scispace - formally typeset
Search or ask a question

Showing papers on "Feature (computer vision) published in 1991"


Journal ArticleDOI
TL;DR: A texture segmentation algorithm inspired by the multi-channel filtering theory for visual information processing in the early stages of human visual system is presented, which is based on reconstruction of the input image from the filtered images.

2,351 citations


Journal ArticleDOI
TL;DR: This paper proves that SV feature vector has some important properties of algebraic and geometric invariance, and insensitiveness to noise, and these properties are very useful for the description and recognition of images.

314 citations



Journal ArticleDOI
01 Feb 1991
TL;DR: The authors develop methodologies for the automatic selection of image features to be used to visually control the relative position and orientation (pose) between the end-effector of an eye-in-hand robot and a workpiece.
Abstract: The authors develop methodologies for the automatic selection of image features to be used to visually control the relative position and orientation (pose) between the end-effector of an eye-in-hand robot and a workpiece. A resolved motion rate control scheme is used to update the robot's pose based on the position of three features in the camera's image. The selection of these three features depends on a blend of image recognition and control criteria. The image recognition criteria include feature robustness, completeness, cost of feature extraction, and feature uniqueness. The control criteria include system observability, controllability, and sensitivity. A weighted criteria function is used to select the combination of image features that provides the best control of the end-effector of a general six-degrees-of-freedom manipulator. Both computer simulations and laboratory experiments on a PUMA robot arm were conducted to verify the performance of the feature-selection criteria. >

223 citations


Journal ArticleDOI
TL;DR: A new set of textural measures derived from the texture spectrum is presented, which extract textural information of an image with a more complete respect of texture characteristics (in all the eight directions instead of only one displacement vector used in the co-occurrence matrix approach).

208 citations


Patent
31 Jul 1991
TL;DR: In this article, a training database stores training records, wherein each training record includes predictor data fields and each predictor data field containes a feature, each feature is a natural language term and a target data field containing a target value representing a classification of the record.
Abstract: Classification of natural language data wherein the natural language data has an open-ended range of possible values or the data values do not have a relative order. A training database stores training records, wherein each training record includes predictor data fields. Each predictor data field containes a feature, wherein each feature is a natural language term, and a target data field containing a target value representing a classification of the record. Features may also include conjunctions of natural language terms and each feature may also be a member of a category subset of features. The training database stores, for each feature, a probability weight value representing the probability that a record will have the target value contained in the target data field if a feature contained in a corresponding predictor data field occurs in the record. Features are extracted from a new record and each feature from the new record is used to query the training records to determine the probability weights from the training records having matching features. The probability weights are accumulated for each training record to determine a comparison score representing the probability that the training record matches the new record and provide an output indicating the training records most probability matching the new record.

180 citations


Patent
29 Jul 1991
TL;DR: In this article, a self-describing feature table, accessible by device drivers, is used to customize the device drivers at installation, or at boot; or the device driver can branch on the data in the feature table.
Abstract: A computer system with self-describing feature table, accessible by device drivers. Thus a simple process can access these feature tables to fully customize the device drivers at installation, or at boot; or the device driver can branch on the data in the feature table. Thus, a new degree of flexibility is achieved without degrading performance.

167 citations


Journal ArticleDOI
TL;DR: The 2-mechanism hypothesis provides a resolution to conflicting findings in the visual search and illusory-conjunction literature and a possible interpretation of the findings with a single guided search mechanism for feature integration is discussed.
Abstract: Four experiments used the visual search paradigm to examine feature integration mechanisms. Reaction time to determine the presence or absence of a conjunctive target is relatively fast and exhaustive for low-density displays. Search rate is slow and self-terminating for high-density displays. Density effects do not arise when the target is defined by a unique feature. Two mechanisms are proposed for feature integration. A fast mechanism integrates features on the basis of coarse location information coded with the initial registration of the features. This coarse location mechanism requires that display items be spaced apart. A second, slower mechanism is used when objects are clumped together. The 2-mechanism hypothesis provides a resolution to conflicting findings in the visual search and illusory-conjunction literature. A possible interpretation of the findings with a single guided search mechanism for feature integration is also discussed.

147 citations


Journal ArticleDOI
TL;DR: Both simulation and experimental tests of the centroid method for subpixel accuracy image feature location prediction are presented for the case of the location of the center lines of the images of projected light stripes in a triangulation-based three-dimensional shape measurement system.
Abstract: An analysis of the properties of the centroid method for subpixel accuracy image feature location is presented. This method is free of systematic error if the maximum spatial frequency of the image incident on the image sensor is less than the sensor's sampling frequency. This can be achieved by using a lens aperture setting such that the modulation transfer function cut-off frequency due to diffraction is appropriately small. Both simulation and experimental tests of this prediction are presented for the case of the location of the center lines of the images of projected light stripes in a triangulation-based three-dimensional shape measurement system.

