scispace - formally typeset
Search or ask a question

Showing papers in "Graphical Models \/graphical Models and Image Processing \/computer Vision, Graphics, and Image Processing in 1988"


Journal ArticleDOI
TL;DR: This paper presents a survey of thresholding techniques and attempts to evaluate the performance of some automatic global thresholding methods using the criterion functions such as uniformity and shape measures.
Abstract: In digital image processing, thresholding is a well-known technique for image segmentation. Because of its wide applicability to other areas of the digital image processing, quite a number of thresholding methods have been proposed over the years. In this paper, we present a survey of thresholding techniques and update the earlier survey work by Weszka (Comput. Vision Graphics & Image Process 7, 1978 , 259–265) and Fu and Mu (Pattern Recognit. 13, 1981 , 3–16). We attempt to evaluate the performance of some automatic global thresholding methods using the criterion functions such as uniformity and shape measures. The evaluation is based on some real world images.

2,771 citations


Journal ArticleDOI
TL;DR: This survey will provide a useful guide to quickly acquaint researchers with the main literature in this research area and it seems likely that the Hough transform will be an increasingly used technique.
Abstract: We present a comprehensive review of the Hough transform, HT, in image processing and computer vision. It has long been recognized as a technique of almost unique promise for shape and motion analysis in images containing noisy, missing, and extraneous data but its adoption has been slow due to its computational and storage complexity and the lack of a detailed understanding of its properties. However, in recent years much progress has been made in these areas. In this review we discuss ideas for the efficient implementation of the HT and present results on the analytic and empirical performance of various methods. We also report the relationship of Hough methods and other transforms and consider applications in which the HT has been used. It seems likely that the HT will be an increasingly used technique and we hope that this survey will provide a useful guide to quickly acquaint researchers with the main literature in this research area.

2,099 citations


Journal ArticleDOI
TL;DR: A new solution to the problem of constructing a shape over a set of cross-sectional contours that can handle the case where the number of contours varies from one cross section to the other and thus, can handle contour-defined objects which are branched and have holes.
Abstract: We propose a new solution to the problem of constructing a shape over a set of cross-sectional contours. We construct a volume whose boundary is a polyhedron with triangular faces intersecting the cutting planes along the given contours. This volume is obtained, slice by slice, by pruning the Delaunay triangulations formed between two adjacent cross sections. The Delaunay triangulation can be computed efficiently in this specific situation: an algorithm is presented that is optimal with respect to the input size and the output size and that only uses 2-dimensional operations. The shape reconstruction method extracts, from that triangulation, a maximum solid with no singularity. It can handle the case where the number of contours varies from one cross section to the other and thus, can handle contour-defined objects which are branched and have holes. This algorithm has been fully implemented and tested on various examples; results on medical images are presented.

500 citations


Journal ArticleDOI
TL;DR: Detecting building structures in aerial images by using a generic model of the shapes of the structures looking for — that they are rectangular or composed of rectangular components to confirm their presence and to estimate their height.
Abstract: Detecting building structures in aerial images is a task of importance for many applications. Low-level segmentation rarely gives a complete outline of the desired structures. We use a generic model of the shapes of the structures we are looking for — that they are rectangular or composed of rectangular components. We also use shadows cast by buildings to confirm their presence and to estimate their height. Our techniques have been tested on images with density typical of suburban areas.

322 citations


Journal ArticleDOI
TL;DR: A modified formulation of this “oriented smoothness constraint” will be compared to an earlier one, and the iterative solution approach could be improved significantly using multigrid methods.
Abstract: Optical flow fields are useful to describe the motion of objects relative to the observer of a scene. Smoothness requirements facilitate the estimation of optical flow fields even for areas with constant or only linearly sloping gray value distributions. Horn and Schunck (Artif. Intell. 17, 1981, 185–203) formulated the estimation of optical flow fields as a minimization problem. In order to avoid problems with the general smoothness constraint of Horn and Schunck (op. cit.), Nagel (in Proceedings, IJCAI, Karlsruhe, Aug. 1983, pp. 945–951) developed the “oriented smoothness constraint.” Here, a modified formulation of this “oriented smoothness constraint” will be compared to an earlier one. The differences between these formulations will be discussed analytically. Starting with an iterative solution approach for the system of nonlinear partial differential equations resulting from the “oriented smoothness constraint,” the iterative solution approach could be improved significantly using multigrid methods. Results obtained from experiments with real world image sequences will be presented.

