A dynamic vision classification system using Fourier descriptions

Kamel, K.; Abnous, R.; Sun, G.

doi:10.1109/secon.1990.117847

Cited by 3 publications

(1 citation statement)

References 1 publication

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The expression for the discrete-time STFT is Similarly, the discrete STFT which is discrete in both time and frequency can be expressed in terms of the DFT. Replacing the fixed variable no with n in (3) (4) Frequency sampling of (4) gives the discrete STFT. Replacing the continuous frequency variable Ct) with a frequency sampling interval of 2it/N we have, (5) is seen as an N-point DFT of each short-time section padded with the appropriate number of zeros.…”

Section: The Stft From the Dftmentioning

confidence: 99%

<title>Framework for object recognition in images using the short-time Fourier transform</title>

Manian

Vásquez

1996

SPIE Proceedings

View full text Add to dashboard Cite

The problem of object recognition in images regardless of their scale and orientation is considered in this paper. A framework is used to train and to recognize or classify a transformed object. A set of features obtained from the short-time Fourier transform of the object is used for scale and rotation invariant recognition. An analysis window is used to compute the short-time Fourier transform. The Fourier magnitudes in the polar domain constitute the scale invariant and rotation invariant features. Since, short time sections are used in this method, features are more separable because of the localization of the window which is useful for discriminating variants of very similar objects. The recognition system is tested for different sets of scales and rotations of several objects. This framework performed well for the range of scales and orientations of the objects considered. The framework is computationally efficient and showed robustness in the presence of noise. The tasks involved are simple and the framework can be used for realtime applications.Object recognition systems are widely used in industrial inspection systems, military systems, in medical applications and scene analysis in vision systems. The goal of an object recognition system is to recognize or classify a given object or image under two dimensional linear transformations such as change in size and/or orientation. Several techniques have been used for two dimensional and three dimensional object recognition. These methods are based on pattern recognition, where a certain set of features are extracted from the pattern. These features are used to train a classification system. The system is then used to extract features from unknown objects and classify them. Traditionally, Fourier descriptors, moment invariants and autoregressive methods have been used1'2'3. The Fourier descriptors are obtained by extracting the boundary of the object and obtaining a set of frequency-domain descriptors by expanding the boundary in a Fourier series4. Moment invariants method uses a set of seven normalized central moments based on the second and third order moments calculated from the objects. Features derived from autoregressive models5 have also been used for object classification.In this paper a framework for classifying objects based on the short-time Fourier transform (STFT) regardless of their size and orientation is presented. This approach to object recognition is used to improve over the traditional Fourier based methods and allows more flexibility in obtaining the descriptors. This framework has been used for identifying objects under different scales and rotations and it gives a good performance under the presence of noise. Section 3 gives an introduction to the STFT also called as windowed Fourier transform. Section 4 defines the framework and explains the algorithm. Experimental results are presented in section 5 and section 6 gives the conclusions. THE WINDOWED FOURIER TRANSFORMThe Fourier transform (Vf) technique has been used for object recognitio...

show abstract

Section: The Stft From the Dftmentioning

confidence: 99%