Object recognition using invariant object boundary representations and neural network models

Bebis, George; Papadourakis, George M.

doi:10.1016/0031-3203(92)90004-3

Cited by 48 publications

(14 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For this work, we will use the centroidal distance profile (CDP) [2,10] as a 1-D representation of the airplane silhouettes. Specifically, if (x(n), y(n)) are an ordered sequence of boundary points which are equidistant along the boundary, then the CDP f (n) is given by…”

Section: Centroidal Distance Profilementioning

confidence: 99%

“…Object recognition from silhouettes is a problem with substantial literature and history in computer vision [1][2][3][4][5][6][7]. The ideal recognition system is one that is robust to orientation variations, scale variations, and boundary perturbations as well as localized distortions.…”

Section: Introductionmentioning

confidence: 99%

“…Our proposed approach is to use the HRP algorithm to extract a set of features which quantify the size and location of the subparts of the object. Since a number of researchers [1,2,3,10] have based their recognition schemes on a 1-D representation of the silhouette, we do as well. In particular, we use HRP to extract features from a 1-D representation of the silhouette, namely, the centroidal distance profile (CDP) [2,4,10].…”

Section: Introductionmentioning

confidence: 99%

“…Since a number of researchers [1,2,3,10] have based their recognition schemes on a 1-D representation of the silhouette, we do as well. In particular, we use HRP to extract features from a 1-D representation of the silhouette, namely, the centroidal distance profile (CDP) [2,4,10]. The HRP algorithm is used to construct a feature vector of low dimension, and recognition is performed using this feature vector and an M-ary hypothesis testing scheme.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Silhouette recognition using high-resolution pursuit

Jaggi

Karl

Mallat

et al. 1999

Pattern Recognition

View full text Add to dashboard Cite

This paper introduces a simple new approach to object recognition from silhouettes. This new approach utilizes features extracted using an adaptive approximation technique called high-resolution pursuit (HRP). In this work, a comparatively small set of HRP features and a simple recognition scheme are used. We demonstrate the strengths of the HRP-based recognition scheme by discriminating among 17 military aircraft. The HRP-based algorithm matches the performance of a widely studied method based on Fourier descriptors in the presence of boundary, scale and orientation variations, and surpasses the performance of the Fourier descriptor-based algorithm in the presence of occlusion and localized silhouette variations.

show abstract

Section: Centroidal Distance Profilementioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Silhouette recognition using high-resolution pursuit

Jaggi

Karl

Mallat

et al. 1999

Pattern Recognition

View full text Add to dashboard Cite

show abstract

“…In this section we review a proposal of a preprocessor, studied by Fuchs and Haken 7 in a non-neural system 10 , which uses Fourier transformations to overcome these di culties. Other non-neural approaches and comparisons with multilayer networks for recognition of boundaries can be found in the literature 20 . Let us consider a two-dimensional function s(x; y).…”

Section: Preprocessing the Input Patternsmentioning

confidence: 99%

Preprocessing in Attractor Neural Networks

Carvalhaes

Costa

Penna

1995

Int. J. Mod. Phys. C

View full text Add to dashboard Cite

Received () Revised () Preprocessing the input patterns seems the simplest approach to invariant pattern recognition by neural networks. The Fourier transform has been proposed as an appropriate and elegant preprocessor. Nevertheless, we show in this work that the performance of this kind of preprocessor is strongly a ected by the number of stored informations. This is because the phase of the Fourier transform plays a more important role than the amplitude in the recognition process.

show abstract