258 citations


Journal ArticleDOI
TL;DR: In this article, the Torrance-Sparrow specular model is used to derive the relationship between the properties of a specular feature in an image and local properties of the corresponding surface.
Abstract: We show that highlights in images of objects with specularly reflecting surfaces provide significant information about the surfaces which generate them. A brief survey is given of specular reflectance models which have been used in computer vision and graphics. For our work, we adopt the Torrance-Sparrow specular model which, unlike most previous models, considers the underlying physics of specular reflection from rough surfaces. From this model we derive powerful relationships between the properties of a specular feature in an image and local properties of the corresponding surface. We show how this analysis can be used for both prediction and interpretation in a vision system. A shape from specularity system has been implemented to test our approach. The performance of the system is demonstrated by careful experiments with specularly reflecting objects.

230 citations


Journal ArticleDOI
TL;DR: This bibliography lists journal publications, conference papers, research technical reports, and articles from trade journals on automated visual inspection for industry which were published during the years from 1981 to 1987 to provide a complete bibliography for those interested in automatedVisual inspection.
Abstract: This bibliography lists journal publications, conference papers, research technical reports, and articles from trade journals on automated visual inspection for industry which were published during the years from 1981 to 1987. More than 600 references are included. References are organized into 13 categories according to subject matter. The categories are (1) books, (2) conferences and workshops, (3) general discussions and surveys, (4) inspection of printed circuit patterns, (5) inspection of solder joints, (6) inspection of microcircuit photomasks, (7) inspection of integrated circuits and hybrids, (8) inspection of other electrical and electronics components, (9) surface inspection, (10) X-ray inspection, (11) other inspection applications, (12) system components, and (13) inspection algorithms. References listed in each category are arranged in chronological order. The purpose is primarily to provide a complete bibliography for those interested in automated visual inspection. Some general observations have been made for the above areas of activity summarizing the advances that have taken place and the problems that remain to be solved.

168 citations


Journal ArticleDOI
TL;DR: An algorithm for the estimation of rigid body motion using straight line correspondences using straightline correspondences is presented and it is found that the rotation and the translation parts are separable.
Abstract: An algorithm for the estimation of rigid body motion using straight line correspondences is presented in this paper. In the case of pure translation, we present a linear algorithm using 5 line correspondences over 3 frames. In the case of general motion, it is found that the rotation and the translation parts are separable. The rotation part can be computed by the iterative solution of nonlinear equations based on 6 or more line correspondences over 3 frames. After the rotation is found, the translation part is determined just as in the pure translation case. For the special case of constant rotation, the convergence range of the iterative method is wide enough so that global search can be used to estimate the rotation matrix. However, for the case of variable rotation, global search appears computationally infeasible at present.

160 citations


Journal ArticleDOI
TL;DR: A new method is presented for the recovery of a bivariate function H(x, y) that describes a “nice,” almost everywhere differentiable height profile, from shading information, based on a recursive way of determining equal-height or level contours of the surface starting at a given level curve.
Abstract: We present a new method for the recovery of a bivariate function H(x, y) that describes a “nice,” almost everywhere differentiable height profile, from shading information. The given shading data is assumed to be a result of diffuse, Lambertian reflection of light from the surface. This implies that, if the scene is uniformly illuminated from above, the shading yields information on the cosine of the angle between the vertical and the surface normal at each point. Given the shading information in the plane, the shape from shading problem is to determine all height profiles consistent with the data, and some boundary conditions, such as points of known height and surface orientation, or height profiles along continuous curves in the image plane. The new shape-from-shading method that we discuss is based on a recursive way of determining equal-height or level contours of the surface starting at a given level curve.