140 citations


Patent
17 Jul 1991
TL;DR: In this article, a method of processing an image including the steps of locating within the image the position of at least one predetermined feature, extracting from the image data representing each feature, and calculating for each feature a feature vector representing the position in an N-dimensional space, such space being defined by a plurality of reference vectors each of which is an eigenvector of a training set of like features in which the image of each feature is modified to normalize the shape of the feature, which step is carried out before calculating the corresponding feature vector.
Abstract: A method of processing an image including the steps of: locating within the image the position of at least one predetermined feature; extracting from the image data representing each feature; and calculating for each feature a feature vector representing the position of the image data of the feature in an N-dimensional space, such space being defined by a plurality of reference vectors each of which is an eigenvector of a training set of like features in which the image data of each feature is modified to normalize the shape of each feature thereby to reduce its deviation from a predetermined standard shape of the feature, which step is carried out before calculating the corresponding feature vector.

128 citations


PatentDOI
TL;DR: In this article, an apparatus for distinguishing benign from malignant tumors in ultrasonic images of candidate tissue taken from a patient is presented. But the method is limited to the detection of benign vs. malignant tumor.
Abstract: An apparatus for distinguishing benign from malignant tumors in ultrasonic images of candidate tissue taken from a patient. A region of interest is located and defined on the ultrasonic image, including substantially all of the candidate tissue and excluding substantially all the normal tissue. The region of interest is digitized, generating an array of pixels intensity values. A first features is generated from the arrays of pixels corresponding to the angular second moment of the pixel intensity values. A second feature is generated from the array of pixels corresponding to the inverse contrast of the pixel intensity values. A third feature is generated from the array of pixels corresponding to the short run emphasis of the pixel intensity values. The first, second and third feature values are provided to a neural network. A set of trained weights are applied to the feature values, which generates a network output between 0 and 1, whereby the output values tend toward 1 when the candidate tissue is malignant and the output values tend toward 0 when the candidate tissue is benign.

Patent
29 Jul 1991
TL;DR: In this article, a color processing system is provided that specifies color in a standardized, perceptually uniform color space and resolves color differences between input and/or output devices. But it is not a color rendering dictionary, but a color mapping system that adjusts the results of color mapping to fit the color gamut of the output device.
Abstract: A color processing system is provided that specifies color in a standardized, perceptually uniform color space and resolves color differences between input and/or output devices. The color processing system utilizes a color mapping techniques to adjust the results of color mapping to fit the color gamut of the output device. The color processing system may be implemented as a color rendering dictionary in a PostScript® Level 2 interpreter. An additional feature involves adaptation of PostScript® Level 1 applications to take advantage of the improved color processing system and PostScript® Level 2 flexibility.

Patent
Flavio M. Manduley1
24 Dec 1991
TL;DR: In this paper, a method for activating an optional feature in a data processing device that includes a microprocessor and memory is described, which includes storing in the memory: a number of applications programs, at least one of which has at least 1 optional feature; application manager software for controlling access to the application programs; a communications software module, accessed by means of the application manager, for controlling data communication by the device; and a program activation software module and including activation data that indicates whether each optional feature is activated or unactivated.
Abstract: A method is provided for activating an optional feature in a data processing device that includes a microprocessor and memory. The method includes storing in the memory: a number of applications programs, at least one of which has at least one optional feature; application manager software for controlling access to the application programs; a communications software module, accessed by means of the application manager, for controlling data communication by the device; and a program activation software module, accessed by means of the application manager and including activation data that indicates whether each optional feature is activated or unactivated. The method also includes receiving a request to operate one of the optional features, determining by reference to the activation data whether the feature is activated or unactivated, operating the feature if it is activated and otherwise preventing operation of the feature. The method further includes receiving a request to activate the feature (if unactivated) and updating the activation data to indicate that the feature has been activated.

Journal ArticleDOI
TL;DR: The time complexity of the shape matching algorithm is O(m), where m is the number of feature points, and the most important advantage of the method is its independence to translation, rotation and scaling of objects.

