Showing papers on "Feature (computer vision) published in 1987"

Dynamic sensor-based control of robots with visual feedback

Abstract: Sensor-based robot control may be viewed as a hierarchical structure with multiple observers. Actuator, feature-based, and recognition observers provide the basis for multilevel feedback control at the actuator, sensor, and world coordinate frame levels, respectively. The analysis and design of feature-based control strategies to achieve consistent dynamic performance is addressed. For vision sensors, such an image-based visual servo control is shown to provide stable and consistent dynamic control within local regimes of the recognition observer. Simulation studies of two- and three-degree-of-freedom systems show the application of an adaptive control algorithm to overcome unknown and nonlinear relations in the feature to world space mapping.

Finding Trajectories of Feature Points in a Monocular Image Sequence

Abstract: Identifying the same physical point in more than one image, the correspondence problem, is vital in motion analysis. Most research for establishing correspondence uses only two frames of a sequence to solve this problem. By using a sequence of frames, it is possible to exploit the fact that due to inertia the motion of an object cannot change instantaneously. By using smoothness of motion, it is possible to solve the correspondence problem for arbitrary motion of several nonrigid objects in a scene. We formulate the correspondence problem as an optimization problem and propose an iterative algorithm to find trajectories of points in a monocular image sequence. A modified form of this algorithm is useful in case of occlusion also. We demonstrate the efficacy of this approach considering synthetic, laboratory, and real scenes.

Anomalies: The January Effect

Abstract: This feature will report successful searches for disconfirming evidence -- economic anomalies. As suggested by Thomas Kuhn, an economic anomaly is a result inconsistent with the present ec...

Anomalies: Weekend, Holiday, Turn of the Month, and Intraday Effects

Abstract: This feature will report successful searches for disconfirming evidence--economic anomalies. As suggested by Thomas Kuhn, an economic anomaly is a result inconsistent with the economics pa...

Segmenting Images Using Localized Histograms and Region Merging

Abstract: THIS PAPER DESCRIBES A WORKING SYSTEM FOR SEGMENTING IMAGES OF COMPLEX SCENES. THE SYSTEM HAS EVOLVED OUT OF MANY YEARS OF RESEARCH IN LOW-LEVEL IMAGE SEGMENTATION AT THE UNIVERSITY OF MASSACHUSETTS. SEGMENTATIONS PRODUCED BY THE SYSTEM ARE USED EXTENSIVELY IN THE IMAGE INTERPRETATION WORK HERE. THE SYSTEM FIRST PRODUCES SEGMENTATIONS BASED UPON AN ANALYSIS OF SPATIALLY LOCALIZED FEATURE HISTOGRAMS. THESE SEGMENTATIONS ARE THEN POST-PROCESSED WITH A REGION MERGING ALGORITHM. A SIMPLE EXTENSION OF THE LOCAL HISTOGRAM ALGORITHM TO MULTIPLE FEATURES IS PRESENTED. ISSUES OF PARAMETER SELECTION FOR THE LOCAL HISTOGRAM SEGMENTATION ALGORITHM ARE ADDRESSED BY PRESENTING A MAPPING FROM THE MULTI-DIMENSIONAL PARAMETER SPACE TO A ONE DIMENSIONAL `SENSITIVITY'' SPACE. RESULTS ARE INCLUDED WHICH DEMONSTRATE THE ROBUST CHARACTER OF THE ALGORITHMS WHEN APPLIED TO OUTDOOR AND AERIAL IMAGERY.

New Methods for Matching 3-D Objects with Single Perspective Views

Abstract: In this paper we analyze the ability of a computer vision system to derive properties of the three-dimensional (3-D) physical world from viewing two-dimensional (2-D) images. We present a new approach which consists of a model-based interpretation of a single perspective image. Image linear features and linear feature sets are backprojected onto the 3-D space and geometric models are then used for selecting possible solutions. The paper treats two situations: 1) interpretation of scenes resulting from a simple geometric structure (orthogonality) in which case we seek to determine the orientation of this structure relatively to the viewer (three rotations) and 2) recognition of moderately complex objects whose shapes (geometrical and topological properties) are provided in advance. The recognition technique is limited to objects containing, among others, straight edges and planar faces. In the first case the computation can be carried out by a parallel algorithm which selects the solution that has received the largest number of votes (accumulation space). In the second case an object is uniquely assigned to a set of image features through a search strategy. As a by-product, the spatial position and orientation (six degrees of freedom) of each recognized object is determined as well. The method is valid over a wide range of perspective images and it does not require perfect low-level image segmentation. It has been successfully implemented for recognizing a class of industrial parts.