155 citations


Journal ArticleDOI
TL;DR: Two classes of point location problems found in visual navigation of a mobile robot are considered, finding the location of a robot using a map of the room where the robot moves and an image taken by a camera carried by the robot.
Abstract: The paper considers two classes of point location problems found in visual navigation of a mobile robot. The problems we consider are finding the location of a robot using a map of the room where the robot moves and an image taken by a camera carried by the robot. In the first class of problems, vertical edges in the image are given, and a possible location for the robot is investigated by establishing a correspondence between the edges in the images and vertical poles given in the map. In the second class of problems, the possible region for the robot is investigated under the assumption that vertical edges are distinguishable from each other, but only the order in which the edges are found when the image is swept from left to right is given. These problems and their variations are considered from a computational geometry point of view, and efficient algorithms for solving them are given.

151 citations


Journal ArticleDOI
TL;DR: This paper divides these algorithms related to octrees into two categories, construction and manipulation of octrees, and gives a detailed survey of them.
Abstract: The octree representation of three-dimensional objects based on the principle of recursive subdivision is a generalization of two-dimensional quadtrees. It has been studied for use in many application areas such as solid modeling, computer graphics, computer-aided design/manufacturing, computer vision, image processing, and robotics. Many algorithms related to octrees have been developed in these application areas. In this paper, we divide these algorithms into two categories, construction and manipulation of octrees, and give a detailed survey of them.

Journal ArticleDOI
TL;DR: A simple method for computing the general position based on the line data is presented and the algorithms of computing the most general eye position by solving this maximin problem are presented.
Abstract: In 3D graphics, the selection of the location of the center of projection (eye) is very important in order to obtain the pictures by which the original shapes of 3D objects can be comprehended easily. The eye position from which the picture of 3D objects with the maximum shape information is obtained is called the “general position”. In the current practice of making pictures, the most general eye position is determined in ad hoc ways, because it is very difficult to determine the view, foreseeing the resultant display image. In this paper, a simple method for computing the general position is presented. First the general position problem based on the line data is formulated as a maximin problem. Then the algorithms of computing the most general eye position by solving this maximin problem are presented. Some examples of performing these algorithms yielding satisfactory results are also presented.

Journal ArticleDOI
TL;DR: A technique called edge matching is described, which provides a direct method of adjusting multiple thresholds so that the edges of the thresholded image closely match the edgesOf the original grey-tone image, which preserves the shape and geometrical structure of the objects in the image.
Abstract: Thresholding is an effective method for simplifying images of solid objects on a solid background. Multiple thresholds are needed when thresholding images containing several objects of different brightness or reflectivity. Also, to account for variations of grey levels due to nonuniform illumination, it is often necessary to allow the thresholds to vary across the image. We describe a technique called edge matching , which provides a direct method of adjusting multiple thresholds so that the edges of the thresholded image closely match the edges of the original grey-tone image. The result is a threshold image that preserves the shape and geometrical structure of the objects in the image.

Journal ArticleDOI
Gary W. Meyer1
TL;DR: Gaussian quadrature with the opponent fundamentals as weighting functions is used to choose the wavelengths of wavelengths for synthetic image generation, and is shown to be superior to using Gaussian Quadratures with the fundamental spectral sensitivity functions or the CIE XYZ matching functions.
Abstract: The efficient synthesis of color in computer graphics is dependent on modelling the correct number and spacing of wavelengths across the visible spectrum. It has recently been shown that the opponent representation of the fundamental spectral sensitivity functions is optimal from the point of view of statistical communication theory. This result is used in this paper to guide the selection of wavelengths for synthetic image generation. Gaussian quadrature with the opponent fundamentals as weighting functions is used to choose the wavelengths. This approach is shown to be superior to using Gaussian quadrature with the fundamental spectral sensitivity functions or the CIE XYZ matching functions. The technique is evaluated by using color difference calculations and by comparisons between a real scene and a computer generated picture of that scene.

Journal ArticleDOI
TL;DR: A criterion is introduced for identifying the local maxima in a specific weighted distance transform, and the reverse weighted distance transformation is introduced to associate a disc with every labeled pixel.
Abstract: Weighted distance transforms computed by using suitable integer weights are an adequate approximation to the Euclidean distance transform. In this paper, a criterion is introduced for identifying the local maxima in a specific weighted distance transform. The reverse weighted distance transformation is also introduced to associate a disc with every labeled pixel, and the discs centered on the local maxima are proved to be maximal discs.

