Wavelets and other multi-resolutional representations are useful due to their ability to resolve edges, yield sparse linear representations of signals and images, and they are related to the way we process visual stimuli. Recent work reduces the dimensionality of image sets by maximizing the energy of certain wavelet subbands of interest (Wavelet PCA). In earlier work, a nonlinear vector multiresolutional analysis was proposed for vector signals. Many image processing tasks have been shown to be better executed in the wavelet domain and wavelet representations of images have a strong appeal. We showed that it is possible to halftone images using error diffusion in the wavelet domain. In related work, we take a multiresolutional approach to global optimization.

Publications:

"A multiresolutional estimated gradient architecture for global optimization," Megan Hazen and Maya R. Gupta, Congress on Evolutionary Computing , 2006.

"Wavelet principal component analysis and its application to hyperspectral imagery," Maya R. Gupta and Nathaniel P. Jacobson, IEEE Intl. Conf. on Image Processing , 2006.

"Segmenting for wavelet compression," Maya R. Gupta and Andrey Stroilov, Proceedings of the Data Compression Conference, p. 462, 2005.

"Halftoning on the wavelet domain," Maya R. Gupta, Proceedings of the SPIE Electronic Imaging Conference, vol. 5008, pp. 431-442, 2003.

"Robust speech recognition using wavelet coefficient features," Maya Gupta and Anna Gilbert, Proceedings of the IEEE Workshop on Automatic Speech Recognition and Understanding, 2001.

"Nonlinear vector multiresolutional analysis," Maya Gupta and Anna Gilbert, Proceedings of the Asilomar Conference on Systems and Signals, 2000.

Related Work:

Global Optimization

Visualizing Remote Sensing Image Datasets