Shape Discrimination Using Fourier Descriptors

Persoon, E.; Fu, King-Sun

doi:10.1109/tsmc.1977.4309681

Cited by 895 publications

(159 citation statements)

References 8 publications

Supporting

Mentioning

139

Contrasting

Unclassified

Order By: Relevance

“…In this study, we examined how IT cortex might extract information about the overall shape of an object from information about local boundaries. We adopted a method of representing shapes in terms of local boundary orientation that is used in computer pattern recognition systems (9,10). The method depends, first, on determining the boundary orientation function for the shape-i.e., the orientation (tangent angle) of the shape's boundary measured at regular intervals around the perimeter.…”

mentioning

confidence: 99%

Shape recognition and inferior temporal neurons.

Schwartz¹,

Desimone²,

Albright³

et al. 1983

Proc. Natl. Acad. Sci. U.S.A.

298

164

View full text Add to dashboard Cite

Inferior temporal cortex plays an important role in shape recognition. To study the shape selectivity of single inferior temporal neurons, we recorded their responses to a set of shapes systematically varying in boundary curvature. Many inferior temporal neurons were selective for stimuli of specific boundary curvature and maintained this selectivity over changes in stimulus size or position. The method of describing boundary curvature was that of Fourier descriptors.What are the neural processes underlying shape recognition? How can we recognize an almost infinite variety of different shapes independent of their size and retinal location? In both man and monkey a likely site for mechanisms of shape recognition is inferior temporal (IT) cortex. Removal of this area impairs the visual recognition of shapes and patterns while leaving basic sensory capabilities, such as acuity, intact (1). Furthermore, many IT neurons are sensitive to the overall shape of objects rather than simply the orientation or location of particular edges (2-4). Finally, IT neurons have large receptive fields that extend into both visual half fields and almost always include the center of gaze (2, 3). These properties of IT neurons depend on the information IT cortex receives from the prestriate visual areas (5, 6).A common feature of neurons in striate and prestriate cortex is sensitivity to the orientation of local contours or boundaries (7,8). In this study, we examined how IT cortex might extract information about the overall shape of an object from information about local boundaries. We adopted a method of representing shapes in terms of local boundary orientation that is used in computer pattern recognition systems (9, 10). The method depends, first, on determining the boundary orientation function for the shape-i.e., the orientation (tangent angle) of the shape's boundary measured at regular intervals around the perimeter. Then, the boundary orientation function is expanded in a Fourier series. Each term in the Fourier expansion is associated with a particular frequency, amplitude, and phase and is known as a Fourier descriptor (FD) (9). Equivalently, individual FDs can be extracted by filtering the boundary orientation function with appropriate bandpass filters. Any shape is fully described by its set of FDs, and a smaller set of only the low-frequency terms can often provide the "gestalt" of a shape (9). Thus, the FDs are a powerful and efficient alphabet for representing and classifying shapes. Furthermore, this method of describing shape is independent of both the position and size of the stimulus. Size invariance is achieved by normalizing the perimeter to 21T before calculating the boundary orientation function. Position invariance is achieved by measuring the boundary orientation relative to the orientation of an arbitrary starting point on the.perimeter.Could IT neurons code shape on the basis of global features like FDs? To explore this possibility, we created a set of stimuli from single FDs. The inverse transform o...

show abstract

mentioning

confidence: 99%

Shape recognition and inferior temporal neurons.

Schwartz¹,

Desimone²,

Albright³

et al. 1983

Proc. Natl. Acad. Sci. U.S.A.

298

164

View full text Add to dashboard Cite

show abstract

“…= 1 n+m (23) was established by Bluml% This stronger form of convergence indicates that, in the limit, it is guaranteed that 0n will equal 6. It was Dvoretzky15 who gave a generalised form of these proofs of Robbins-Monro and Blum and showed that the convergence criteria Eqns.…”

Section: Learning Using Stochastic Approximationmentioning

confidence: 90%

Trends in Pattern Recognition and Machine Learning

Majumdar¹

1985

DSJ

View full text Add to dashboard Cite

show abstract

“…Region-based descriptors (e.g., position, size, skeleton, curvature, and moment) have a smaller calculation amount, but they are easily affected by noise, and the shape described by a single descriptor is indistinguishable. Boundary-based descriptors fit shape by approximately using a string or function, such as shape context [31], Fourier transform [32], and the turning function [33], that can better analyze the structural features of a shape.…”

Section: Shape Recognition In Building Simplificationmentioning

confidence: 99%

Template Matching and Simplification Method for Building Features Based on Shape Cognition

Yan

Zhang

2017

IJGI

View full text Add to dashboard Cite

This study proposes a template matching simplification method from the perspective of shape cognition based on the typical template characteristics of building distributions and representations. The method first formulates a series of templates to abstract the building shape by generalizing their polygons and analyzing their symbolic meanings, then conducts the simplification by searching and matching the most similar template that can be used later to replace the original building. On the premise of satisfying the individual geometric accuracy on a smaller scale, the proposed method can enhance the impression of well-known landmarks and reflect the pattern in mapping areas by the symbolic template. The turning function that describes shape by measuring the changes of the tangent-angle as a function of the arc-length is employed to obtain the similar distance between buildings and template polygons, and the least squares model is used to control the geometry matching of the candidate template. Experiments on real datasets are carried out to assess the usefulness of this method and compare it with two existing methods. The experiments suggest that our method can preserve the main structure of building shapes and geometric accuracy.

show abstract

Shape Discrimination Using Fourier Descriptors

Cited by 895 publications

References 8 publications

Shape recognition and inferior temporal neurons.

Shape recognition and inferior temporal neurons.

Trends in Pattern Recognition and Machine Learning

Template Matching and Simplification Method for Building Features Based on Shape Cognition

Contact Info

Product

Resources

About