Journal ArticleDOI
TL;DR: By generalizing Blinn's “blobby” approach to modeling complex objects, this work construct single equations for nonconvex, composite shapes starting with the basic convex primitives.
Abstract: We present a new approach to the problem of modeling smoothly deformable shapes with convex polyhedral bounds. Our hyperquadric modeling primitives, which include superquadrics as a special case, can be viewed as hyperplanar slices of deformed hyperspheres. As the original hypersphere is deformed to its bounding hypercube, the slices undergo corresponding smooth deformations to convex polytopes. The possible shape classes include arbitrary convex polygons and polyhedra, as well as taperings and distortions that are not naturally included within the conventional superquadric framework. By generalizing Blinn's “blobby” approach to modeling complex objects, we construct single equations for nonconvex, composite shapes starting with our basic convex primitives. Hyperquadrics are of potential interest for the generation of synthetic images, for automated image interpretation and for psychological models of geometric shape representation, manipulation, and perception.

Journal ArticleDOI
TL;DR: Proofs are presented which show that under a few, easily realizable restrictions, there exists a geometrical image transformation that produces a change in a density image that is identical to the change that would be produced in the image by the motion of the objects being imaged.
Abstract: Proofs are presented which show that under a few, easily realizable restrictions, there exists a geometrical image transformation that produces a change in a density image that is identical to the change that would be produced in the image by the motion of the objects being imaged. Both two and three-dimensional images of three-dimensional scenes are considered subject to restrictions that are appropriate to conventional radiography, computerized axial tomography, gamma-ray scintigraphy, and magnetic resonance imaging. There is no restriction on the motion of the objects being imaged except that they behave as a conserved medium. There is no restriction on formation of the image except that it include convolution with a differentiable point-function. The convolution can be the result of blurring inherent in the acquisition of the image or of explicit image processing. The results have special significance with regard to the problem of motion artifacts in digital subtraction angiography. The case of incompressible flow is considered and comparisons are made with problems in optical flow.

Journal ArticleDOI
TL;DR: In this new depth recovery method, an arbitrary depth must be first preset for a point somewhere in the image, then path-independent line integrals are computed to get the relative depths at every pont in theimage.
Abstract: This paper describes a new method for constructing a depth map from surface normals. In this new depth recovery method, an arbitrary depth must be first preset for a point somewhere in the image, then path-independent line integrals are computed to get the relative depths at every pont in the image. The validity of the proposed method is discussed and its efficiency was tested using surface normals obtained by shape from a shading algorithm. The comparison of the new method to the previous ones is given. Theoretical analysis and experimental results show that our method is both powerful and easy to implement.

Journal ArticleDOI
TL;DR: The results indicate that the method works well for a variety of images and pair types (e.g., motion, stereo), and the speed of the implementation on a small computer demonstrates that it is practical for many applications.
Abstract: The paper presents a new approach to match straight line segments extracted from image pairs. Important information can be obtained from the location of a pair of corresponding lines, such as distance from the camera and relative movement. Other line matching methods rely primarily on correspondence constraints to determine matches. The algorithm described here takes a unique approach. It relies on parameter values attached to each line that describe the regions in the image from which the lines are extracted. The values are compared to find similar lines in image pairs. A match function is defined that determines the strength of the similarity between a pair of lines. Pairs which are a mutually best match, based on the match function, are considered corresponding lines. The implementation and testing of the algorithm are also described. The results indicate that the method works well for a variety of images and pair types (e.g., motion, stereo). The speed of the implementation on a small computer also demonstrates that it is practical for many applications.