Patent
08 Nov 1991
TL;DR: The adaptive dynamic signature verification system of the invention includes a number of types of signature verification terminals which digitize signatures for further verification by statistical analysis as discussed by the authors, which allows for a degree of inconsistency in both speed and size of genuine signatures without losing the ability to discriminate against forgery attempts.
Abstract: The adaptive dynamic signature verification system of the invention includes a number of types of signature verification terminals which digitize signatures for further verification by statistical analysis. A feature set is employed that consists of a number of personalized features, some static and some dynamic, each of which is normalized with respect to both time and spatial dimensions and allows for a degree of inconsistency in both speed and size of genuine signatures without losing the ability to discriminate against forgery attempts. The statistical approach also develops and employs either an individual-specific or a common feature subset for accurate signature discrimination.

Proceedings ArticleDOI
09 Apr 1991
TL;DR: A method is presented to determine viewpoints for a robotic vision system for which object features of interest will simultaneously by visible, inside the field-of-view, in-focus, and magnified as required.
Abstract: A method is presented to determine viewpoints for a robotic vision system for which object features of interest will simultaneously by visible, inside the field-of-view, in-focus, and magnified as required. A technique that poses the problem in an optimization setting in order to determine viewpoints that satisfy all requirements simultaneously and with a margin is presented. The formulation and results of the optimization are shown, as well as experimental results in which a robot vision system is positioned and its lens is set according to this method. Camera views are taken from the computed viewpoints in order to verify that all feature detectability requirements are satisfied. >

Journal ArticleDOI
TL;DR: This paper found that language is a feature to distinguish the human from the animal that has seemed a more enduring diagnostic character than some. But how is the breath of words to be made visible in the stony traces of the Pleistocene?
Abstract: Language is a feature to distinguish the ‘human’ from the ‘animal’ that has seemed a more enduring diagnostic character than some. But how is the breath of words to be made visible in the stony traces of the Pleistocene?

Journal ArticleDOI
TL;DR: This paper reviews the different approaches developed to estimate motion parameters from a sequence of two range images and gives the mathematical formulation of the problem along with the various modifications by different investigators to adapt the formulation to their algorithms.
Abstract: The estimation of motion of a moving object from a sequence of images is of prime interest in computer vision. This paper reviews the different approaches developed to estimate motion parameters from a sequence of two range images. We give the mathematical formulation of the problem along with the various modifications by different investigators to adapt the formulation to their algorithms. The shortcomings and the advantages of each method are also briefly mentioned. The methods are divided according to the type of feature used in the motion estimation task. We address the representational and the computational issues for each of the methods described. Most of the earlier approaches used local features such as corners (points) or edges (lines) to obtain the transformation. Local features are sensitive to noise and quantization errors. This causes uncertainties in the motion estimation. Using global features, such as surfaces, makes the procedure of motion computation more robust at the expense of making the procedure very complex. A common error is assuming that the best affine transform is the best estimate of the desired motion, which in general is false. It is important to make the distinction between the motion transform and the general affine transform, since an affine transform may not be realized physically by a rigid object.

Proceedings ArticleDOI
W. Jang1, Zeungnam Bien1
09 Apr 1991
TL;DR: By means of various examples, the method of feature-based servoing of a robot proposed is proved to be very effective for conducting object-oriented robotic tasks.
Abstract: A method is presented for using image features in servoing a robot manipulator. Specifically, the concept of a feature is mathematically defined, and the differential relationship between the robot motion and feature vector is derived in terms of a feature Jacobian matrix and its generalized inverse. The feature-based PID (proportional-integral-derivative) controller is established with three scalar gains and an n*n matrix. By means of various examples, the method of feature-based servoing of a robot proposed is proved to be very effective for conducting object-oriented robotic tasks. >

Journal ArticleDOI
TL;DR: A new fast method is presented for the quantification of patient set-up errors during radiotherapy with external photon beams, which proved to be a source of large systematic errors, which could be reduced by offering a precise protocol for the feature alignment.
Abstract: A new fast method is presented for the quantification of patient set-up errors during radiotherapy with external photon beams. The set-up errors are described as deviations in relative position and orientation of specified anatomical structures relative to specified field shaping devices. These deviations are determined from parameters of the image transformations that make their features in a portal image align with the corresponding features in a simulator image. Knowledge of some set-up parameters during treatment simulation is required. The method does not require accurate knowledge about the position of the portal imaging device as long as the positions of some of the field shaping devices are verified independently during treatment. By applying this method, deviations in a pelvic phantom set-up can be measured with a precision of 2 mm within 1 minute. Theoretical considerations and experiments have shown that the method is not applicable when there are out-of-plane rotations larger than 2 degrees or translations larger than 1 cm. Inter-observer variability proved to be a source of large systematic errors, which could be reduced by offering a precise protocol for the feature alignment.

