Showing papers on "3D reconstruction published in 1988"

Projective invariants of shapes

Abstract: A major goal of computer vision is object recognition, which involves matching of images of an object, obtained from different, unknown points of view. Since there are infinitely many points of view, one is faced with the problem of a search in a multidimensional parameter space. A related problem is the stereo reconstruction of 3-D surfaces from multiple 2-d images. The author proposes to solve these fundamental problems by using geometrical properties of the visible shape that are invariant to a change in the point of view. To obtain such invariants, he starts from classical theories for differential and algebraic invariants not previously used in image understanding. As they stand, these theories are not directly applicable to vision. He suggests extensions and adaptations of these methods to the needs of machine vision. He then studies general projective transformations, which include both perspective and orthographic projections as special cases. >

Computer vision: basic principles

Abstract: The author provides a general introduction to computer vision. He discusses basic techniques and computer implementations, and also indicates areas in which further research is needed. He focuses on two-dimensional object recognition, i.e. recognition of an object whose spatial orientation, relative to the viewing direction is known. >

3D reconstruction: the registration problem

Abstract: Three dimensional (3D) reconstruction of serial sections through biological tissue is a very important method for visualizing and quantifying the 3D relationships between internal structures. This project involves the automation of the 3D reconstruction process, focusing on the problem of registering or aligning successive serial sections. Serial sections can be misregistered with respect to rotation, translation and scaling (if magnification is not constant). A linear least squares fitting procedure is developed to determine the required registration transformation based on a set of labeled feature points. The results obtained from using two different types of feature points are evaluated and compared: (1) manually selected feature points and (2) characteristic shape points computed from the contours which describe specific shape properties (e.g., centroid and radius weighted mean point). These results demonstrate that good registration accuracy is obtained using both kinds of feature points under conditions in which the contours do not have exact matches.

Automated 3D reconstruction of tree-like structures from two orthogonal views

Abstract: The common use of orthogonal X-ray images for the diagnosis of vascular abnormalities provides a natural motivation for three dimensional reconstruction from two perpendicular views. Reconstruction from two projections is a problem with a non-unique solution. Ambiguity in the 3D reconstruction has previously been resolved by manually matching feature points between the two views. In this study, a priori knowledge of the object's tree-like structure was incorporated into a rule-base in order to help automate the reconstruction process. The procedure for reconstruction of a tree-like structure from two orthogonal images contains three stages: extraction of the skeletal knowledge in each image, determination of the correspondence of tree segments between views by application of a heuristic rule-base, and algorithmic 3D reconstruction from the corresponding segment pairs. >

A new stereo vision model based upon the Binocular fusion concept

Abstract: In computer vision, the idea of using stereo cameras for depth perception has been motivated by the fact that in human vision one percept can arise from two retinal images as a result of the process called “fusion”. Nevertheless, most of the stereo algorithms are generally concerned with finding a solution to obtaining depth and three-dimensional shape irrespective of its relevance to the human system. Recent progress in the study of the brain mechanisms of vision has opened new vistas in computer vision research. This paper investigates this knowledge base and its applicability to improving the technique of computer stereo vision. In this regard, (1) a stereo vision model in conjunction with evidences from neurophysiology of the human binocular system is established herein; (2) a computationally efficient algorithm to implement this model is developed. This algorithm has been tested on both computer generated and real scene images. The results from all directional subimages are combined to obtain a complete description of the target surface from disparity measurements.

Some computational aspects of low-level computer vision

Abstract: Some computational aspects of low-level computer vision are addressed, discussing the following example problems: 2-1/2 dimensional sketch, shape from shading, and optical flow. The existence and uniqueness of the solutions, schemes for optimal solutions, reliable and efficient algorithms for the computation, discretization errors, and coping with discontinuities are studied. >

Image analysis and computer vision: 1987

Abstract: This paper presents a bibliography of over 1400 references related to computer vision and image analysis, arranged by subject matter. The topics covered include architectures; computational techniques; feature detection, segmentation, and image analysis; matching, stereo, and time-varying imagery; shape and geometry; color and texture; and three-dimensional scene analysis. A few references are also given on related topics, such as computer graphics, image input/output, image processing, optical processing, neural nets, visual perception, pattern recognition, and artificial intelligence.

Fast Automatic Recognition and 3D Reconstruction of the Coronary Tree from DSA-Projection Pairs

Abstract: For a reliable quantitative assessment of vascular lesions the three-dimensional orientations and connections have to be known accurately. This information can be extracted systematically from two DSA-projections using a digital image processing system. Automatic identification and calculation of three-dimensional coordinates of the vessel centrelines are possible at angles of 15–45° between the two projections. The knowledge of the three-dimensional coordinates allows the determination of the true length and corrected densitometric values of foreshortened vessel sections.