Journal ArticleDOI
TL;DR: It is suggested that both filtering and edge detection should take place at the same time, and representation of the neighborhood by its mean and variance can be generalized by Haralick's sloped-facet model, which has a more complete characterization of the local changes of intensities.
Abstract: The conventional way of edge detection is to first filter the image and then use simple techniques to detect edges. However, filtering the noise will also blur the edges since edges correspond to the high frequencies. Our suggestion is that both filtering and edge detection should take place at the same time. The way of doing this is by statistical theory of hypothesis testing. A simple form of decision rule is derived and the generalization of this result to more complicated situations is also discussed in detail. The decision rule can make a decision whether in a given small neighborhood there is an edge, or a line, or a point, or a corner edge, or just a smooth region. During the computation of the decision rule, the by-products are the mean and variance of the neighborhood and these can be used for split and merge analysis. Calculation of the mean acts as filtering of the neighborhood pixels. In fact, representation of the neighborhood by its mean and variance can be generalized by Haralick's sloped-facet model, which has a more complete characterization of the local changes of intensities.

Journal ArticleDOI
TL;DR: Several methods of Hough transform computation suitable for implementation on a mesh-connected SIMD parallel processor, such as Goddard Space Flight Center's Massively Parallel Processor (MPP) or Martin Marietta Corp.'s Geometric Arithmetic Parallel processor (GAPP), are compared.
Abstract: Hough transform techniques for straight line detection play a key role in the road following algorithms developed by the University of Maryland for the DARPA Autonomous Land Vehicle Project. This report compares several methods of Hough transform computation suitable for implementation on a mesh-connected SIMD parallel processor, such as Goddard Space Flight Center's Massively Parallel Processor (MPP) or Martin Marietta Corp.'s Geometric Arithmetic Parallel Processor (GAPP).

Journal ArticleDOI
TL;DR: The approach here is first to transform images of line segments to the center of the image plane as if the camera were rotated to aim at them, and the 3D information extracted in this canonical position is transformed back to the original configuration.
Abstract: Given a perspective projection of line segments on the image plane, the constraints on their 3D positions and orientations are derived on the assumption that their true lengths or the true angles they make are known. The approach here is first to transform images of line segments to the center of the image plane as if the camera were rotated to aim at them. The 3D information extracted in this canonical position is then transformed back to the original configuration. Examples are given, by using real images, for analytical 3D recovery of a rectangular corner and a corner with two right angles.

Journal ArticleDOI
TL;DR: A new algorithm for converting a binary image into chain codes using its run-length codes that has an advantage that it can detect easily the inclusion relationship between boundaries at the same time as chain-code generation.
Abstract: This paper presents a new algorithm for converting a binary image into chain codes using its run-length codes. The basic idea of conventional chain-coding algorithm is to follow boundary pixels by convolving a 3 × 3 window with the image and to sequentially generate chain codes. The proposed algorithm has two phases, namely run-length coding and chain-code generation. We use connectivity information between runs as well as their coordinates in the phase of run-length coding. In the second phase (chain-code generation) the connectivity information extracted in the first phase is utilized for sequentially tracking runs containing the boundary pixels to be followed. This algorithm has an advantage that we can detect easily the inclusion relationship between boundaries at the same time as chain-code generation.

Journal ArticleDOI
TL;DR: Using two new shape operators called Minkowski addition and decomposition operators, a simple shape model is presented and mathematical characteristics of these operators are explored in some detail, with the aim of eventually arriving at a formal theory of shape description.
Abstract: Using two new shape operators called Minkowski addition and decomposition operators, a simple shape model is presented. Mathematical characteristics of these operators are explored in some detail, with the aim of eventually arriving at a formal theory of shape description. A few application areas of the shape model, particularly some important uses of the shape operators are briefly mentioned.

Journal ArticleDOI
Henry S. Baird1
TL;DR: Large-scale statistically-significant trials, in the context of a mixed-font, variable-size optical character recognition (OCR) system, have shown that the technique is superior to simpler, fixed mappings, and is effective in generalizing common characteristics in mixtures of fonts.
Abstract: A general technique for combining the strengths of structural shape analysis with statistical classification is proposed. The approach is to construct a function, called a feature identification mapping, from the representation generated by structural analysis to the one required for statistical classification. It is shown that if a certain continuity property holds for the parameterizations of the structural shape types, then it is possible to infer the mapping automatically. Inference is slow and heuristic, but is highly automated, controlled by only a few statistical parameters, and is applicable uniformly to all shape types. In addition, if the shape types are sufficiently elementary, the resulting mapping can be computed quickly using kD-trees. Large-scale statistically-significant trials, in the context of a mixed-font, variable-size optical character recognition (OCR) system, have shown that the technique is superior to simpler, fixed mappings, and is effective in generalizing common characteristics in mixtures of fonts.

