Early orientation selection: Tangent fields and the dimensionality of their support

doi:10.1016/0734-189X(85)90003-9

Computer Vision, Graphics, and Image Processing

Volume 32, Issue 1, October 1985, Pages 74-103

https://doi.org/10.1016/0734-189X(85)90003-9 Get rights and content

Orientation selection is the inference of orientation information out of images. It is one of the foundations on which other visual structures are built, since it must precede the formation of contours out of pointillist data and surfaces out of surface markings. We take a differential geometric view in defining orientation selection, and develop algorithms for actually doing it. The goal of these algorithms is formulated in mathematical terms as the inference of a vector field of tangents (to the contours), and the algorithms are studied in both abstract and computational forms. They are formulated as matching problems, and algorithms for solving them are reduced to biologically plausible terms. We show that two different matching problems are necessary, the first for 1-dimensional contours (which we refer to as Type I processes) and second for 2-dimensional flows (or Type II processes). We conjecture that this difference is reflected in the response properties of “simple” and “complex” cells, respectively, and predict several other psychophysical phenomena.

References (36)

MayhewJ. et al.
Psychophysical and computational studies towards a theory of human stereopsis
Artif. Intell.
(1981)
SchumerR. et al.
Length summation in simple cells of cat striate cortex
Vision Res.
(1984)
TerzopoulosD.
Multilevel computational processes for visual surface reconstruction
Computer Vision Graphics Image Process
(1983)
ZuckerS.W. et al.
Receptive fields and the reconstruction of visual information
BinfordT.
Inferring surfaces from images
Artif. Intell.
(1981)
FleetD. et al.
A Spatio-Temporal Model for Early Visual Processing
GibsonJ.
Perception of the Visual World
(1950)
GilbertC.
Microcircuitry of the visual cortex
Ann. Rev. Physiol.
(1983)
GlassL.
Moire effect from random dots
Nature (London)
(1969)
HornB.
Obtaining shape from shading information

HubelD. et al.

Functional architecture of macaque monkey visual cortex

HummellR.A. et al.

On the foundations of relaxation labeling processes

IEEE Trans. Pattern Anal. Mach. Intell.

(1983)

JuleszB.

Foundations of Cyclopean Perception

(1971)

KanizsaG.

Organization in Vision

(1979)

KassA. et al.

Analyzing oriented textures

KoffkaK.

Principles of Gestalt Psychology

(1935)

LeclercY. et al.

The local structure of intensity discontinuities in one dimension

LinkN. et al.

Sensitivity to Corners in Flow Patterns

Cited by (78)