Journal ArticleDOI
TL;DR: The concept of a form feature is progressively developed by consideration of dimensionality, contiguity, closure, finiteness, uniqueness, and representation.
Abstract: Feature definition, definition methods, properties, and representation are classified and formalized. Fundamental issues related to the definition and semantics of form features are explored. The concept of a form feature is progressively developed by consideration of dimensionality, contiguity, closure, finiteness, uniqueness, and representation. Issues related to feature-feature interactions are identified and the types of interactions classified. The consequences of viewpoint dependence of features and feature transformations are also discussed, specifically with regard to the possibility of feature standardization and data exchange.

Journal ArticleDOI
TL;DR: This research shows that emergent features can be effectively exploited to support tasks involving both integration of information and attention to individual data sources.
Abstract: Visual display design for dynamic systems may be helped by exploiting emergent features that allow subjects to easily distinguish different states of the system. Three different types of displays were compared--two object displays and one separated display--each in a version that had a salient emergent feature that distinguished normal and failed states and in a version that did not have such an emergent feature. Subjects monitored for global and local failures simultaneously, which presented demands similar to integrated and separated tasks, respectively. Displays with salient emergent features supported superior global failure detection and may also have helped local failure detection. An object display with a salient emergent feature supported both types of failures better than or at least as well as the other displays. This advantage was attributed to the fact that the display had a salient emergent feature rather than to the fact that it was an object display. This research shows that emergent features can be effectively exploited to support tasks involving both integration of information and attention to individual data sources.

Patent
08 Mar 1991
TL;DR: In this paper, a method for detecting change points between cuts from motion picture images including a plurality of consecutive images is presented, where the change rate exceeds a predetermined allowable value is detected as a change point between cuts of the motion picture image images.
Abstract: There is provided a method for detecting change points between cuts from motion picture images including a plurality of consecutive images. First of all, in this method, motion picture images which are the subject of detection of change point between cuts are inputted (21) in a time series by taking a frame as the unit, and a predetermined feature quantity including a color histogram possessed by image data of said motion picture images is produced (22) while taking a frame as the unit. A correlation coefficient with respect to feature quantity between the above described feature quantity and a feature quantity produced in an immediately preceding frame is then produced (23). Further, a change rate between the correlation coefficient of the current frame and a correlation coefficient produced for a frame preceding the current frame is produced (24). A time point whereat the change rate exceeds a predetermined allowable value is detected (25) as a change point between cuts of the motion picture images.

Patent
Amir Israel1
28 Feb 1991
TL;DR: In this paper, a line of light having a gradient profile is directed at an acute angle at the feature and its background to illuminate a strip of area thereacross, and the reflectance intensity of the illuminated strip will vary in accordance with the height thereof.
Abstract: To obtain an enhanced two-dimensional image of a raised, light-reflective reflective feature (12), and its lower height surrounding background such that the contrast between them is enhanced, a line of light having a gradient profile is directed at an acute angle at the feature and its background to illuminate a strip of area thereacross. The reflectance intensity of the illuminated strip will vary in accordance with the height thereof because of the gradient intensity profile of the line of light. When the line of light is spanned across the feature and its background, the resultant image of the feature and its background, captured by spanning a linescan camera (24) thereacross, will exhibit an enhanced contrast between the feature and its background because of the greater height of the feature.

Proceedings ArticleDOI
07 Oct 1991
TL;DR: The authors show that a matrix of image measurements can be factored by singular value decomposition into the product of two matrices that represent shape and motion, respectively.
Abstract: Recovery scene geometry and camera motion from a sequence of images is an important problem in computer vision. If the scene geometry is specified by depth measurements, that is, by specifying distances between the camera and feature points in the scene, noise sensitivity worsens rapidly with increasing depth. The authors show hat this difficulty can be overcome by computing scene geometry directly in terms of shape, that is, by computing the coordinates of feature points in the scene with respect to a world-centered system, without recovering camera-centered depth as an intermediate quantity. More specifically, the authors show that a matrix of image measurements can be factored by singular value decomposition into the product of two matrices that represent shape and motion, respectively. The results in this paper extend to three dimensions the solution the authors described in a previous paper for planar camera motion (ICCV, Osaka, Japan, 1990). >

