&lt;title&gt;Local discriminant bases&lt;/title&gt;

Saito, Naoki; Coifman, Ronald R.

doi:10.1117/12.188763

Cited by 102 publications

(69 citation statements)

References 4 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A point-wise, pseudo distribution measure like KL, is fooled by the energy voids in a periodic texture aligning with energy peaks from the other texture at the same coordinate. KL, however, is widely used in WP basis selection [10] because it is additive and can be used to compare the relative energy when choosing child subbands over parent for an image region. Figure 4 shows four most discriminant feature images and corresponding classification results for the straw/grass image.…”

Section: Experiments and Discussionmentioning

confidence: 99%

Discriminant Feature Selection for Texture Classification

Bhalerao

Rajpoot

2003

Procedings of the British Machine Vision Conference 2003

View full text Add to dashboard Cite

The computational complexity of a texture classification algorithm is limited by the dimensionality of the feature space. Although finding the optimal feature subset is a NP-hard problem [1], a feature selection algorithm that can reduce the dimensionality of problem is often desirable. In this paper, we report work on a feature selection algorithm for texture classification using two subband filtering methods: a full wavelet packet decomposition and a Gabor type decomposition. The value of a cost function associated with a subband (feature) is used as a measure of relevance of that subband for classification purposes. This leads to a fast feature selection algorithm which ranks the features according to their measure of relevance. Experiments on a range of test images and both filtering methods provide results that are promising.

show abstract

Section: Experiments and Discussionmentioning

confidence: 99%

Discriminant Feature Selection for Texture Classification

Bhalerao

Rajpoot

2003

Procedings of the British Machine Vision Conference 2003

View full text Add to dashboard Cite

show abstract

“…The details of finding a discriminating basis from a wavelet packet table can be found in Saito & Coifman (1994), a description of the algorithms and methods we used is given in the Appendix. Although creating complete wavelet packet tables might be time consuming, once the basis is chosen, then only a small part of the tree has to be rebuilt for the classification process.…”

Section: Description Of the Wavelet Algorithmmentioning

confidence: 99%

“…The discriminating basis construction is described in detail in Saito & Coifman (1994). After a WPT has been generated for every click in the training set, a representation of the signal is searched that maximizes the distance between the different classes.…”

Section: Discriminating Basis Searchmentioning

confidence: 99%

“…To create a class vector, the strongest coefficients from this basis are chosen. In this report, these coefficients were selected with Fisher's class separability (Saito & Coifman, 1994), which calculates the discriminating power of a wavelet coefficient at position (j,k,m) as follows: , )) , j k m c å mean i and var i take the mean and variance of the coefficient over all samples in the class and mean c takes the mean over all classes. In our case, choosing the 15 strongest coefficients gave optimal results.…”

Section: Class Vector Creation and Classificationmentioning

confidence: 99%

“…For the second method, we tried to find characteristic features in a click using a local discriminant basis, constructed with a wavelet packet approach as described in Saito & Coifman (1994). Similar methods have been applied successfully in the classification of signals from other underwater mammals (Delory & Potter, 1998;Huynh et al, 1998;Delory et al, 1999), and a discrete wavelet transform has been used with some success in an attempt to classify sperm whales (Dougherty, 1999).…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

A comparison of model and non-model based time-frequency transforms for sperm whale click classification

et al. 2007

View full text Add to dashboard Cite

We tried to find discriminating features for sperm whale clicks in order to distinguish between clicks from different whales, or to enable unique identification. We examined two different methods to obtain suitable characteristics. First, a model based on the Gabor function was used to describe the dominant frequencies in a click, and then the model parameters were used as classification features. The Gabor function model was selected because it has been used to model dolphin sonar pulses with great precision. Additionally, it has the interesting property that it has an optimal time-frequency resolution. As such, it can indicate optimal usage of the sonar by sperm whales. Second, the clicks were expressed in a wavelet packet table, from which subsequently a local discriminant basis was created. A wavelet packet basis has the advantage that it offers a highly redundant number of coefficients, which allow signals to be represented in many different ways. From the redundant signal description a representation can be selected that emphasizes the differences between classes. This local discriminant basis is more flexible than the Gabor function, which can make it more suitable for classification, but it is also more complex. Class vectors were created with both models and classification was based on the distance of a click to these vectors. We show that the Gabor function could not model the sperm whale clicks very well, due to the variability of the changing click characteristics. Best performance was reached when three subsequent clicks were averaged to smoothen the variability. Around 70% of the clicks classified correctly in both the training and validation sets. The wavelet packet table adapted better to the changing characteristics, and gave better classification. Here, also using a 3-click moving average, around 95% of the training sets classified correctly and 78% of the validation sets. These numbers lowered by only a few per cent when single clicks, instead of a moving average, were classified. This indicates that, while the features may show too much variability to enable unique identification of individual whales on a click by click basis, the wavelet approach may be capable of distinguishing between a small group of whales.

show abstract

Spectral Data, Modern Classification Methods for

Mallet

Coomans

Vel

2000

Encyclopedia of Analytical Chemistry

View full text Add to dashboard Cite

A major concern which arises from the classification of spectral data, such as near‐infrared (NIR) spectra, is that the number of variables often exceeds the number of cases. In this instance each spectrum represents a case or observation, and each variable corresponds to the wavelengths at which a response (e.g. reflectance) is measured. Classification refers to the process of assigning a spectrum whose class identity is unknown into one of several predefined classes. There are two main strategies for combating the high‐dimensional (i.e. more variables than cases) scenario. One strategy is to use a high‐dimensional classifier, that is, one which is suited to the ill‐posed nature of the data. The second strategy is to reduce the number of variables by a feature extraction method and supply the variables to a low‐dimensional classifier, one which is suited to high observation‐to‐variable ratios. This article presents two high‐dimensional classification methods – regularized discriminant analysis (RDA) and penalized discriminant analysis (PDA). The low‐dimensional classifiers used include Bayesian linear discriminant analysis (BLDA), Bayesian quadratic discriminant analysis (BQDA), RDA and flexible discriminant analysis (FDA). Common stepwise selection routines are employed to investigate the effectiveness of the wavelet transform as a feature extraction technique. This includes implementation of the local discriminant bases (LDBs) algorithm and the adaptive wavelet algorithm (AWA).

show abstract

<title>Local discriminant bases</title>

Cited by 102 publications

References 4 publications

Discriminant Feature Selection for Texture Classification

Discriminant Feature Selection for Texture Classification

A comparison of model and non-model based time-frequency transforms for sperm whale click classification

Spectral Data, Modern Classification Methods for

Contact Info

Product

Resources

About