12 Computational neural models of spatial integration in perceptual grouping
2001, Advances in Psychology
Recent developments in the neural computational modeling of perceptual grouping are described with reference to a newly proposed taxonomy to formalize mechanisms of spatial integration. This notational framework and nomenclature is introduced in order to clarify key properties common to all or most models, while permitting unique attributes of each approach to be independently examined. The strength of spatial integration in the models that are considered is always some function of the distances and relative alignments in perceptual space of the centers of units representing orienrational features or energy in a visual scene. We discuss the significance of variations of the constituents of an activation function for spatial integration, and also consider the larger modeling framework in which this function is applied in each approach. We also discuss the relationship of feedforward and feedback mechanisms and the issues of self-organization as core principles underlying the establishment of spatial integration mechanisms. The relationship of the grouping models to models of other visual competencies is considered with respect to prospects for future research.
The role of objects in perceptual grouping
1999, Acta Psychologica
Perceptual organization can be viewed as the selection of the best or “most reasonable” parse of a given scene. However, the principles that determine which interpretation is most reasonable have resisted most attempts to define them formally. This paper summarizes a formal theory of human perceptual organization, called minimal model theory, in which the best interpretation of a given scene is expressed as the formally minimal interpretation in a well-defined space of possible interpretations. We then focus specifically on the role of types of grouping units, in particular the difficult notion of “object”. Although grouping is often thought of as the process of dividing the image into objects, most research in perceptual grouping actually focuses on simpler types of units, such as contours and surfaces. Minimal model theory characterizes grouping units at a logical level, demonstrating how formal assumptions about units induce the observer to place a certain preference ranking on interpretations. The theory is then applied to the more subtle problem of objects, culminating in a definition for objects that is formally rigorous but at the same time captures some of the flexibility of human intuitions about objects.
Curvilinearity, covariance, and regularity in perceptual groups
1997, Vision Research
A curvilinear pattern among a series of visual items (e.g., dots) can be regarded as a kind of probabilistic inference, in which each consecutive angle, regarded independently, is more nearly collinear than would be expected by chance alone. This paper investigates judgments of curvilinearity as a function of the joint distribution of successive inter-dot angles. Subjects were asked to classify 4 and 5-dot configurations as having been generated by a curvilinear generating process, vs independently. Their results are distributed as a gaussian over the inter-dot angles with mean 0 dog (collinear), with a negative correlation between successive angles, but negligible correlation between non-successive angles. This suggests that curvilinearity is evaluated in a 4-dot window moving along the chain of dots, evaluating collinearity and smoothness but ignoring higher-order relationships. Moreover, the probabilistic model provides a remarkably precise numeric prediction of the magnitude of the correlation. Subjects also showed a reliable preference for equal spacing of dots along the virtual curve.
Curve detection in a noisy image
1997, Vision Research
In this paper we propose a new theory of the Gestalt law of good continuation. In this theory perceptual processes are modeled by an exponential pyramid algorithm. To test the new theory we performed three experiments. The subject's task was to detect a target (a set of dots arranged along a straight or curved line) among background dots. Detectability was high when: (a) the target was long; (b) the density of target dots relative to the density of background dots was large; (c) the local change of angle was small along the entire line; (d) local properties of the target were known to the subject. These results are consistent with our new model and they contradict prior models. © 1997 Elsevier Science Ltd. All rights reserved.
An algorithm for real-time vessel enhancement and detection
1997, Computer Methods and Programs in Biomedicine
In this paper we present an algorithm for the real-time enhancement and detection of blood vessels in medical images. The algorithm is based on a set of linear filters sensitive to vessels of different orientation and thickness. Such filters are obtained as linear combinations of properly shifted Gaussian kernels. The output of multiple oriented vessel-enhancing filters can be integrated to obtain images in which vessels are highly enhanced independently of their direction and thickness. To avoid spurious responses in the presence of step edges or to enhance the skeletons of vessels, the output of directional filters can be validated before integration. Skeleton detection and vessel segmentation can be performed via thresholding with hysteresis. Experimental results on synthetic images and real coronary arteriograms are reported.
Vision based navigation system for an endoscope
1996, Image and Vision Computing
A vision based navigation system to guide an endoscope inside a human colon has been designed and tested. It uses low level vision techniques to extract two types of navigational landmarks, dark regions and curved contours. Dark regions correspond to the distant inner space of the colon, called the lumen. The curved contours represent occlusions due to the inner colon muscles. A hierarchical search space and environment representation, called the QL-tree, was developed to integrate the visual features and implement the navigation system. It uses multiple quadtrees which are linked at all hierarchical levels. A multiprocessor system was employed to achieve real-time performance. The endoscope navigation system has been used successfully in artificial colon models.

View all citing articles on Scopus

View full text

Early orientation selection: Tangent fields and the dimensionality of their support

Artif. Intell.

Vision Res.

Computer Vision Graphics Image Process

Inferring surfaces from images

Artif. Intell.

A Spatio-Temporal Model for Early Visual Processing

Perception of the Visual World

Microcircuitry of the visual cortex

Ann. Rev. Physiol.

Moire effect from random dots

Nature (London)

Obtaining shape from shading information