Proceedings ArticleDOI
03 Jun 1991
TL;DR: An integrated segmentation technique that combines the strengths of the previous two techniques while eliminating their weaknesses is proposed and is truly unsupervised, since it eliminates the need for knowing the exact number of texture categories in the image.
Abstract: Multichannel filtering techniques are presented for obtaining both region- and edge-based segmentations of textured images. The channels are represented by a bank of even-symmetric Gabor filters that nearly uniformly covers the spatial-frequency domain. Feature images are obtained by subjecting each (selected) filtered image to a nonlinear transformation and computing a measure of energy around each pixel. Region-based segmentations are obtained by using a square-error clustering algorithm. Edge-based segmentations are obtained by applying an edge detector to each feature image and combining their magnitude responses. An integrated segmentation technique that combines the strengths of the previous two techniques while eliminating their weaknesses is proposed. The integrated approach is truly unsupervised, since it eliminates the need for knowing the exact number of texture categories in the image. >

Patent
29 Apr 1991
TL;DR: In this article, a parallel processing computer system for clustering data points in continuous feature space by adaptively separating classes of patterns is presented, which is based upon the gaps between successive data values within single features.
Abstract: A parallel processing computer system for clustering data points in continuous feature space by adaptively separating classes of patterns. The preferred embodiment for this massively parallel system includes preferably one computer processor per feature and requires a single a priori assumption of central tendency in the distributions defining the pattern classes. It advantageously exploits the presence of noise inherent in the data gathering to not only classify data points into clusters, but also measure the certainty of the classification for each data point, thereby identifying outliers and spurious data points. The system taught by the present invention is based upon the gaps between successive data values within single features. This single feature discrimination aspect is achieved by applying a minimax comparison involving gap lengths and locations of the largest and smallest gaps. Clustering may be performed in near-real-time on huge data spaces having unlimited numbers of features.

Journal ArticleDOI
TL;DR: The Form Feature Information Model (FFIM) that was developed by the pdes committee in the USA and accepted into the International Standards Organisation step standard was tested by means of experiments described in the paper, and a number of factors that increase the complexity of the feature data-exchange process were identified.
Abstract: Exchanging product data between feature-based geometric modelers presents some challenges. This is partly because features technology is still evolving, and also because there are major philosophical differences between the feature concepts on which different systems are based. The Form Feature Information Model (FFIM) that was developed by the pdes committee in the USA and accepted into the International Standards Organisation step standard was tested by means of experiments described in the paper. The modeler used in the experiments was the ASU Features Testbed, which was developed at the Arizona State University, USA, independently of the FFIM. The purpose of the experiments was to identify problems in the FFIM, and to suggest solutions where possible. In addition to the traditional problems of data transfer, such as numerical inaccuracies, a number of other factors that increase the complexity of the feature data-exchange process were identified. Some of these were the lack of relational positioning/locating information, multiple representations of a single feature, representation of certain popular profiles in rather tedious data structures, and the nonunique mapping of features between the two systems. The loss of semantic information, ecapsulated in rules, methods and constraints, seems to place some serious limits on information that can be exchanged with the current version of the FFIM. The pdes/step models are essentially static snapshots that are adequate for exchanging geometric and topological data, but are insufficient for dynamic exchange of feature models that encode engineering semantics in addition to geometric shapes.

Proceedings ArticleDOI
M.A. Shackleton1, W.J. Welsh1
03 Jun 1991
TL;DR: A facial feature classification technique that independently captures both the geometric configuration and the image detail of a particular feature is described and results show that features can be reliably recognized using the representation vectors obtained.
Abstract: A facial feature classification technique that independently captures both the geometric configuration and the image detail of a particular feature is described. The geometric configuration is first extracted by fitting a deformable template to the shape of the feature (for example, an eye) in the image. This information is then used to geometrically normalize the image in such a way that the feature in the image attains a standard shape. The normalized image of the facial feature is then classified in terms of a set of principal components previously obtained from a representative set of training images of similar features. This classification stage yields a representation vector which can be used for recognition matching of the feature in terms of image detail alone without the complication of changes in facial expression. Implementation of the system is described and results are given for its application to a set of test faces. These results show that features can be reliably recognized using the representation vectors obtained. >

Book ChapterDOI
01 Jan 1991
TL;DR: A model for simulating the nijimi effect with computers has been developed, based on the physical analysis of the construction of the paper, the characteristics of liquid flow and particle adsorption, and the interaction between the liquid and the paper.
Abstract: Diffuse painting, or ‘Nijimi’, which is the most remarkable feature of black ink painting (called ‘Sumie’ in Japanese), is produced by letting the ink diffuse into the absorbent paper. The diffusion of the ink usually results in a delicate blurred image. A model for simulating the nijimi effect with computers has been developed. It is based on the physical analysis of the construction of the paper, the characteristics of liquid flow and particle adsorption, and the interaction between the liquid and the paper. The images generated by this model are proved to fit well with the observed patterns.