Journal ArticleDOI
TL;DR: This paper presents a linear algorithm for determining 3D motion parameters of a rigid object based on straight line correspondences that is uniquely determined by the physical constraints of 3D rotations and translations.
Abstract: This paper presents a linear algorithm for determining 3D motion parameters of a rigid object based on straight line correspondences. The algorithm requires a minimum number of thirteen line correspondences over three frames. It includes three steps: first, three intermediate matrices are computed; then, several candidate solutions of the rotation matrices and translation vectors are obtained from the intermediate matrices; finally, motion parameters are uniquely determined by the physical constraints of 3D rotations and translations. Some simulation results are also given.

Journal ArticleDOI
TL;DR: A new structural approach based on the Hough method of line detection is introduced,based on the relative orientation and location of the lines within the texture, which is independent of geometrical transformations such as rotation, translation, and/or scaling.
Abstract: Texture is one of the important image characteristics and is used to identify objects or regions of interest. The problem of texture classification has been widely studied. Texture classification techniques are either statistical or structural. Some statistical texture classification approaches use Fourier power-spectrum features, while others are based on first- and second-order statistics of gray level differences. Periodic textures that consist of mostly straight lines are of particular interest. In this paper, a new structural approach based on the Hough method of line detection is introduced. This classification is based on the relative orientation and location of the lines within the texture. With proper normalization, the classification is independent of geometrical transformations such as rotation, translation, and/or scaling. Experimental results will also be presented.

Journal ArticleDOI
TL;DR: A linear algorithm is developed to determine the mode of motion, the instantaneous rotation velocity, the direction of the instantaneous translation velocity, and the relative depth map (or surface structure) under the rank assumption.
Abstract: The purpose of the article is to establish the relationship between rigid body motion and the optic flow image and solve motion parameters from optic flow image points A basic equation relating optic flow image points to rigid body motion which involves only the instantaneous rotation and translation velocities without depths is established A linear algorithm is developed to determine the mode of motion (whether the instantaneous translation is zero or not), the instantaneous rotation velocity, the direction of the instantaneous translation velocity, and the relative depth map (or surface structure) under the rank assumption The algorithm represents a simplification to the linear optic flow-motion algorithm proposed in ( Zhuang and Haralick , in Proceedings, IEEE First Conf on Artificial Intelligence Applications, Denver, CO, 1984, pp 366–375)

Journal ArticleDOI
TL;DR: It is shown that spatial stability analysis leads to an edge detection scheme with good noise resilience characteristics and that it can lead to improvements in “shape from texture” methods.
Abstract: The scale-space S(x, σ) of a signal I(x) is defined as the space of the zero-crossings from {∇2G(σ)* I(x)}, where G is a Gaussian filter. We present a new method for parsing scale-space, spatial stability analysis, that allows the localization of region boundaries from scale space. Spatial stability analysis is based on the observation that zero-crossings of region boundaries remain spatially stable over changes in filter scale. It is shown that spatial stability analysis leads to an edge detection scheme with good noise resilience characteristics and that it can lead to improvements in “shape from texture” methods.

Journal ArticleDOI
TL;DR: The influence of using distinct optimization criteria for determining the coefficients of a distance transform is studied, and emphasis is given to isotropy, or invariance with respect to rotation, and the use of unbiased distance estimates.
Abstract: The influence of using distinct optimization criteria for determining the coefficients of a distance transform is studied. The criteria studied are (1) minimizing the maximum of the absolute value of the difference between the distance transform and Euclidaan distance, and (2) minimizing the root-mean-square difference between the distance transform and Euclidean distance. By allowing an overall scaling factor to have other than integer values, other integer approximations of the distance transform's coefficients result as optimal. Emphasis is given to isotropy, or invariance with respect to rotation, and to the use of unbiased distance estimates.