Method and apparatus for identifying features of an image on a video display

Abstract: A computer graphic interface allows a user to obtain descriptive information concerning a feature of a displayed image by pointing to the location of the feature. Conversely, the user may enter descriptive textual information, and the locations of responsive features are indicated. The data processing and memory storage requirements are minimized by encoding information about the image as a pixel bit map, and a color map in which the addresses or indices of the color map are correlated with the addresses or pointers to strings of descriptive information. Each color map address corresponds to a predefined set of features and descriptive information about those features. Since the pixel bit map defines a color map address for each location on the image, suitable programming of the color map can insure proper correlation of descriptive information with corresponding locations on the image. The correlation between color map addresses and the descriptive information about the features is represented most compactly by arranging or sorting the entries in the color map so that there is a correspondence between each predefined feature and a continuous range of color map addresses. Therefore, for a specified color map address, the corresponding set of features and their pointers can be found by comparing the specified color map address to the limits of the color map address ranges for the various features.

Adaptive tracking vision and guidance system

Abstract: An adaptive tracking vision and guidance system for use with robot systems includes a structured light vision system and an expert system based look ahead feature tracker. The adaptive tracking vision system may be used to track a feature on which the robot is to operate and to supply guidance information in real time to the robot or to an adaptive stage on the robot arm train. The vision system may be a structured light system capable of scanning a relatively large window, and a control computer. The feature tracker includes an expert system software package and interface packages for the robot controller, host computer, vision camera and operator console. The vision system does an initial search to locate a feature based on a priori knowledge of characteristics of the feature. Then the system tracks the feature by scanning ahead, adaptively planning future scans, predicting the feature path and choosing the data to be used to reduce processing time. The system can supply guidance information to an adaptive stage controlling an end effector tool so that the robot can perform the desired operation precisely on the feature. If the feature is lost while scanning ahead, the vision system does a recovery search to relocate the feature, including switching to alternative mapping systems if necessary. If the feature cannot be recovered a signal is given to the operator.

A Unification Method for Disjunctive Feature Descriptions

Abstract: Although disjunction has been used in several unification-based grammar formalisms, existing methods of unification have been unsatisfactory for descriptions containing large quantities of disjunction, because they require exponential time. This paper describes a method of unification by successive approximation, resulting in better average performance.

New approach for evaluating form errors of engineering surfaces

Abstract: Manufactured features generally deviate in size, form and relations with respect to other features from the features desired by the designer. Designers specify certain limits for these deviations that depend on functional requirements. The specification of different form errors is dealt with by the International Organization for Standardization (ISO). ISO also specifies that an ideal geometric feature must be established from the actual measurements such that the maximum deviation between it and the actual feature concerned is the least possible value. In practice, the least squares (LS) method is sometimes used for establishing the ideal feature, as this method is based on sound mathematical principles. However, the least squares procedure does not yield a minimum value. Therefore some attempts have been made to arrive at a form error based on the minimum deviation (MD) principles. A stray peak or valley on the actual feature introduces considerable variation in the results obtained by the minimum deviation method. This paper suggests a new approach based on the minimum average deviation (MAD). In this method, the ideal feature is established by using a search technique for different geometric features such as straight lines, circles, planes, cylinders and spheres. Using simple numerical examples, the values obtained by this approach are compared with the values obtained by the least squares and minimum deviation methods.

Color halftone image processing apparatus producing various screen angles and having an adaptive color image data conversion look-up table and a small-capacity masking memory

Abstract: An image processing apparatus includes a generator for generating predetermined image data representing an image and an image reproducing device for reproducing the image. A reading device reads the image reproduced by the reproducing device and generates read data representative thereof. A converter converts the read data generated by the reading device to converted image data which is substantially equal to the predetermined image data generated by the generating device. The converting means includes a table adapted to receive the read data as address data for converting and outputting the converted image data. The converted image data may be transmitted to the reproducing device which is arranged to reproduce an image represented by either the predetermined image data or the converted image data. The image processing apparatus of the present invention may also incorporate a digitizer having a dither processing circuit for specifically processing data in accordance with a plurality of kinds of basic cells. Each cell has a number of threshold values and is of the same shape but has different arrangements of threshold values. The plurality of kinds of basic cells are alternately arranged in two perpendicular directions and are shifted relative to one another in row and column directions of the dither matrices by a and b respectively. The variables a and b represent displacements between the plural kinds of basic cells and each has a value other than zero. The dither matrix has a size N×N determined in accordance with the equation N=m(a+b) +n(a-b) where M and N are minimum values of integers and satisfy the equation M/N=(a+b)/a-b). In accordance with still a further feature of the present invention, the apparatus thereof may include an input device for inputting the image data as a plurality of bits. A first converter receives as an address the upper l bits of the image date in order to output converted image data and a second converter receives as an address the upper m bits of the image data in order to output converted data. Correcting means correct the converted image data by utilizing the converted data from the second converter and the lower n bits of the image data input by the input device.

