applied some of the kinds of decisions that need to be made in medical imaging. This tutorial will present an introduction to statistical pattern recognition and show how those techniques can be applied to several kinds of image analysis problems, including analysis of multiple modalities and multiple scales. The course introduction will review problems in biomedical image analysis and provide both a framework and taxonomy for approaching vision problems. The course will include detailed discussion on the structure of images, statistical pattern recognition techniques, image pattern recognition using this technique, and statistical representations of image geometry.
This artificial visual system (AVS) is a computational framework for computer vision based on spatial filtering and statistical pattern recognition. Computer vision tasks are often poorly defined; the AVS clarifies the kinds of visual tasks that can be defined and what constitutes a well-defined task. `Segmentation'' is not a well-defined task. Edge detection is revealed to be an absurd task. A filter set composed of multiscale Gaussians alone captures the structure of Koenderink''s generic neighborhood operators when a pattern is constructed from the responses at a pixel and neighboring locations, where the distance to the selected neighbors increases with larger scale. Prior studies of the feature space formed by multiscale Gaussians reveal surprising power in the multiscale Gaussians alone. New studies support this observation. Contrary to common belief, we show how nonlocal, spatial, geometric structure can be captured using statistical pattern recognition operations in the AVS framework. A procedure is defined for deriving a single composite filter providing optimal separation of two clusters in feature space.
The computation of structural descriptions of objects in images can contribute to many image analysis tasks including measurement, registration, and object formation and identification. Edge-based structural descriptions fail to support the creation of coherent objects. Medial descriptions provide better support for object formation and measurement. We demonstrate an artificial visual system that uses outputs of Gaussian derivative filters to infer a multiscale medial axis (MMA) in 2-D grayscale images. Properties of the MMA are illustrated.
Image segmentation involves labelling pixels according to their membership in image regions. This requires that we understand what a region is. Using supervised pixel classification, we investigate how groups of pixels labelled manually according to perceived image semantics map onto the feature space created by an Artificial Visual System. We investigate multiscale structure of regions and show that pixels form clusters based on their geometric roles in the image intensity function, not by image semantics. A tentative abstract definition of a "region" is proposed based on this behavior.