Local feature analysis: a general statistical theory for object representation

Penev, P.S.; Atick, Joseph J.

doi:10.1088/0954-898x/7/3/002

Cited by 339 publications

(141 citation statements)

References 53 publications

Supporting

Mentioning

139

Contrasting

Unclassified

Order By: Relevance

“…Redundancy reduction requires that the lengths of sensory messages be proportional to their information contents, given by the negative logarithms of their respective probabilities (Shannon and Weaver, 1949). One framework for redundancy reduction involves factorial codes (Barlow et al, 1989;Linsker, 1988;Atick and Redlich, 1992), in which the probability of observing a particular signal is a product of independent factors, e.g., the features that code for it (Penev and Atick, 1996). Furthermore, if the strength of the factorial code output is proportional to its information content, the code can directly represent not only the sensory signal itself, but also its likelihood.…”

Section: Feature Extractionmentioning

confidence: 99%

“…One can now easily see that if the entropy H for some reconstructed image goes down, its likelihood goes up. One can draw S-H diagrams and observe how much of the SNR can be obtained relatively "cheaply"; afterward, even if one increases the SNR, i.e., gets a better approximation, the reconstruction obtained is very improbable, not at all likely (Penev, 1998).…”

Section: Feature Extractionmentioning

confidence: 99%

“…Although the basis images for PCA are eigenfaces, some of which resemble distorted versions of the whole face, the NMF basis is radically different: its images are localized features that correspond better with intuitive notions of the parts of faces. Penev and Atick (1996) have advanced the concept of local feature analysis (LFA) as a substitute for global (holistic) PCA for face recognition. In terms of local processing for feature extraction, LFA is conceptually similar to the use of Gabor jets by Malsburg's group (Okada et al, 1998) or to the attempts to define eigenimages corresponding to specific facial landmarks such as the eyes, nose, and mouth.…”

Section: Natural Basesmentioning

confidence: 99%

See 2 more Smart Citations

Pattern recognition: Historical perspective and future directions

Rosenfeld

Wechsler

2000

Int. J. Imaging Syst. Technol.

View full text Add to dashboard Cite

What being walks sometimes on two feet, sometimes on three, and sometimes on four, and is weakest when it has the most?" -The Sphinx's Riddle ABSTRACT: Pattern recognition is one of the most important functionalities for intelligent behavior and is displayed by both biological and artificial systems. Pattern recognition systems have four major components: data acquisition and collection, feature extraction and representation, similarity detection and pattern classifier design, and performance evaluation. In addition, pattern recognition systems are successful to the extent that they can continuously adapt and learn from examples; the underlying framework for building such systems is predictive learning. The pattern recognition problem is a special case of the more general problem of statistical regression; it seeks an approximating function that minimizes the probability of misclassification. In this framework, data representation requires the specification of a basis set of approximating functions. Classification requires an inductive principle to design and model the classifier and an optimization or learning procedure for classifier parameter estimation. Pattern recognition also involves categorization: making sense of patterns not previously seen. The sections of this paper deal with the categorization and functional approximation problems; the four components of a pattern recognition system; and trends in predictive learning, feature selection using "natural" bases, and the use of mixtures of experts in classification.

show abstract

Section: Feature Extractionmentioning

confidence: 99%

Section: Feature Extractionmentioning

confidence: 99%

Section: Natural Basesmentioning

confidence: 99%

See 1 more Smart Citation

Pattern recognition: Historical perspective and future directions

Rosenfeld

Wechsler

2000

Int. J. Imaging Syst. Technol.

View full text Add to dashboard Cite

show abstract

“…More robust approaches such as using linear discriminant analysis (LDA) [15], local feature analysis [16], independent component analysis [17] and Gabor wavelets [18] were also applied in facial action recognition. These techniques were evaluated by Donata et al [10] for the recognition of only 12 out of 46 facial actions for two independent regions of the face.…”

Section: Introductionmentioning

confidence: 99%

A face emotion tree structure representation with probabilistic recursive neural network modeling

Wong

Cho

2009

Neural Comput & Applic

View full text Add to dashboard Cite

show abstract

“…Much research [1][2][3] shows that human facial images usually consist of two kinds of features: global and local. Local features generally hold strong discriminatory characteristics of the facial image [4,5].…”

Section: Introductionmentioning

confidence: 99%

An Integrated Face Rec gniti n System Based n Multiscale L cal Discriminatory Features

Hong¹,

Tang

2001

Pattern Analysis & Applications

View full text Add to dashboard Cite

Despite some successes, the process of Automatic Facial Recognition (AFR) remains a significant challenge when unconstrained imaging conditions are involved. The authors believe that this occurs because an effective feature extraction method of facial images has not been found so far. In this paper a new approach to extract powerful local discriminatory features is described. First, the wavelet transform is used for extraction of multi-resolution coarse features, and then the emphasis is placed on the extraction of Multiscale fine Local Discriminatory Features (MLDFs). Instead of using traditional wavelet features, the authors examine the multiscale local statistical characteristics to derive stronger discriminatory features based on some important wavelet subbands. To efficiently utilise potentials of the extracted multi-MLDFs, an integrated recognition system is developed where the multi-classifiers first conduct the corresponding coarse classification, then a decision making scheme is used to associate different priorities with each of the classifiers to make the final recognition. Experiments have shown that this scheme provides superior performance to popular methods, such as Principal Components Analysis (PCA or Eigenface), wavelet features, neural networks, etc.

show abstract

Local feature analysis: a general statistical theory for object representation

Cited by 339 publications

References 53 publications

Pattern recognition: Historical perspective and future directions

Pattern recognition: Historical perspective and future directions

A face emotion tree structure representation with probabilistic recursive neural network modeling

An Integrated Face Rec gniti n System Based n Multiscale L cal Discriminatory Features

Contact Info

Product

Resources

About