Co-ordinate measuring

Abstract: A co-ordinate measuring machine has a probe (62) used to touch points on an article (61) under manual control by a user, e.g. from a joystick (69). A computer (71) determines a vector describing the direction of probe movement prior to touching each point. On the basis of the number of points touched by the user and their respective vectors, the computer makes a decision as to the shape of a feature of the article under inspection, e.g. whether it is a flange, slot, boss, bore, internal sphere, external sphere, cylinder or cone. It can then either calculate a relevant geometrical parameter such as a dimension of the feature, or write a part program for future inspection of the feature on further, nominally identical articles (61). This can be done automatically without the need for the user to tell the computer anything about the shape of the feature concerned.

Automated cartographic generalization for linear map features

Abstract: Cartographic generalization is the task of making smaller-scale maps (for example, 1:100,000) from larger-scale maps (for example, 1:20,000). An automated method is presented for generalization of linear features that are already in a topologically structured computer-readable form. The method consists of three stages: feature elimination, feature simplification, and interference detection and resolution. The first stage is implemented using English-like rules. These rules specify which map features are to be eliminated at the generalized map scale. They also control when the second and third stages are invoked. The second stage consists of the simplification of linear map features. A procedure for polyline simplification is presented, along with new algorithms for subdivision of linear features and riverbank combination. The third stage deals with interference and detection among linear map features. Interference may occur when a simplified feature is added to the generalized map. Due to the width of the symbol representing this feature, it may overlap another feature already on the generalized map. If that occurs, one or both of the features have to be displaced so as to eliminate the interference between them. This displacement may cause the end nodes of the displaced feature to be moved, which results in the propagation of the displacement to the features that are connected to these displaced end nodes. The algorithms used to detect feature interference and to perform the subsequent feature displacement propagation are described. A FORTRAN 77 computer program called MAPEX was written to perform automated generalization of linear map features. This program assumes a topologically structured input data file similar to the U.S. Geological Survey 1:24,000 scale Digital Line Graph map data files. To integrate the English-like rules with MAPEX, a rule compiler was written to translate the English-like rules to a FORTRAN 77 subroutine. The results of running MAPEX on a topologically structured test data file are given, along with the algorithmic details used in the implementation of MAPEX.

Area feature sorting mechanism for neighborhood-based proximity correction in lithography processing of integrated circuit patterns

Abstract: Proximity correction of an area feature in lithography processing of an integrated circuit device is facilitated by a procedure for locating those area features which lie within a prescribed proximity correction neighborhood of an area feature of interest. The method comprises examining the pattern, for each of a plurality of area windows associated with respective distance increments extending in a first direction across the pattern, to identify those area features which are located within a respective area window extending in a second direction from a respective distance increment, and storing information representative of characteristics of each identified area feature in a group that is associated with that respective area window. For each of the area features whose identities have been stored, a boundary window which delineates the prescribed proximity correction neighborhood relative to that respective area feature is defined. Then, for a respective area feature, those area features that have been stored and geometry data for which are contained within the boundary window for the respective area feature are selected as lying within the prescribed proximity correction neighborhood of the area feature of interest.

Determining object motion in a sequence of stereo images

Abstract: The motion of a three-dimensional object is determined from a sequence of stereo images by extracting three-dimensional features, establishing correspondences between these features, and finally, computing the rigid motion parameters. Three-dimensional features are extracted from the depth map of a scene. A two-pass relaxation method is developed for matching features extracted from successive depth maps. In each iteration, geometrical relationships between a feature and its neighbors in one map are compared to those between a candidate in the other map and its neighbors to update the matching probability of the candidate. The comparison of the geometrical relationship is based on the principle of conservation of distance and angle between features during rigid motion. The use of three-dimensional features allows one to find the rotation and translation components of motion separately via solving linear equations. Experimental results using several sets of real data are presented to illustrate results and difficulties.

3D Positional Integration from Image Sequences.

Abstract: An explicit three-dimensional (3D) representation is constructed from feature points extracted from a sequence of images taken by a moving camera. The points are tracked through the sequence, and their 3D locations are accurately determined by use of Kalman filters. The egomotion of the camera is also determined.