Three-dimensional vascular reconstruction from projections: a theoretical review

Abstract: The three-dimensional reconstruction of vascular networks from X-ray projection images is reviewed in analogy to general three-dimensional computed tomography. Intrinsic vascular bed properties (connectivity, density (spareness), lumen dimensions, radio-opaque contrast uniformity, etc.) suggest a general data representation, provide a basis for the algorithmic nature of reconstruction, and can be used to predict tradeoffs between reconstruction accuracy, number of views, and view orientation. >

3d Movement Detecticn: A Hierarchical Approach

Abstract: To study quantitatively human body movement the modelling approach is used in biomechanics; body's segments are schematically represented by links, and joint articulations by hinges. For an exhaustive description of the movement of such a structure, it is enough to detect and follow hinges movement from which the trajectories of all the structure can be recovered. A common approach is to mark these points by active or passive markers and then to use a suitable machine to detect them. For this purpose, a 3D movement detection system (Elite) has been developed. It is a vision system with a two level architecture; the first level, hardware implemented, receives the image from a set of TVcameras and recognizes, in real-time. (100Hz), any number of markers on the scene. By a cross-correlation filtering algorithm, the markers are recognized on the basis of their shape. This process gives a high signal to noise ratio, which allows the operator to disregard reflexes and other objects present in the environment (which are rejected); moreover the markers (small,light pieces of reflecting paper) do not cause any constraints on the subject to be analyzed. The second level, software implemented, performs a more intelligent task, that of matching the surveyed markers' coordinates to the marker arrangement predefined in a model. Each model constituted by links and hinges interconnected among them is the visual schematic representation of the subject to be analyzed. Starting from it, a special designed program (KAT - Knowledge based Automatic Tracking) automatically tracks down each marker frame by frame reconstructing the coordinates even when the overlapping of body segments occurs. After this procedure, a 3D reconstruction is carried out by means of a generalized triangulation algorithm starting from the images picked up by a couple of TVcameras. The description of the system is completed with the 3D resection and calibration algorithm. The first allows the recovering of TVcameras geometrical internal and external parameters which are requested in 3D reconstruction; calibration algorithm allows distortions correction and leads to a final high accuracy in the computation of 3D markers position.

Spatial Reconstruction of the Bone Trabecular Network. A Simple Algorithm for the Leitz TAS+ and Related Image Analyzers

Abstract: An algorithm is proposed for 3D reconstruction of trabecular bone from bone biopsies with a Leitz image analyzer (TAS+). Images are directly acquisited and processed by the analyzer, allowing demonstration of branching patterns of trabeculae with hidden line removal. The complete algorithm is provided, allowing an easy adaptation to other image analyzers. (The J Histotechnol 11:133, 1988.)

A Spherical Approach To Optical Flow

Abstract: In the computer vision literature, the vision model used most frequently has incorporated Monge surfaces and either orthographic or planar perspective. In recent years, a vision model based on spherical surfaces and spherical perspective has arisen as an alternative that avoids the limitations of these standard models. In this paper we discuss the use of the spherical vision model in the study of optical flow for smooth surfaces.

2 - Description d'un algorithme de reconstruction de volume à partir de projections coniques par déconvolution tri-dimensionnelle

Abstract: This paper presents the discretization and the application of a 3D reconstruction method front cone beam X-ray projections . The generalized back projection theorem, established in a previous work, is the theoretical basis for the method. It allows to reduce the reconstruction problem to a 3D deconvolution problem . The proposed algorithm essentially consists in two steps : (i) computation of the discrete corrected back projection of all the cone beam projections ; (ii) deconvolution of the result . After vectorization, this algorithm has been implemented on a CDC CYBER 205 computer. A simple and comprehensive test function is proposed to evaluate the algorithm relatively to various error criteria . The first simulations show that the reconstruction results are very satisfying when the X-ray sources are located in the whole space around the object, in accordance with the theory (4 it geometry). Furthermore, even in poor acquisition conditions the algorithm seems to give a first approximation of the object which can be sufficient to study its morphological aspect .

Scene reconstruction from rays application to stereo data

Abstract: The problem of reconstructing shapes of objects from sparse measurements such as points on the boundary of an object is considered. In most situations, the points are the endpoints of a curve or a ray which does not cross the objects. For example, if the sensor is an optical device, the ray is the straight line (the optical ray) joining the camera center to the point. It is shown that the information provided by rays is crucial when determining the shapes of objects, and nonheuristic reconstruction methods in 2-D and 3-D space are described. An efficient method is derived for the reconstruction of surfaces from 3-D segments provided by a stereo vision process. >