The introduction of wavelets in signal and image processing has provided a new tool to create innovative and novel methods for solving problems in such areas as data compression, signal analysis, and noise removal, to name a few. Although wavelets are popular and used extensively in research and in engineering applications, their usage in signature detection and classification is still an area open to extensive research. This dissertation discusses wavelet signal processing working in synergy with other processing techniques to detect, recognize, and classify abnormal and cueing signatures that are important to the military, industry, and medicine. This dissertation presents three different but related topics: (1) 1-D two-class military ground vehicle acoustics recognition; (2) 1-D multiple-class railroad wheel-bearing fault acoustic detection and identification; and (3) 2-D signal abnormality detection, such as microcalcifications in mammograms. In the two-class problem, a discrete wavelet transform-based acoustic signal processing algorithm that remotely recognizes military vehicles by sound is presented. This algorithm is implemented in a general-purpose DSP microprocessor for the real-time processing. Results from both computer simulation and real-time processing provide reliable and robust recognition of the vehicles. In the multiple-class problem, a novel method to detect, recognize, and classify a variety of railroad wheel-bearing defects using audible acoustics is presented. This algorithm consists of modules that mimic the human cochlea's cortex linkage and processing. Four different transform feature extractions are implemented to simulate cochlea processing: the fast Fourier transform, the continuous wavelet transform, the discrete wavelet transform, and the wavelet packet. The designed wavelet-based neural network provides reliable and highly accurate fault identification. In the 2-D application in medicine, an innovative detection algorithm that takes advantages of wavelet multiresolution analysis and synthesis is developed to assist radiologists looking for clusters of microcalcifications in digitized mammograms. Microcalcification regions may not be detectable by visual inspection or other detection techniques because of the inherent complexity revealed in mammograms and surrounding false positives. The developed algorithm successfully unmasks the complexity and limits the false positives. In all three topics, a thorough analysis, algorithm description and examples are provided.
展开▼
