Feature Extraction by Best-Basis and Wavelet Methods

摘要

In the past year the AFOSR, through its award number F49620-92-J-0106, hasdirectly supported the research of Professors Wickerhauser and Weiss, as well as the work of temporary visitors, graduate students and postdoctoral researchers. In the third year of the original 3-year research plan, work has been completed on all proposed projects and have expanded work on some new projects that were not envisioned. During the period from 1 March 1995 to 29 February 1996, principal investigators Weiss and Wickerhauser wrote more than a dozen articles and books on harmonic and wavelet analysis and applications. The main focus was those properties of wavelets which need to be known before implementing feature detection algorithms, such as their time and frequency localization properties and their algebraic characterizations. This AFOSR contract permitted wider dissemination of these fundamental results. It supported two visiting consultants who lectured at Washington University, and allowed Weiss and Wickerhauser to attend and lecture at 8 scientific meetings in 5 cities. It also bought one additional fast workstation on which some numerical experiments were performed and which serves the research group in wavelet analysis at the Washington University mathematics department.