Pattern recognition method

Abstract: A pattern recognition method comprises a contour tracing step of a contour of a binary image in a predetermined direction along a predetermined type of pixels At this time, four adjacent pixels on upper, lower, left-hand and right-hand sides of a point of interest are sequentially checked Then, one pixel of a predetermined type out of the four adjacent pixels which is the first pixel of the predetermined type found by the check is designated as a next trace point subsequent to the point of interest In this process, a feature of each of the traced pixels is extracted The feature is represented by a directional code which indicates a direction of motion from the point of interest to the next trace point The features are successively registered in a one-dimensional form Histogram is produced for each sub-region storing a part of the one-dimensional feature sequence A histogram for each sub-region is compared with a related reference histogram for a known character

Conduction in granular metals with potential disorder

Abstract: The author presents a detailed analysis of conduction in discontinuous metals based on the traditional model but taking into account both the statistical aspects of relevant structural features and the presence of potential disorder. The analysis predicts conductivity with a temperature dependence close to simple activation at all temperatures. It is argued that an essential feature of the physics of transport in these systems has yet to be identified and incorporated into a satisfactory detailed model.

Map display system for vehicle navigation

Abstract: A map display system includes a vehicle symbol positioned in relation to background map data, and provides a large amount of relevant, easy-to-read map data to a user in a brief period of time. A first feature of the invention involves the drawing of aircraft landing and take-off areas in true size and orientation relative to background map data. A second feature involves display of navigation control areas in relation to an aircraft symbol. In accordance with a third feature, a user can select for display only objects meeting preselected criteria, such as airports having runways of a minimum length. A user can select an object for which text appears on the screen adjacent to that object and which may indicate relative position or distance to such object from the vehicle, in accordance with a fourth feature. A user can select full screen text of information concerning an object on a screen display, in accordance with a fifth feature. A sixth feature provides an automatic alert for an increased oncoming minimum safe altitude for aircraft flight. In an seventh feature, background map data is held stationary while a vehicle symbol moves in relation to the background map data. An emergency aircraft mode of operation, in accordance with an eighth feature, causes automatic display of at least a predetermined number of suitable landing sites according to preselected criteria, such as minimum runway length. A ninth feature involves display of planned and actual travel patterns of a vehicle.

Rough approximation of shapes in pattern recognition

Abstract: The paper approaches the nature of a feature recognition process through the description of image features in terms of the rough sets. Since the basic condition for representing images must be satisfied by any recognition result, elementary features are defined as equivalence classes of possible occurrences of specific fragments existing in images. The names of the equivalence classes (defined through specific numbers of objects and numbers of background parts covered by a window) constitute the best lower approximation of window contents (i.e., names of recognized features). The best upper approximation is formed by the best lower approximation, its attributes, and parameters, all referenced to the object fragments situated in the window. The rough approximation of shapes is resistant to accidental changes in the width of contours and lines and to small discontinuities and, in general, to possible positions or changes in shape of the same feature. The rough sets are utilized also on the level of image processing for noiseless image quantization.

System for forming design pattern data

Abstract: In a design pattern data forming system, a design pattern such as an apparel pattern having marks for discriminating feature points of the pattern is read by an image reader to prepare binary picture element data. The picture element data is processed by a computer to prepare an X-Y point series data consisting of X and Y addresses corresponding to outer profile lines of the design pattern at every three scanning lines sequentially scanned in the Y direction. From the X-Y point series data, positions corresponding to the feature points are extracted and data between adjacent feature points are divided as line data. The X and Y addresses of X-Y point series data interconnecting start and end points corresponding to the feature points and an intermediate point(s) of the line data are stored in a memory. The X-Y point series data have line data arranged as one complete trace data for the pattern profile, and feature point codes representing contents of the feature points are associated with the start and end points of the line data. The system can automatically read design pattern data without requiring a specially skilled operator by merely loading a sheet of paper or a cutout pattern piece into the image reader to recognize feature points at a high speed in a simple manner.

Automatic discrimination of features in grey‐scale images

Abstract: SUMMARY Automatic setting of discriminator thresholds to select features from grey-scale images is desirable to reduce variations due to operator variability, and to adapt to changing illumination or sample characteristics. This can be carried out in several ways. One locates minima in the original brightness histogram. For images in which minima are absent or poorly defined other techniques are preferred. Settings that minimize the change in total feature area or perimeter with changes in the threshold setting are suitable for a wide variety of images. After processing to extract edges a fixed percentage of the number of pixels in the image may be useful.